สำเนาบทสนทนา 🔗
1
00:00:00,196 --> 00:00:03,960
Welcome to This Dumb Week, a weekly live Twitter Space hosted by Alex Marinos and
2
00:00:04,460 --> 00:00:08,323
yours truly, Dr. Rollergator. Available in podcast form on Apple Podcasts,
3
00:00:08,661 --> 00:00:11,972
Spotify, Amazon Music, Audible, iHeartRadio,
4
00:00:12,200 --> 00:00:15,360
and others. Alex, welcome back. Have you had a sufficiently
5
00:00:15,376 --> 00:00:18,276
dumb week? Yeah, it's been, uh,
6
00:00:18,907 --> 00:00:22,403
it's been pretty good as far as these things go. Nobody— no new war started.
7
00:00:22,468 --> 00:00:25,187
No news. We can all go home early. Thank you for joining,
8
00:00:25,284 --> 00:00:29,051
everybody. We'll see you. see you next week. I think we might pick
9
00:00:29,551 --> 00:00:33,097
up from where we stopped last week this week, I think. Maybe. Perhaps. I mean,
10
00:00:33,691 --> 00:00:36,646
it was the tail end. No, I mean, we won't be starting there. We will
11
00:00:36,758 --> 00:00:40,130
be getting to it, of course, Alex. But we
12
00:00:40,630 --> 00:00:44,433
will try to get through other things that happened, because apparently other things did
13
00:00:44,449 --> 00:00:47,853
happen this week. We will— I do hope we go through that
14
00:00:48,142 --> 00:00:51,370
dollar thread I sent to you, because it's a museum exhibit.
15
00:00:52,332 --> 00:00:54,929
We can do that, too. We can certainly do that, too.
16
00:00:55,538 --> 00:00:59,225
The first news we have on the miscellaneous block— we're going to have a short
17
00:00:59,725 --> 00:01:02,335
miscellaneous block, Alex, before we get to other things this week.
18
00:01:03,217 --> 00:01:07,481
There's a new alert this morning about those popular GLP-1 weight loss
19
00:01:07,981 --> 00:01:11,536
drugs. The FDA is now mandating a warning
20
00:01:11,649 --> 00:01:14,935
label that the drugs carry a potential increased risk
21
00:01:14,999 --> 00:01:19,152
of a rare eye condition that may cause permanent
22
00:01:19,168 --> 00:01:22,845
vision loss. The problem is caused by decreased blood flow to the optic nerve.
23
00:01:23,262 --> 00:01:26,859
Top GLP-1 drugmakers are responding, saying safety
24
00:01:26,971 --> 00:01:30,664
is their top priority. Eli Lilly adding that the FDA's
25
00:01:31,164 --> 00:01:34,823
decision does not reference any findings specific to its
26
00:01:35,323 --> 00:01:39,110
medications, while Novo Nordisk says it continues to believe the data
27
00:01:39,174 --> 00:01:42,466
does not establish a cause-and-effect relationship.
28
00:01:44,943 --> 00:01:48,074
I just have to say, Alex, you know, who, who really cares if you can't
29
00:01:48,574 --> 00:01:51,764
see that you look fabulous? Everyone else can see you look fabulous. Does it really
30
00:01:51,780 --> 00:01:55,689
matter if the GLP-1s may cause a little blindness?
31
00:01:57,513 --> 00:02:00,969
In, in fact, you know, if you wake up sick and tired
32
00:02:01,469 --> 00:02:06,087
of seeing yourself in the
Load full transcript
mirror, you know, getting that, getting that middle-aged
33
00:02:06,232 --> 00:02:09,778
physique, um, GLPs could help you one way
34
00:02:10,278 --> 00:02:13,616
or another. Yes, and when you rise through adversity,
35
00:02:13,712 --> 00:02:17,111
you become a stronger person at the end. So, you know, if you— That's right.
36
00:02:17,239 --> 00:02:19,195
Whatever doesn't kill you makes you blind.
37
00:02:20,525 --> 00:02:23,507
There you go. There it goes. All right, Alex.
38
00:02:23,748 --> 00:02:27,675
Now, do you remember the before times, before what
39
00:02:28,175 --> 00:02:31,491
they call artificial intelligence, or as the Trump administration
40
00:02:31,587 --> 00:02:34,889
has now donned it, superintelligence? Do you remember the
41
00:02:35,389 --> 00:02:38,932
before times at all? Uh, sometimes I
42
00:02:38,980 --> 00:02:42,190
have flashbacks, but I, I can't put it together. I'm not sure
43
00:02:42,690 --> 00:02:45,897
exactly what, what, what it was, but I, I'm— I know it existed.
44
00:02:46,844 --> 00:02:49,621
Well, do you remember how people used to hack things themselves?
45
00:02:50,600 --> 00:02:54,227
Uh, where they didn't— Yeah. Okay. They would type on keyboards and
46
00:02:54,727 --> 00:02:59,491
shit. This next story, this next story is a couple of rapscallions
47
00:02:59,587 --> 00:03:02,219
engaged in that old antiquated pastime.
48
00:03:04,334 --> 00:03:07,812
Just before midnight, this man pulls up to this Glendale ATM. He's there
49
00:03:08,004 --> 00:03:11,353
just like you and me, to get cash. But instead of entering his
50
00:03:11,385 --> 00:03:15,215
debit card or his PIN number, he's getting under the hood. He's not a worker
51
00:03:15,296 --> 00:03:18,837
or a security guard. He's a criminal about to hit the jackpot.
52
00:03:19,398 --> 00:03:22,475
Video shows him using a key to unlock panels underneath. He pulls
53
00:03:22,523 --> 00:03:26,417
up the entire front of the machine. Watch the screen go to black and
54
00:03:26,433 --> 00:03:29,671
then appear to reboot. He and his driver then pull
55
00:03:29,719 --> 00:03:33,470
away. About an hour later, they return to the ATM at the UW Credit
56
00:03:33,486 --> 00:03:37,254
Union on Port Washington Road, and it looks like the jackpot
57
00:03:37,754 --> 00:03:41,277
delivered. Pulling stack after stack of money, even fanning out
58
00:03:41,777 --> 00:03:45,333
the bills, boldly showing it off for a selfie, then showing it to someone
59
00:03:45,429 --> 00:03:48,059
on FaceTime. They took off undetected,
60
00:03:48,219 --> 00:03:52,387
$110,000 richer. Criminal organizations, they are always
61
00:03:52,579 --> 00:03:55,994
looking for ways to increase their profits.
62
00:03:56,106 --> 00:04:00,306
The FBI now revealing to 12 News robbers like these are sophisticated crooks who install
63
00:04:00,450 --> 00:04:04,538
malware, then hack into the ATM with their own computer or hard drive.
64
00:04:04,826 --> 00:04:07,487
It doesn't say that, hey, I'm pulling this from, uh,
65
00:04:08,049 --> 00:04:11,591
SAC Carr's account, or I'm pulling this from James Stratton's account. It tells the ATM,
66
00:04:11,976 --> 00:04:15,647
hey, it doesn't matter, we're just going to start spitting out 20s. Exactly, yes.
67
00:04:15,807 --> 00:04:18,757
Special Agent in Charge Alan Carr leads Milwaukee's FBI office.
68
00:04:19,093 --> 00:04:22,139
Think of this as the new bank robbery without the, the force of the
69
00:04:22,639 --> 00:04:25,986
violence of a bank robbery. It's a trend on the rise. Days Days after the
70
00:04:26,486 --> 00:04:29,689
Glendale heist, thieves hit the U.S. Bank on 27th Street in Greenfield,
71
00:04:30,026 --> 00:04:33,424
smearing mud on the camera and opening the hood. And before long,
72
00:04:33,617 --> 00:04:37,047
another jackpot payout. In Greenfield, the suspects returned again and
73
00:04:37,547 --> 00:04:40,830
again and got away with some $230,000 from 2
74
00:04:41,330 --> 00:04:43,988
different machines. Investigators call it jackpotting.
75
00:04:44,132 --> 00:04:47,611
It's generally a much more sophisticated
76
00:04:47,627 --> 00:04:51,089
crew and generally more people than just somebody traditionally
77
00:04:51,589 --> 00:04:54,712
showing up at a bank. Carr says the groups work for days surveilling the bank,
78
00:04:54,792 --> 00:04:57,405
its cameras, and vulnerabilities. Then they strike.
79
00:04:58,307 --> 00:05:01,732
Generally, what we have seen here in Wisconsin is that people come in, they hit
80
00:05:02,232 --> 00:05:04,263
the ATMs, and they leave, or they're part of a larger crew. And they head
81
00:05:04,311 --> 00:05:07,756
to another state? They head to another state or another area, yeah. Banks report
82
00:05:08,256 --> 00:05:11,169
more than 1,900 jackpotting incidents since 2020,
83
00:05:11,602 --> 00:05:15,255
resulting in more than $20 million lost last year alone.
84
00:05:15,672 --> 00:05:18,860
We always encourage the banks to take common-sense approaches and measures,
85
00:05:19,373 --> 00:05:22,994
updating software to overcome vulnerabilities, increasing the
86
00:05:23,494 --> 00:05:26,584
physical protections of ATMs, Cameras in the area.
87
00:05:26,825 --> 00:05:30,095
Greenfield police closed their case, telling 12 News they don't have
88
00:05:30,595 --> 00:05:33,589
any leads on who did it. But Glendale police say the 2 cases may be
89
00:05:33,637 --> 00:05:36,747
related. Search warrants show at least one of the suspects may be from
90
00:05:36,763 --> 00:05:40,611
Chicago, and Glendale police are working to identify everyone.
91
00:05:43,320 --> 00:05:46,734
Now, Alex, a couple things come to mind. One, it's really nice you hear
92
00:05:46,798 --> 00:05:50,661
that, that the Gen Z and below, uh, cohort, they are,
93
00:05:50,870 --> 00:05:54,509
they are rebelling against AI and they are doing more things IRL.
94
00:05:54,781 --> 00:05:58,349
They are touching grass more frequently because they are
95
00:05:58,414 --> 00:06:02,676
anti-AI or anti-artificial— artificiality.
96
00:06:02,773 --> 00:06:06,950
And so these young, you know, rapscallions are demonstrating
97
00:06:07,030 --> 00:06:11,353
that to us. They are actually going outside and actively participating in
98
00:06:11,449 --> 00:06:15,210
real-life things. They are going to the ATMs and hacking
99
00:06:15,242 --> 00:06:19,334
them manually. That is a very— they're
100
00:06:19,834 --> 00:06:22,991
not just sitting in their basements, Alex, trapped in a virtual world. They're not doing
101
00:06:23,491 --> 00:06:27,112
this from Second Life. No, no, no. And let's not forget, they're returning to
102
00:06:27,612 --> 00:06:31,359
the traditional money of cashness. They're touching
103
00:06:32,144 --> 00:06:36,246
green, but not necessarily grass.
104
00:06:36,310 --> 00:06:40,156
Fiat. Yeah. Yeah. They're not hijacking Binance.
105
00:06:40,493 --> 00:06:43,489
Only because the government said so. It's amazing. It's amazing.
106
00:06:43,682 --> 00:06:46,662
Think about this, Alex. Now, here's something.
107
00:06:47,832 --> 00:06:49,979
What happened to the benefits of flock cameras?
108
00:06:51,377 --> 00:06:55,443
Because apparently these guys are all untraceable
109
00:06:55,459 --> 00:06:59,528
with that network. We can, we can find out what, where the cop's
110
00:06:59,608 --> 00:07:02,957
ex-wife is, or the cop's girlfriend is, what she's been up to,
111
00:07:03,005 --> 00:07:06,850
but we can't find out where the, the hundreds of thousands
112
00:07:07,350 --> 00:07:11,160
of dollars in ATM theft is going by following the—
113
00:07:11,480 --> 00:07:14,893
Like, use their mobile phones to take selfies and
114
00:07:15,069 --> 00:07:18,562
FaceTiming people. Surely this is trackable
115
00:07:18,594 --> 00:07:21,768
somehow. No, just that— I'm not sure, man.
116
00:07:22,232 --> 00:07:25,486
I'm not sure. I don't understand how investigations go. I thought that you,
117
00:07:25,502 --> 00:07:28,212
like, you know, got a bunch of dots and then you started to connect them,
118
00:07:28,228 --> 00:07:31,482
but apparently it's just, uh, you beg for people in the
119
00:07:31,578 --> 00:07:35,762
audience to tell you who did it. I think that's the cheapest
120
00:07:36,262 --> 00:07:39,929
path to go. I think the real answer is that the NSA got the
121
00:07:40,429 --> 00:07:42,366
call and they were like, how much did they steal?
122
00:07:43,199 --> 00:07:47,864
$100,000? Yeah, sorry, man, like, I don't— Call
123
00:07:48,364 --> 00:07:51,327
me when they steal more, or at least do a January 6th.
124
00:07:52,112 --> 00:07:56,088
All right, Alex, now that closes off our miscellaneous stories.
125
00:07:56,184 --> 00:07:59,887
Can you believe it? That's a very short miscellaneous story segment. Like, it's just,
126
00:08:00,111 --> 00:08:03,702
just 2 and we're out. Now we're gonna have another short segment,
127
00:08:03,798 --> 00:08:06,667
Alex, uh, but we get to resurrect a theme song.
128
00:08:06,876 --> 00:08:10,531
Uncle Jeffy, hey there,
129
00:08:10,611 --> 00:08:14,250
he's the uncle everybody trusts.
130
00:08:14,939 --> 00:08:17,857
Uncle Jeffy, hey there.
131
00:08:18,370 --> 00:08:20,491
He's never eating babies.
132
00:08:22,650 --> 00:08:26,709
He's got parties on the weekend. Lots of
133
00:08:27,209 --> 00:08:31,703
friends you've seen on TV. Life's an adventure
134
00:08:32,203 --> 00:08:36,295
every day when you're staying with Uncle
135
00:08:36,795 --> 00:08:36,845
Jeffy.
136
00:08:41,593 --> 00:08:45,140
This is from NPR, Alex. The federal government failed. This better
137
00:08:45,640 --> 00:08:49,004
not be your intro to the Eric Weinstein segment. It is not.
138
00:08:49,469 --> 00:08:52,932
It is not, Alex. Federal government failed to hold Epstein
139
00:08:53,432 --> 00:08:57,420
accountable, Republican committee says. From NPR, the Republican-controlled House Oversight
140
00:08:57,920 --> 00:09:01,171
Committee released a report Tuesday on its investigation into convicted
141
00:09:01,671 --> 00:09:04,906
sex offender Jeffrey Epstein and his associates, concluding that, that the federal
142
00:09:04,954 --> 00:09:08,785
government mishandled criminal investigations into the disgraced financier
143
00:09:08,802 --> 00:09:12,376
and failed survivors. Quote, Epstein operated with impunity,
144
00:09:12,408 --> 00:09:16,544
shielded and enabled by powerful connections and a justice system that repeatedly
145
00:09:16,560 --> 00:09:20,553
failed its victims, said committee chair James Comer, a Republican from Kentucky.
146
00:09:21,067 --> 00:09:24,566
The scathing report comes after 15 months of interviews with 19 people,
147
00:09:24,678 --> 00:09:28,225
including former President Bill Clinton and former Secretary of State Hillary Clinton,
148
00:09:28,771 --> 00:09:31,917
Microsoft co-founder Bill Gates, and Commerce Secretary Howard Lutnick.
149
00:09:32,222 --> 00:09:35,753
Many of the interviews were conducted behind closed doors, with transcripts released
150
00:09:36,253 --> 00:09:39,670
days later. Appearing in the Department of Justice Epstein files or
151
00:09:40,170 --> 00:09:43,345
before the committee is not necessarily an indication of criminal wrongdoing.
152
00:09:44,180 --> 00:09:47,855
The report found that the facts revealed by investigation do not suggest
153
00:09:48,355 --> 00:09:52,190
President Donald Trump was implicated in any of Epstein's crimes. The report also
154
00:09:52,690 --> 00:09:56,687
accuses Democrats of launching an unfounded witch hunt by attempting to insinuate that Trump
155
00:09:57,187 --> 00:10:01,360
was involved in Epstein's crimes. Epstein was arrested on federal sex trafficking
156
00:10:01,425 --> 00:10:04,942
charges in 2019 and later died in prison. Authorities determined
157
00:10:04,958 --> 00:10:08,571
his death was a suicide. In January, Democrats will
158
00:10:08,587 --> 00:10:12,924
take control of the House, with
Representative Robert Garcia expected
159
00:10:13,424 --> 00:10:16,537
to be at the helm of the committee. In a statement Tuesday, Garcia criticized the
160
00:10:17,037 --> 00:10:20,595
report as partisan. Quote, oversight Republicans have released a report based on
161
00:10:20,627 --> 00:10:24,762
an incomplete investigation and zero public hearings on Jeffrey Epstein,
162
00:10:24,874 --> 00:10:28,849
Garcia said. We intended to hold real public hearings uplifting
163
00:10:28,897 --> 00:10:32,647
survivors and asking tough questions on the powerful men who associated
164
00:10:33,147 --> 00:10:36,141
with Epstein, including Donald Trump. Our work is just getting started.
165
00:10:37,359 --> 00:10:41,366
Isn't that good news, Alex? We still— we don't have enough Epstein stuff,
166
00:10:41,382 --> 00:10:44,523
so we get more Epstein stuff, more promises of more Epstein stuff. We got a
167
00:10:45,023 --> 00:10:48,195
new season. Oh, well, Trump, uh, I, I don't know if
168
00:10:48,695 --> 00:10:52,703
you know, but, um, uh, Massie has, uh, gotten enough signatures to force,
169
00:10:53,633 --> 00:10:56,168
um, the, uh,
170
00:10:56,857 --> 00:10:59,568
the vote on the— his second Epstein bill.
171
00:11:00,611 --> 00:11:03,900
All right, well, more, more and more Epstein. I know
172
00:11:04,060 --> 00:11:07,284
I haven't had my fill. I, I do have 2 things to
173
00:11:07,784 --> 00:11:10,060
say about that bill. Uh, first of all, um,
174
00:11:10,878 --> 00:11:14,891
the, uh, the, the— sorry. I'm hearing
175
00:11:14,923 --> 00:11:16,476
some noises which are distracting me.
176
00:11:18,507 --> 00:11:22,291
Sorry. Okay. Never mind. Move on. Trump has long
177
00:11:22,324 --> 00:11:26,252
maintained that he had a falling out with the disgraced financier before Epstein was
178
00:11:26,365 --> 00:11:30,212
indicated for— or sorry, indicted, my apologies— for soliciting
179
00:11:30,228 --> 00:11:34,285
prostitution in 2006. In February, White House spokeswoman Abigail
180
00:11:34,333 --> 00:11:37,739
Jackson told NPR in a statement that just as President Trump
181
00:11:38,239 --> 00:11:41,265
has said, he's been totally exonerated on anything related to Epstein.
182
00:11:42,443 --> 00:11:45,778
The report puts the next steps in the Epstein investigation into the hands of the
183
00:11:45,794 --> 00:11:50,026
DOJ, saying that the agency needs to interview or reinterview witnesses
184
00:11:50,122 --> 00:11:54,226
and investigate crimes committed by Epstein's associates. Maxwell is the only
185
00:11:54,726 --> 00:11:58,201
associate of Epstein to be charged with a crime and sentenced to prison. In June,
186
00:11:58,233 --> 00:12:02,160
the committee interviewed Sarah Kellen, a longtime assistant of Epstein who
187
00:12:02,660 --> 00:12:05,976
says she is also a survivor of his abuse. According to the committee's report,
188
00:12:06,232 --> 00:12:09,775
Kellen revealed the names of 2 men in her transcribed testimony that
189
00:12:10,275 --> 00:12:13,494
the committee sent the DOJ for further investigation. The DOJ has not
190
00:12:13,542 --> 00:12:17,281
announced any investigations into the men. In January,
191
00:12:17,618 --> 00:12:20,989
Democrats will take control of the House, with Representative Robert Garcia expected
192
00:12:21,489 --> 00:12:24,102
to be at the helm of the committee. In a statement Tuesday— oh wait,
193
00:12:24,134 --> 00:12:27,264
why did that repeat? Apologies. Did I
194
00:12:27,986 --> 00:12:31,613
accidentally copy this twice? Apologies. That was,
195
00:12:31,982 --> 00:12:35,047
uh, that was the end of it there. The DOJ has not announced any investigation
196
00:12:35,547 --> 00:12:39,268
into the men. I apologize for duplicating the copying and pasting of that story.
197
00:12:40,376 --> 00:12:42,913
So that takes care of the Uncle Jeffy segment there, Alex.
198
00:12:43,475 --> 00:12:46,622
We get more, more Uncle Jeffy. Now, do you catch your
199
00:12:47,087 --> 00:12:50,427
thoughts on the, the Massey thing? Yeah, so the, the,
200
00:12:50,684 --> 00:12:53,847
the first thing which relates to our, our theme the last few
201
00:12:53,895 --> 00:12:58,488
weeks is that apparently the first Epstein Transparency
202
00:12:58,988 --> 00:13:02,630
Act bill was quite likely the first
203
00:13:02,727 --> 00:13:06,710
AI-generated bill to be voted into law. So there might have been others
204
00:13:06,903 --> 00:13:10,514
that had been submitted, but this was almost certainly the first
205
00:13:11,014 --> 00:13:13,259
one to actually be enacted.
206
00:13:14,088 --> 00:13:17,679
Okay. All right. I did not know if you knew that.
207
00:13:19,783 --> 00:13:23,069
I did not know that. So do we know what
208
00:13:23,165 --> 00:13:26,821
model might have participated in this generation? Was it '04,
209
00:13:26,966 --> 00:13:30,480
that lovely '04 that has caused so many problems
210
00:13:30,980 --> 00:13:33,570
in the court? system. We could just put this on it, on its, on its
211
00:13:34,070 --> 00:13:37,995
rap sheet as well. Uh, all right, I don't know what they used, but the,
212
00:13:38,075 --> 00:13:41,586
uh, the new— the— because the first one was generated
213
00:13:42,086 --> 00:13:44,135
in a rush and machine-generated,
214
00:13:44,151 --> 00:13:47,854
apparently the second one is, is plugging all the
215
00:13:47,918 --> 00:13:52,070
holes basically that the first one left, uh, including allowing
216
00:13:52,198 --> 00:13:55,581
states and state AGs to sue the federal
217
00:13:55,629 --> 00:13:59,267
government to force the release of documents that the government isn't releasing.
218
00:13:59,315 --> 00:14:02,840
Because right now, even though by the original law,
219
00:14:03,706 --> 00:14:07,311
um, the AG of the United States has to release the documents,
220
00:14:07,808 --> 00:14:11,429
nobody has standing to take
221
00:14:11,494 --> 00:14:14,778
this to court. Uh, so they're like, all right, who's— you
222
00:14:15,278 --> 00:14:19,393
know, we gotta, um, create some sort of counterparty
223
00:14:19,426 --> 00:14:22,759
to, to, to— that has legal authority to,
224
00:14:22,871 --> 00:14:25,711
uh, you Pursue this. Well,
225
00:14:26,000 --> 00:14:29,147
as long as the rest of our lives, as long as the rest of our
226
00:14:29,275 --> 00:14:33,032
lives are dedicated to talking about Donald Trump and Jeffrey Epstein,
227
00:14:33,626 --> 00:14:37,062
I'm satisfied because that's all, all I want to talk about for the
228
00:14:37,562 --> 00:14:39,775
rest of my life. I hope it never ends,
229
00:14:39,871 --> 00:14:43,275
Alex. I get it, right? I get where you're coming from.
230
00:14:43,644 --> 00:14:47,160
At the same time, if we do let them just frigging
231
00:14:47,192 --> 00:14:49,360
slide with like, oh,
232
00:14:50,639 --> 00:14:52,849
How about I give you half the documents? The best I can do.
233
00:14:53,970 --> 00:14:56,878
And we're like, all right, whatever. I'm bored with this anyway.
234
00:14:58,745 --> 00:15:02,378
You know, that's not. It doesn't teach. It doesn't teach
235
00:15:02,475 --> 00:15:05,813
good manners to the government. So even on principle,
236
00:15:05,829 --> 00:15:09,380
I think you should support this. Even even if you are sick of talking about
237
00:15:09,880 --> 00:15:13,204
it. All right, I'll I'll I'll I'll put a ribbon on
238
00:15:13,782 --> 00:15:17,388
to show my support. Is that sufficient or a colored
239
00:15:17,437 --> 00:15:20,784
wristband? We need to find a new color for our
240
00:15:20,849 --> 00:15:23,479
color revolution, dammit. Oh, okay.
241
00:15:24,442 --> 00:15:28,051
All right, now moving forward, Alex, because that's the only thing on the stack of
242
00:15:28,551 --> 00:15:32,142
the Epstein list there, we're going to breeze through this next
243
00:15:32,735 --> 00:15:36,088
semi-major story of the week. Not much to go into, but it
244
00:15:36,553 --> 00:15:39,986
was pretty much a big story. Tonight, email has now been down for
245
00:15:40,162 --> 00:15:44,253
hours for people across the country who use Microsoft Outlook Copilot
246
00:15:44,753 --> 00:15:47,540
and other services. So what's behind this? Here's Trevor Ault.
247
00:15:48,533 --> 00:15:52,188
Tonight, a major outage impacting Microsoft platforms across North
248
00:15:52,688 --> 00:15:56,196
America, knocking out critical tools including email. The outage
249
00:15:56,276 --> 00:15:59,482
crippling communications for hours, leaving thousands of people
250
00:15:59,982 --> 00:16:03,730
unable to access their email accounts on Microsoft service Outlook. The company's
251
00:16:03,810 --> 00:16:06,936
AI assistant Copilot was also knocked out for a time.
252
00:16:07,337 --> 00:16:10,623
Outage tracker Downdetector appeared to show a spike in reports
253
00:16:11,123 --> 00:16:14,647
after 1 PM today, with tens of thousands of users reporting problems
254
00:16:15,147 --> 00:16:18,302
sending and receiving emails.
Late today, Microsoft Microsoft said it's slowly
255
00:16:18,318 --> 00:16:21,704
restoring service. In a series of online statements, they say the outage may
256
00:16:22,204 --> 00:16:26,083
have originated from a specific component that's responsible for processing server health,
257
00:16:26,372 --> 00:16:28,827
but it's not clear what change triggered the issue.
258
00:16:29,725 --> 00:16:32,854
And David, this has had a massive impact. Outlook is one of the most used
259
00:16:33,354 --> 00:16:36,913
email services in the world, and many corporations and government agencies
260
00:16:37,105 --> 00:16:40,458
depend on it for communication. David Trevor reporting tonight.
261
00:16:40,539 --> 00:16:43,507
Trevor, thank you with us here tonight. Lee, we appreciate it. Well,
262
00:16:43,523 --> 00:16:47,551
tonight, given the ground stops across this country, it's important to point out that Microsoft
263
00:16:48,051 --> 00:16:51,819
is apologizing for its global outage affecting millions of people and their email
264
00:16:51,948 --> 00:16:55,366
around the world. And of course, here in the U.S., the outage disrupting schools,
265
00:16:55,382 --> 00:16:58,223
hospitals, government agencies in more than 70 countries.
266
00:16:58,608 --> 00:17:02,186
Millions here in this country with no email yesterday and early this morning.
267
00:17:02,475 --> 00:17:05,958
It lasted more than 24 hours. Tonight, Microsoft now blaming,
268
00:17:06,070 --> 00:17:09,889
quote, a technical issue not related to cyber or AI agents.
269
00:17:10,379 --> 00:17:14,591
Adding, we have no indication this is the result of malicious or cyber activity.
270
00:17:15,362 --> 00:17:19,358
When we come back on this very busy Friday night, a college student— So,
271
00:17:19,534 --> 00:17:22,824
Alex, now there is a— there's been a nice trend this week,
272
00:17:22,921 --> 00:17:26,178
which is that any single thing that has been reported has
273
00:17:26,210 --> 00:17:29,436
gotten a lot of people going, is this, you know, the AI?
274
00:17:29,677 --> 00:17:33,320
Is the AI gone rogue again? And the doomers have been really
275
00:17:33,352 --> 00:17:37,813
hoping, really hoping that it is an attack and then announcing
276
00:17:38,327 --> 00:17:41,823
pretty much that it must be. early, early on and then finding
277
00:17:41,919 --> 00:17:45,959
out later that nobody is claiming that it's an AI rogue cyberattack
278
00:17:46,459 --> 00:17:50,335
or something like that. And then getting— Look, it didn't
279
00:17:50,835 --> 00:17:54,199
prevent it. So— This is true. It did nothing to help.
280
00:17:55,321 --> 00:17:59,569
Like having to read the 3 Laws of Robotics by Asimov.
281
00:18:00,290 --> 00:18:04,378
By inaction, allow human to come to harm. Though, in this particular
282
00:18:04,878 --> 00:18:08,129
case, I'm not sure if Uh, shutting down Outlook
283
00:18:08,530 --> 00:18:11,720
is counting as harm. Correct. Correct.
284
00:18:11,896 --> 00:18:15,808
I, I know I enjoyed a little bit of a break from seeing any
285
00:18:15,904 --> 00:18:19,543
emails come through. I know that I— So it's a lot of alignment, really.
286
00:18:20,168 --> 00:18:23,710
Yes. I think we can count this as progress. It was, uh, it was
287
00:18:23,775 --> 00:18:26,275
a mental health break for many people, uh,
288
00:18:26,724 --> 00:18:28,840
a mandatory mental health break.
289
00:18:30,010 --> 00:18:34,595
So since you are our, our resident Outlook aficionado,
290
00:18:34,819 --> 00:18:37,400
um, do you— Oh, okay. Yeah, I am. All right.
291
00:18:38,398 --> 00:18:41,957
It's definitely not me, so it's gotta be
292
00:18:42,457 --> 00:18:45,840
you. Okay. Do you know what the details of the, uh, of the incident were?
293
00:18:47,028 --> 00:18:50,606
No, no, I, I don't, I don't have too much on the forensics on
294
00:18:50,639 --> 00:18:54,185
that, mostly because, you know, I'm not, I'm not concerned about the,
295
00:18:54,329 --> 00:18:57,443
the why so much as the— or at least the,
296
00:18:57,459 --> 00:19:00,395
the, the minutia of the technical cause.
297
00:19:00,909 --> 00:19:04,792
I am more interested in the fact that Microsoft is
298
00:19:05,486 --> 00:19:09,796
Currently mandatory— doing the voluntary layoffs of 60,000 people
299
00:19:10,597 --> 00:19:14,590
and had one of these events somewhat simultaneously
300
00:19:14,622 --> 00:19:18,727
to that event, wondering what they have in their infrastructure
301
00:19:19,497 --> 00:19:22,480
that they are ill-equipped to support,
302
00:19:23,041 --> 00:19:26,665
that they are banking on AI replacing.
303
00:19:26,729 --> 00:19:30,513
I'm not saying that AI was responsible or caused this or anything like
304
00:19:30,578 --> 00:19:34,443
that. I'm simply wondering. How much of their future planning is
305
00:19:34,508 --> 00:19:38,439
them banking on outsourcing some responsibility that used
306
00:19:38,939 --> 00:19:42,531
to be in the human capital department out to AI in order
307
00:19:43,031 --> 00:19:46,173
to find out that they don't know as much about the infrastructure as they thought
308
00:19:46,286 --> 00:19:50,618
they did? I'm wondering— Well, given that most other engineers
309
00:19:51,118 --> 00:19:54,710
are probably still working there, presumably they— well, the 60,000 are phased
310
00:19:55,210 --> 00:19:58,706
out and whatnot, phased over time and blah, blah, blah. So I think we can
311
00:19:58,738 --> 00:20:02,832
count this on Team Human. And therefore even increase
312
00:20:03,332 --> 00:20:06,428
the layoffs. I think this is how you need to
313
00:20:06,928 --> 00:20:10,667
play it as management. You just tell your humans
314
00:20:11,245 --> 00:20:14,473
every incidence of downtime— Every mistake,
315
00:20:14,874 --> 00:20:18,872
we put on a new AI agent. We add another 5,000
316
00:20:19,372 --> 00:20:24,669
to the layoffs, yeah. Yes, okay, well, you know, the CEO,
317
00:20:26,178 --> 00:20:29,742
What's his last name
again? Um, Satya Natella.
318
00:20:29,887 --> 00:20:33,307
Natella. Okay. Yeah, I was, I was
319
00:20:33,355 --> 00:20:37,112
merging my, uh, this was a duck moment. This is not supposed to— you are
320
00:20:37,612 --> 00:20:40,404
not supposed to take this seriously. Okay. Well, I listen, I asked for help.
321
00:20:40,484 --> 00:20:43,101
I'm not, I'm not gonna question you after the fact, Alex.
322
00:20:43,696 --> 00:20:46,843
I'm getting my consonants all mixed up.
323
00:20:46,907 --> 00:20:50,279
Satya, Satya, Satella. Anyway.
324
00:20:51,482 --> 00:20:55,458
All right. Okay. So anyway, he recently had
325
00:20:55,570 --> 00:20:57,365
been promoting, uh,
326
00:20:58,840 --> 00:21:02,880
uh, oh, what's their agent? Their AI agent
327
00:21:03,380 --> 00:21:07,192
that I'm now blanking on. Uh, Copilot. Your favorite. Copilot. Yes. Copilot. Yes. The worst,
328
00:21:07,272 --> 00:21:10,927
the worst of all of the AI agent harnesses. Uh, Copilot has,
329
00:21:11,183 --> 00:21:14,566
is absolutely miserable. He's been trying to hype it by saying that he, you know,
330
00:21:14,790 --> 00:21:18,846
uses Copilot all throughout the day to help him manage his busy schedule
331
00:21:18,894 --> 00:21:22,549
and help him write email, respond to emails and, and, and do all of
332
00:21:22,565 --> 00:21:26,062
his, uh, calendaring and, and things like that. So since
333
00:21:26,110 --> 00:21:28,549
Copilot was down for the day,
334
00:21:29,464 --> 00:21:32,801
I wonder if he could, you know, possibly be receiving an email from
335
00:21:32,914 --> 00:21:35,529
Elon snarkily asking him, what did you get done today?
336
00:21:36,845 --> 00:21:40,631
Just like Elon did when he came in to work for Doge. That's right.
337
00:21:40,936 --> 00:21:44,049
That's right. Did you accomplish anything today? Because Copilot was down.
338
00:21:44,306 --> 00:21:47,852
Hopefully I won't have any brain farts going through, but I
339
00:21:48,352 --> 00:21:51,205
just blanked out on his name and Copilot's name, and Copilot was easier to remember.
340
00:21:53,981 --> 00:21:57,640
I mean, you know, the name itself is like, you know, I won't take responsibility
341
00:21:58,140 --> 00:22:01,989
for piloting. I'm just gonna sit here on the side and bother you. I mean,
342
00:22:02,054 --> 00:22:05,793
when you start out that way, I mean, it's not gonna get any better.
343
00:22:06,772 --> 00:22:09,628
All right. Now, Alex, we're moving into a new,
344
00:22:10,061 --> 00:22:13,464
new segment here. We're gonna be talking about some of the midterm fallout and
345
00:22:13,785 --> 00:22:17,363
some adjacent paths that we are gonna go from the
346
00:22:17,863 --> 00:22:21,515
midterm fallout. Now, Alex, do you recall one of my main
347
00:22:22,108 --> 00:22:25,729
points from the aftermath of the midterms?
348
00:22:25,793 --> 00:22:29,367
Do you recall what one of my, basically my central thesis was
349
00:22:29,960 --> 00:22:33,934
in the aftermath? No, go for it. Okay. Central thesis was basically
350
00:22:34,527 --> 00:22:37,860
that there is a strong core of
351
00:22:37,956 --> 00:22:41,866
Democrat voters, progressives mostly, but Democrat voters
352
00:22:42,459 --> 00:22:47,159
who have been entrenched in propaganda since 2016
353
00:22:47,191 --> 00:22:50,931
that has had their passions for revenge rising
354
00:22:51,044 --> 00:22:54,575
and rising and rising the entire time. And that
355
00:22:55,075 --> 00:22:58,476
the Democratic win, the blue team's win in this midterm election,
356
00:22:58,508 --> 00:23:01,847
was not simply just because of progressives
357
00:23:02,345 --> 00:23:05,876
in those networks. And so the desire for
358
00:23:06,376 --> 00:23:09,873
revenge is going to be met with competition for the people
359
00:23:09,889 --> 00:23:13,389
who actually pushed them over the edge, actually wanting them
360
00:23:13,517 --> 00:23:17,339
to handle things like affordability
361
00:23:17,435 --> 00:23:21,032
that they were promising to vote for, because they switched teams
362
00:23:21,080 --> 00:23:24,821
this election to vote the blue people in because they were promised
363
00:23:25,321 --> 00:23:28,578
certain things about the cost of living by the Trump administration,
364
00:23:28,691 --> 00:23:31,212
and that these things were going to be in contention and that this is going
365
00:23:31,712 --> 00:23:34,391
to be— Well, guess what, Alex? The rest of the news is picking up with
366
00:23:35,097 --> 00:23:38,804
my central thesis. We have here from Reuters, After midterm sweep,
367
00:23:38,868 --> 00:23:42,203
Democrats split over Trump and cost-cutting agenda. Fresh from
368
00:23:42,251 --> 00:23:45,986
a midterm election landslide, Democrats met in Denver this week, united in wanting
369
00:23:46,018 --> 00:23:49,400
to deliver voters relief from high costs, but split over how to achieve
370
00:23:49,900 --> 00:23:53,440
that, what lessons to draw from their victory, and how combative they should be towards
371
00:23:53,472 --> 00:23:56,886
President Donald Trump. Despite the celebratory mood after
372
00:23:56,918 --> 00:24:00,749
the November 3rd rout of Republicans, tensions between moderates
373
00:24:00,830 --> 00:24:04,741
and a restive progressive wing surfaced at the party's first major gathering
374
00:24:04,773 --> 00:24:08,588
since its victory, along with anxiety that the party could squander the opportunity
375
00:24:08,620 --> 00:24:11,974
voters had handed candidate. The gathering offered an early glimpse
376
00:24:12,474 --> 00:24:16,549
of the battle to define its future as prospective 2028 presidential candidates and
377
00:24:17,049 --> 00:24:20,257
senior leaders debated how to turn electoral success into a
378
00:24:20,757 --> 00:24:24,077
governing agenda while navigating longstanding ideological and generational
379
00:24:24,109 --> 00:24:27,625
divides. The conference, hosted by Majority Democrats,
380
00:24:27,641 --> 00:24:31,236
a group that
advocates for younger party leaders and against left-wing activism,
381
00:24:31,637 --> 00:24:35,185
offered a forum for the jockeying now underway as Democrats debate
382
00:24:35,685 --> 00:24:39,149
what lessons to draw from their midterm win. And begin the search for a 2028
383
00:24:39,197 --> 00:24:42,391
White House standard bearer. Speakers disagreed over
384
00:24:42,891 --> 00:24:46,708
whether to embrace a bolder progressive agenda or tack towards the center, exposing competing
385
00:24:46,740 --> 00:24:50,239
approaches to lowering food, health care, and housing costs while working
386
00:24:50,335 --> 00:24:54,829
alongside Trump, who retains veto power for 2 more years. California Representative
387
00:24:54,909 --> 00:24:58,728
Ro Khanna, a progressive and potential presidential candidate, urged Democrats—
388
00:24:59,081 --> 00:25:02,708
Democrats to stop chasing the political center and embrace a bolder agenda,
389
00:25:02,772 --> 00:25:05,982
including universal health care, a wealth tax, a $25
390
00:25:06,255 --> 00:25:09,561
minimum wage, deep Pentagon cuts, and abolishing U.S.
391
00:25:09,609 --> 00:25:12,530
Immigration and Customs Enforcement. By contrast,
392
00:25:12,819 --> 00:25:16,558
Pennsylvania Governor Josh Shapiro, a moderate and another presidential 20—
393
00:25:16,912 --> 00:25:20,089
potential 2028 contender, argued the party should
394
00:25:20,137 --> 00:25:24,021
move beyond ideological litmus tests, saying voters don't give a damn about such
395
00:25:24,069 --> 00:25:27,873
debates and care instead about whether Democrats can deliver results. The breadth
396
00:25:27,921 --> 00:25:31,404
of the Democratic victories had allowed moderates and progressives to argue that voters
397
00:25:31,452 --> 00:25:34,964
have endorsed their vision for the party's future. Quote, it's to be expected there
398
00:25:35,464 --> 00:25:38,281
will be a big debate because the Democratic victory was so broad. It swept in
399
00:25:38,329 --> 00:25:42,463
centrists, Southern Democrats, and more progressive and left-wing Democrats, Democratic strategist
400
00:25:42,559 --> 00:25:44,386
Elaine Carmack told Reuters.
401
00:25:45,844 --> 00:25:49,641
Trump retains significant executive authority to block much of the Democrats' agenda,
402
00:25:49,865 --> 00:25:53,695
complicating the party's efforts to lower costs, the central issue voters gave
403
00:25:54,195 --> 00:25:57,653
Democrats control of Congress to fix. Democrats are projected to hold at least
404
00:25:58,153 --> 00:26:01,659
255 seats to Republicans' 169. In the U.S.
405
00:26:01,723 --> 00:26:05,008
House, with 12 contests still undecided in the Senate,
406
00:26:05,152 --> 00:26:08,870
and the Senate breakdown will be at least 55 to 44, with Democrats
407
00:26:08,918 --> 00:26:13,037
possibly gaining one more seat in Alaska when vote counting is completed.
408
00:26:14,447 --> 00:26:18,052
One of the most striking debates in Denver centered on how to govern
409
00:26:18,552 --> 00:26:21,738
with Trump still in the White House. James Talarico, the first Democrat
410
00:26:22,238 --> 00:26:25,600
elected to the U.S. Senate from Texas since 1988, urged Democrats to work
411
00:26:26,100 --> 00:26:29,142
with Trump. Quote, we should make the final 2 years of his presidency the most
412
00:26:29,190 --> 00:26:32,539
productive of any presidency in American history, Talarico said.
413
00:26:32,812 --> 00:26:35,906
Drawing tepid applause. We're tired of being pitted against each
414
00:26:35,938 --> 00:26:39,352
other. He found an ally in Iowa Governor-elect Rob Sand,
415
00:26:39,609 --> 00:26:43,360
who said he encouraged Democrats at the campaign events to applaud Republicans and that
416
00:26:43,392 --> 00:26:46,758
partisan warfare will not deliver results. Quote, the only way we're going to fix this
417
00:26:47,258 --> 00:26:51,198
is by fixing it together, Sand said. Minutes later, U.S. Representative Jamie Raskin
418
00:26:51,231 --> 00:26:54,260
of Maryland and Eric Holder, who was attorney general in the Obama administration,
419
00:26:54,757 --> 00:26:58,075
largely rejected accommodation with Trump. Now keep that in mind here,
420
00:26:58,123 --> 00:27:00,768
Jamie Raskin, that little paragraph that's going to come up a little bit later.
421
00:27:00,849 --> 00:27:04,536
Raskin expected to become chairman of the House Judiciary Committee in January and
422
00:27:05,036 --> 00:27:08,255
oversee investigations of Trump next year, said the Republican Party had become
423
00:27:08,755 --> 00:27:12,807
an authoritarian cult of personality. We're going to, quote, we're going to reassert
424
00:27:12,871 --> 00:27:17,215
all of the powers of Congress that have been surrendered and abdicated by MAGA
425
00:27:17,715 --> 00:27:21,287
Congress to King Trump, Raskin said. We're going to enforce the Constitution against
426
00:27:21,787 --> 00:27:25,102
him. Holder warned Democrats against pursuing bipartisanship.
427
00:27:26,144 --> 00:27:29,302
Quote, if we retreat into caution, if we operate out
428
00:27:29,802 --> 00:27:33,614
of fear, if we take refuge in the niceties of old Washington and seek approval
429
00:27:33,646 --> 00:27:36,708
from those who have been and are actively dismantling our institutions, solutions,
430
00:27:36,756 --> 00:27:39,964
we will fail. There has been a long simmering debate inside the
431
00:27:40,464 --> 00:27:43,107
party over whether to move further left or to pursue a more centrist agenda,
432
00:27:43,428 --> 00:27:45,881
and this election has only intensified those tensions.
433
00:27:46,860 --> 00:27:50,436
Almost everyone who spoke at the conference agreed the party's top priority was lowering
434
00:27:50,484 --> 00:27:54,269
costs and improving life for working families. It was a theme echoed by
435
00:27:54,769 --> 00:27:58,775
New York Representative Hakeem Jeffries, set to become U.S. House Speaker, and several prospective
436
00:27:58,807 --> 00:28:04,244
2028 presidential candidates, including Shapiro, former Transportation
437
00:28:04,744 --> 00:28:07,794
Secretary Pete Buttigieg and Khanna. But splits emerged over how to achieve that.
438
00:28:07,971 --> 00:28:11,584
Jeffries, a moderate, was heckled by a group of protesters who accused him of
439
00:28:12,084 --> 00:28:15,196
being too timid to deliver change and called for a House leader who embraced a
440
00:28:15,696 --> 00:28:19,274
more populist, progressive agenda. Khanna and Maryland Senator Chris
441
00:28:19,307 --> 00:28:22,823
Van Hollen, another progressive who this week challenged more centrist
442
00:28:23,323 --> 00:28:26,789
Senator Chuck Schumer of New York for the party's top leadership job in the Senate,
443
00:28:27,062 --> 00:28:29,855
called for populist government funded economic solutions.
444
00:28:30,930 --> 00:28:34,349
There was a collective anxiety among the speakers that the party
445
00:28:34,381 --> 00:28:37,382
might squander one of the biggest victories it has achieved in a generation.
446
00:28:37,896 --> 00:28:41,138
Many speakers described the win not as an enduring mandate, but rather
447
00:28:41,638 --> 00:28:44,589
a temporary opportunity voters had given them to fix the country's problems.
448
00:28:44,974 --> 00:28:48,649
Arizona Senator Ruben Gallego, another potential presidential
449
00:28:49,149 --> 00:28:53,111
contender, noted that many Hispanic voters who voted for Trump in 2024
450
00:28:53,159 --> 00:28:56,562
backed Democrat this month. Quote, it was not, though, a love
451
00:28:56,674 --> 00:28:59,723
letter to the Democratic Party, Gallego said. In a short term, it's a
452
00:29:00,223 --> 00:29:02,949
short-term loan we need to keep paying or we will lose it. Now, Alex,
453
00:29:02,965 --> 00:29:06,640
I told you to pin that Raskin situation because I
454
00:29:06,705 --> 00:29:10,219
want to, I want to play a clip from Raskin on election
455
00:29:10,332 --> 00:29:13,991
night and a promise that he was making on election night as the Democrats
456
00:29:14,087 --> 00:29:17,120
were certified to have been taking the House and the Senate.
457
00:29:17,698 --> 00:29:21,325
We will not hesitate to go after the private
458
00:29:21,825 --> 00:29:25,256
corporations that are collaborating in all of the lawlessness with
459
00:29:25,756 --> 00:29:29,283
Donald Trump. And we've done it before and we will absolutely do
460
00:29:29,783 --> 00:29:33,486
it again. The law firms you mentioned, we know that there were many
461
00:29:33,678 --> 00:29:37,929
heroic law firms that have stood up against
462
00:29:38,395 --> 00:29:41,860
all of the lawbreaking and then the attempt to shake
463
00:29:42,360 --> 00:29:45,726
them down for the time of their partners and their young
464
00:29:45,790 --> 00:29:49,688
associates. And yet there are major law firms in America,
465
00:29:49,704 --> 00:29:53,938
including Paul Weiss, that have capitulated
466
00:29:54,355 --> 00:29:57,498
to these
coercive tactics by Donald Trump and have basically
467
00:29:57,998 --> 00:30:01,572
surrendered and made all kinds of outrageous concessions. We've sent
468
00:30:02,072 --> 00:30:05,517
a number of letters to those law firms, and they have been not
469
00:30:06,017 --> 00:30:09,991
forthcoming up until now. But when we have the power of subpoena and we
470
00:30:10,491 --> 00:30:14,016
have the power to conduct oversight, we believe a
471
00:30:14,081 --> 00:30:17,241
lot of these law firms are going to be coming forward in telling the truth.
472
00:30:17,610 --> 00:30:21,028
Otherwise, they are law firms that have no respect for the rule of law,
473
00:30:21,220 --> 00:30:24,623
and we're going to make sure that they comply with a congressional subpoena.
474
00:30:24,703 --> 00:30:28,651
It is the exact same legal force as a judicial subpoena does.
475
00:30:28,699 --> 00:30:32,101
Subpoena means under the power or under the penalty of law.
476
00:30:33,529 --> 00:30:37,413
So of course, Alex, I would interpret that, and many interpret that, as echoes
477
00:30:37,525 --> 00:30:41,907
of retribution, right? Speaking of retribution, going after companies
478
00:30:41,939 --> 00:30:45,962
that were partnering with Trump in some way. What is he talking about?
479
00:30:46,010 --> 00:30:50,111
That is— that was a very weird and tortured use of language
480
00:30:50,611 --> 00:30:55,895
there. What? Well, you know, they have resisted the demands
481
00:30:55,959 --> 00:30:59,484
for the use of the time of their partners and associates.
482
00:30:59,644 --> 00:31:03,553
What? He means that they have represented Trump? Is that
483
00:31:04,053 --> 00:31:07,543
what his problem is? There's— well, there's certain things where certain
484
00:31:07,639 --> 00:31:11,068
law firms had cooperated with Trump on certain things,
485
00:31:11,164 --> 00:31:14,387
and I guess they're getting They're going to
486
00:31:14,887 --> 00:31:17,355
be in trouble. They want to know, I guess, what conversations were had or something
487
00:31:17,419 --> 00:31:21,269
along those lines. There was— What conversations
488
00:31:21,769 --> 00:31:24,976
were had between a client and his attorney? Possibly. I don't
489
00:31:25,476 --> 00:31:28,810
know. I'm going to try to look to see what more he
490
00:31:28,890 --> 00:31:33,383
wants. But outside of the law
491
00:31:33,883 --> 00:31:37,361
firms, he's talking about private corporations, basically. So any corporation that
492
00:31:37,410 --> 00:31:40,830
was cooperating with Trump in
493
00:31:40,926 --> 00:31:46,017
any way is going to be under their oversight. It's going to be held to—
494
00:31:46,065 --> 00:31:50,112
Here's the thing. I think my understanding anyway of
495
00:31:50,160 --> 00:31:53,581
the voting public in the US is that people are
496
00:31:53,693 --> 00:31:58,415
pissed at Trump. The people who not only did
497
00:31:58,768 --> 00:32:02,269
prior MAGA voters that not only stayed home,
498
00:32:02,301 --> 00:32:06,078
but actually got up and went and voted blue, have no
499
00:32:06,158 --> 00:32:09,251
problems with Trump being prosecuted for,
500
00:32:09,944 --> 00:32:12,811
you know, all sorts of things, quite frankly. Like, some of the crypto stuff is
501
00:32:13,085 --> 00:32:15,518
wild, and there's, there's more.
502
00:32:17,193 --> 00:32:20,592
But that's one thing, okay? Especially if
503
00:32:21,092 --> 00:32:23,460
they do it properly, as in, you know,
504
00:32:25,397 --> 00:32:28,530
not, not do the thing where they invent some fucking hoax
505
00:32:28,562 --> 00:32:32,677
to go after, like, Russiagate, but go after the actual things. things that
506
00:32:33,398 --> 00:32:37,740
have been done. And I think pretty much most
507
00:32:38,240 --> 00:32:41,441
of the same people are, even on the MAGA, formerly on
508
00:32:41,941 --> 00:32:46,553
the MAGA side, are pretty sure there's smoke and fire.
509
00:32:47,594 --> 00:32:51,103
But that's one thing. The other, to say that anybody who provided
510
00:32:51,603 --> 00:32:54,933
you with representation, we're gonna go after you, and he named one
511
00:32:55,029 --> 00:32:58,474
firm as well. I don't know what that was, man.
512
00:32:58,650 --> 00:33:02,957
That's something else. Now, in the— in the— Oddly
513
00:33:02,989 --> 00:33:06,664
specific, you know what I mean? It wasn't like the crypto firms and the lawyers
514
00:33:06,696 --> 00:33:10,180
and the this and the that and the, you know, the Russian businessman. He was,
515
00:33:10,309 --> 00:33:12,782
he was going after a specific law firm.
516
00:33:14,916 --> 00:33:18,329
What? That was wild. Well,
517
00:33:18,377 --> 00:33:22,175
remember, when the opposition is in power, it's authoritarianism,
518
00:33:22,191 --> 00:33:25,989
but when you're in power, it's justice. You're pursuing justice and upholding the rule of
519
00:33:26,489 --> 00:33:29,578
law. That's right. That's what you do. Now,
520
00:33:30,686 --> 00:33:34,043
in light of that article that I just read, Alex, where it's like there's a
521
00:33:34,543 --> 00:33:38,202
lot of contention because the people who have been really ensconced
522
00:33:38,378 --> 00:33:41,976
in the outrage propaganda for 10 years want heads
523
00:33:42,072 --> 00:33:45,428
to roll. They want revenge. They do not want anyone
524
00:33:45,525 --> 00:33:48,913
who participated in the Trump administration to get away free. They want every
525
00:33:49,413 --> 00:33:52,928
single person who even said something nice about Trump to at least get
526
00:33:53,428 --> 00:33:56,861
a little jail time. But the rest of the voters, as we've mentioned, the rest
527
00:33:57,361 --> 00:34:00,552
of the public, they don't necessarily want that. And so this week,
528
00:34:00,793 --> 00:34:04,003
Jamie Raskin was asked somewhat directly and gave a little bit of a different
529
00:34:04,503 --> 00:34:08,465
tune. The American people didn't necessarily vote for retribution.
530
00:34:08,561 --> 00:34:12,044
So where is that line as you see it between— No, there's no
531
00:34:12,092 --> 00:34:16,201
line. We're all— the line is essentially
532
00:34:16,701 --> 00:34:21,465
between everything that happened on Tuesday after from what happened
533
00:34:21,965 --> 00:34:25,581
before, because the entire Trump Department of Justice has been based on the principle
534
00:34:25,918 --> 00:34:29,020
of revenge and retribution against Donald Trump's enemies.
535
00:34:29,132 --> 00:34:32,764
19 different grand juries have refused to bring charges
536
00:34:33,264 --> 00:34:37,006
against people that they want to bring charges against. We're talking about James Comey,
537
00:34:37,472 --> 00:34:41,152
Letitia James, Adam Schiff, Mark Kelly,
538
00:34:41,666 --> 00:34:44,864
Maggie Goodlander, Chrissy Houlihan.
539
00:34:45,938 --> 00:34:49,285
You know, it just happened the other day with Cassidy Hutchinson. It happened to one
540
00:34:49,785 --> 00:34:53,004
of my constituents they tried to set up on the reflecting pool job. Again,
541
00:34:53,344 --> 00:34:55,683
nothing like this in American history has ever happened before.
542
00:34:58,170 --> 00:35:01,166
So he's promising no retribution now. He's, nope, nope,
543
00:35:01,247 --> 00:35:05,177
that's them, not us. So that's, that's a different tune. But I'm thinking
544
00:35:05,193 --> 00:35:08,518
he's secretly still going to be pushing for retribution because
545
00:35:09,018 --> 00:35:11,251
that's kind of— I still think we need to figure out what the fuck is
546
00:35:11,267 --> 00:35:13,758
going on with those law firms and what exactly he was talking about. He said
547
00:35:14,258 --> 00:35:18,113
they wrote a letter Can we find that letter? Because I, I think
548
00:35:18,613 --> 00:35:22,174
there's something a lot more specific about what he was talking about in
549
00:35:22,223 --> 00:35:25,546
that little rant of his. Well, I'll try to flag a follow-up for that then,
550
00:35:25,562 --> 00:35:28,804
Alex. Yeah, I'll try to— it was like a
551
00:35:29,304 --> 00:35:33,283
follow-up. Such an odd little, little piece of— I don't—
552
00:35:33,379 --> 00:35:36,750
is this guy a lawyer? Probably. Let's just assume. Um, I don't
553
00:35:37,601 --> 00:35:40,715
know, there's some kind of inside
554
00:35:40,747 --> 00:35:44,372
baseball going on that I've not I'm not fully satisfied we've
555
00:35:44,757 --> 00:35:48,235
gotten to the bottom of there. Interesting. Very interesting piece. I will pin
556
00:35:48,267 --> 00:35:51,698
that to follow up, Alex. I will try to do my best to have some
557
00:35:51,714 --> 00:35:55,224
material next week on that angle. Now, one thing that they have
558
00:35:55,256 --> 00:35:59,280
been floating in these trial balloons, these whiffs
559
00:35:59,344 --> 00:36:02,694
in the air, this is from The Washington Post. Movement to
560
00:36:02,775 --> 00:36:05,804
end Electoral College gets boost after Democratic wins.
561
00:36:07,039 --> 00:36:10,485
A movement to give the presidency to whoever receives the most votes
562
00:36:10,790 --> 00:36:14,461
nationally has new momentum after Democrats swept state races
563
00:36:14,557 --> 00:36:18,006
in the midterm elections, but still faces major hurdles in legislative
564
00:36:18,506 --> 00:36:22,177
chambers and courtrooms. For 2 decades, reformers have pressed states to agree
565
00:36:22,677 --> 00:36:25,931
to give their electoral votes to the winner of the national popular vote,
566
00:36:26,188 --> 00:36:29,669
regardless of who wins their state. The agreement would
567
00:36:29,717 --> 00:36:33,102
render the Electoral College moot and end aberrations like the
568
00:36:33,166 --> 00:36:36,647
ones in 2000 and 2016 that saw Republicans George W.
569
00:36:36,728 --> 00:36:40,161
Bush and Donald Trump win the presidency despite Democrats receiving more
570
00:36:40,661 --> 00:36:43,872
votes. There is no other election in this country where the person
571
00:36:43,888 --> 00:36:47,191
who loses can win the election, where the person who gets the most
572
00:36:47,691 --> 00:36:51,049
votes doesn't actually win, said Wisconsin State Kelda Roys.
573
00:36:51,469 --> 00:36:55,302
And it's fundamentally unfair and anti-democratic, small d democratic. Now, Alex,
574
00:36:55,351 --> 00:36:58,811
I would like to point out that there still isn't an election where the person
575
00:36:58,908 --> 00:37:02,790
who loses wins. It's the person
576
00:37:03,290 --> 00:37:06,591
who won that won, and they want to have a different way of deciding who
577
00:37:06,655 --> 00:37:09,874
won. I mean, that's a It's a very specific claim that's being
578
00:37:09,906 --> 00:37:12,615
made here. Yep,
579
00:37:13,288 --> 00:37:16,975
yep. And I mean, there are many elections where the person who loses wins.
580
00:37:17,280 --> 00:37:21,544
Anything that has to do with caucuses or counties, or like,
581
00:37:22,265 --> 00:37:25,102
it's, it's not hard to understand. Like, you know,
582
00:37:25,439 --> 00:37:28,693
um, for example, uh, who gets the congressional
583
00:37:28,966 --> 00:37:32,669
majority in the House and Senate? They don't count votes.
584
00:37:33,165 --> 00:37:36,306
I don't know if you've heard about gerrymandering and whatnot. There's a reason for
585
00:37:36,806 --> 00:37:40,279
this. It, it doesn't change the total number
586
00:37:40,779 --> 00:37:44,108
of votes. Hint, hint. Yeah. Yeah. Finding a
587
00:37:44,608 --> 00:37:48,354
way to sidestep the Electoral College once seemed all but impossible, but with Democrats winning
588
00:37:48,434 --> 00:37:51,830
full control of more state governments last week, it now appears within reach,
589
00:37:51,878 --> 00:37:55,419
at least politically. If supporters get their way, they will be met with a
590
00:37:55,919 --> 00:37:58,911
slew of litigation, and any one lawsuit could block their path.
591
00:37:59,424 --> 00:38:02,532
Such a change would transform
the country politically, reshaping who can
592
00:38:03,032 --> 00:38:06,730
mount viable presidential bids and how they run their campaigns. Instead of concentrating
593
00:38:06,778 --> 00:38:10,162
on half a dozen or so swing states, they would seek votes across the
594
00:38:10,662 --> 00:38:13,978
nation. Supporters say that that would guarantee presidential candidates find broad
595
00:38:14,010 --> 00:38:17,778
backing. Detractors say they would ignore smaller states and spend most of their time
596
00:38:18,278 --> 00:38:22,108
in the most populous states— California, Texas, and Florida. Opponents say those who want
597
00:38:22,608 --> 00:38:26,566
to overhaul the Electoral College must go through the arduous process of amending the Constitution
598
00:38:26,678 --> 00:38:29,885
rather than rewriting the rules with state laws. Under the
599
00:38:30,385 --> 00:38:34,086
Constitution, states decide how to award their electoral votes. With most giving all
600
00:38:34,150 --> 00:38:37,178
of theirs to whoever wins their state.
601
00:38:37,723 --> 00:38:41,473
To get around that setup, reformers have asked states to agree to give their electoral
602
00:38:41,505 --> 00:38:45,479
votes to whoever wins the national popular vote. The agreement wouldn't go into effect
603
00:38:45,511 --> 00:38:49,165
until states with at least 270 electoral votes needed to win the presidency
604
00:38:49,181 --> 00:38:52,514
join. Over the last 20 years, 18 states and Washington,
605
00:38:52,546 --> 00:38:56,600
D.C. have signed onto the agreement. Together, they have 222 electoral
606
00:38:56,633 --> 00:38:58,507
votes. Now, Alex, historically speaking,
607
00:38:59,943 --> 00:39:03,543
I want to point out that this is another one of those things that the
608
00:39:03,623 --> 00:39:07,408
early progressive era purposely destroyed so that it no longer made
609
00:39:07,908 --> 00:39:10,519
sense, so that they could later make the argument that it made no sense and
610
00:39:11,019 --> 00:39:14,672
get rid of it entirely. So I want to point
611
00:39:15,172 --> 00:39:19,226
out that the original sort of process for the Electoral College
612
00:39:19,659 --> 00:39:23,540
was more or less supposed to mirror the election
613
00:39:23,588 --> 00:39:26,814
of the House of Representatives. But what you were doing was you
614
00:39:26,863 --> 00:39:30,621
were electing a person whom would engage
615
00:39:30,798 --> 00:39:34,733
in the process of the Electoral College sometime later
616
00:39:35,233 --> 00:39:37,640
to then choose who the president would become.
617
00:39:39,069 --> 00:39:42,908
Okay? So the process was an intellectual
618
00:39:43,021 --> 00:39:46,602
delegation. What you basically would be doing in concept
619
00:39:46,875 --> 00:39:50,200
would be going, this is the wisest person in our community. This is the
620
00:39:50,473 --> 00:39:54,320
most informed, confirmed brightest person. This person in our
621
00:39:54,820 --> 00:39:58,278
community is the one we are going to elect to have
622
00:39:58,778 --> 00:40:02,011
the knowledge and wisdom to vote for the president and choose the president as part
623
00:40:02,511 --> 00:40:06,514
of the myriad of balances of power that the
624
00:40:07,014 --> 00:40:10,648
Constitution was intending to set up. Right. And effectively,
625
00:40:11,065 --> 00:40:14,526
what happened was the progressive push started to
626
00:40:14,830 --> 00:40:18,291
change the state's behaviors so that you did not
627
00:40:18,791 --> 00:40:22,586
put the electors really as the people you were voting for. You had electors
628
00:40:22,698 --> 00:40:26,272
pledge to vote for the president of the party's
629
00:40:26,384 --> 00:40:29,573
selection. And then the presidents, the people who were running for
630
00:40:30,073 --> 00:40:33,002
president under the parties were the ones that you were voting for. And then as
631
00:40:33,034 --> 00:40:36,159
a technicality, when you vote for president,
632
00:40:37,105 --> 00:40:40,406
your vote is being counted towards a pledged elector.
633
00:40:41,511 --> 00:40:44,716
So they completely destroyed the original intent of the whole
634
00:40:45,216 --> 00:40:49,332
thing, made it convoluted, made it so that people are going through the mirage of
635
00:40:49,380 --> 00:40:52,653
voting for a president, They are going through the practice of
636
00:40:52,686 --> 00:40:55,939
selecting a person that they are choosing for president.
637
00:40:56,327 --> 00:40:59,855
Their vote goes to the elector who promised to vote for that guy for
638
00:40:59,871 --> 00:41:03,307
president. And now they say this system
639
00:41:03,807 --> 00:41:06,101
is antiquated and doesn't make sense because they freaking broke it.
640
00:41:07,804 --> 00:41:11,273
Right. So it doesn't make sense. Well, here's the thing though, right?
641
00:41:12,144 --> 00:41:15,193
I'm gonna try to argue the other side of this. Oh boy. Not because I'm
642
00:41:16,032 --> 00:41:20,348
fully agreeing with it, but because I like to see you get
643
00:41:20,898 --> 00:41:23,194
worked up. So,
644
00:41:24,439 --> 00:41:27,915
you know, arguably when the original Constitution
645
00:41:28,415 --> 00:41:33,387
was set up, the US was a lot more actually
646
00:41:33,484 --> 00:41:37,261
like a coalition of nations, right? Like the European
647
00:41:37,293 --> 00:41:40,350
Union is in a way. I mean, the European Union is weaker.
648
00:41:40,912 --> 00:41:44,623
There was a strong concern for federal concentration of power,
649
00:41:44,768 --> 00:41:48,062
and the federal government was there in concept
650
00:41:48,898 --> 00:41:53,069
argued against just a strict confederation where it
651
00:41:53,569 --> 00:41:56,473
was just a mutual agreement among states,
652
00:41:57,163 --> 00:42:01,049
partially because of the concern over not
653
00:42:01,097 --> 00:42:04,324
having the ability to have a coherent army for defense,
654
00:42:05,111 --> 00:42:08,804
largely. Right, right. So for example, I remember one
655
00:42:08,820 --> 00:42:11,598
statement that— I don't know as much of the US history as you do,
656
00:42:11,614 --> 00:42:15,706
but I remember one interesting statement by Robert E. Lee that,
657
00:42:15,786 --> 00:42:19,619
you know, he kind of, you know, both sides in the Civil War
658
00:42:20,119 --> 00:42:24,334
wanted him to join their sides. I guess he was considered a very competent general
659
00:42:25,233 --> 00:42:28,937
before the war broke out. But, and maybe
660
00:42:29,001 --> 00:42:32,081
his personal loyalties laid with the Union more,
661
00:42:32,530 --> 00:42:36,106
like if he were to choose. But the big factor for him
662
00:42:36,154 --> 00:42:39,570
is that Virginia went with the South. So he's like, well, my state is
663
00:42:39,602 --> 00:42:43,067
with the South, so I'm not gonna fight against my state. That was his reasoning
664
00:42:43,115 --> 00:42:46,652
was like— Yeah. my country effectively is on that side,
665
00:42:46,685 --> 00:42:50,224
so I'm on that side. There's no, you know, what I think, you know,
666
00:42:50,241 --> 00:42:54,059
in this sort of political squabble is irrelevant. Travel was obviously much more
667
00:42:55,240 --> 00:42:58,960
difficult in earlier days to make the—
668
00:42:59,121 --> 00:43:02,630
You could take an army and go, though. Well,
669
00:43:02,905 --> 00:43:05,346
you know, I'm just saying it would take, it would take a long time to
670
00:43:05,411 --> 00:43:09,324
travel by horse to another location if you were in the center of
671
00:43:09,824 --> 00:43:12,913
a deep decent state, actually getting from one
672
00:43:13,010 --> 00:43:16,480
end to— from Philadelphia to Ohio was not just
673
00:43:16,980 --> 00:43:20,513
a 6-hour drive. It's like— Right. So I
674
00:43:20,562 --> 00:43:24,097
guess my point is that now I feel like
675
00:43:24,597 --> 00:43:31,472
most people who are US Americans are feeling
676
00:43:31,970 --> 00:43:35,296
like that is their primary national
677
00:43:35,796 --> 00:43:39,772
identity. designation. You don't
678
00:43:40,012 --> 00:43:42,723
feel like a whatever, New Yorker, whatever it is you are,
679
00:43:43,204 --> 00:43:46,749
or a Pennsylvanian, or a this or a that.
680
00:43:46,829 --> 00:43:50,599
There is still a lot. It is not as strong. I will absolutely grant
681
00:43:51,099 --> 00:43:54,529
you that. It is not as strong as it would've been back in the earlier
682
00:43:55,029 --> 00:43:58,362
days, particularly closer to the founding, because each state had its
683
00:43:58,443 --> 00:44:01,554
own constitution and therefore reflected part of
684
00:44:01,586 --> 00:44:04,937
the ethos of the people that formed that state
685
00:44:04,953 --> 00:44:08,767
constitution. So there was a lot more culture embedded
686
00:44:09,344 --> 00:44:13,125
in the state itself for the people that inhabited
687
00:44:13,625 --> 00:44:17,227
it. So there was a stronger connection, but people from Pennsylvania
688
00:44:17,727 --> 00:44:21,216
still shit on people from Ohio. Okay. Of course you do. But so
689
00:44:21,716 --> 00:44:25,126
I'm getting— And everyone makes fun of West Virginia. I can
690
00:44:25,626 --> 00:44:29,372
bring you something from growing up in Greece where we are under the EU and
691
00:44:29,872 --> 00:44:32,978
the EU is this weird convoluted bureaucratic mess. that nobody
692
00:44:33,478 --> 00:44:36,698
fully understands, but it's sort of there and whatever. However, if there
693
00:44:37,198 --> 00:44:40,899
was a war between Greece and Bulgaria, there would be no question as
694
00:44:40,995 --> 00:44:45,116
to, you know, where people's loyalties lie, you know. Yeah, like within
695
00:44:45,180 --> 00:44:48,884
Greece, we shit on other, like, you know, uh, regions of
696
00:44:49,384 --> 00:44:51,690
the country, right? We're like, oh, people from the north, people from the south,
697
00:44:51,754 --> 00:44:55,506
people from the islands, people from the mainland, whatever. Sure, but like,
698
00:44:56,308 --> 00:44:59,498
you know what your country is, right? And, and, and yes, the EU is
699
00:44:59,595 --> 00:45:02,860
there, But that's kind of like NAFTA or something.
700
00:45:02,925 --> 00:45:06,343
Sure. Sure. It's more than that, unfortunately, and people aren't into that.
701
00:45:07,952 --> 00:45:11,979
Part of that is the things that the federal government was
702
00:45:12,043 --> 00:45:15,861
meant to do that it ameliorated certain contentions that
703
00:45:16,167 --> 00:45:19,945
just having a confederacy would do. So for example, in the Federalist
704
00:45:20,445 --> 00:45:23,786
Papers, Hamilton, who is not one of my favorite people,
705
00:45:23,883 --> 00:45:27,329
but Hamilton was Explaining that a
706
00:45:27,377 --> 00:45:31,540
source of contention among states that could cause disunion was
707
00:45:31,572 --> 00:45:35,460
the fact that a state could hypothetically put
708
00:45:35,573 --> 00:45:39,310
import tariffs against another state somewhat
709
00:45:39,810 --> 00:45:43,451
targeted. So several states could gang together and then pick out
710
00:45:43,951 --> 00:45:47,452
a different state and then just start to put tariffs on them so that imports
711
00:45:47,952 --> 00:45:50,887
from that state would be taxed at a different rate and that it would cause
712
00:45:51,387 --> 00:45:54,686
tension. Yeah. between states and could cause war and
713
00:45:55,186 --> 00:45:58,444
conflict. So the federal— that's actually where the Commerce Clause comes
714
00:45:58,620 --> 00:46:02,346
in the federal Constitution. That is why the power is granted to
715
00:46:02,491 --> 00:46:05,783
the federal government to make regular the commerce
716
00:46:06,283 --> 00:46:09,862
between states, which is— the bastardization of the word
717
00:46:09,878 --> 00:46:14,375
regulation gets caught up in there. And unfortunately, that's where the government
718
00:46:14,875 -->
00:46:17,330
usurps a lot of power is through the Commerce Clause. But that was the original
719
00:46:17,830 --> 00:46:21,500
purpose of the Commerce Clause. to somewhat make regular—
720
00:46:21,532 --> 00:46:23,856
And that's why I can't grow food in my backyard.
721
00:46:24,962 --> 00:46:28,776
That is why, yes, you can't grow food in your backyard
722
00:46:29,113 --> 00:46:32,879
without the FDA busting in. Something works.
723
00:46:33,841 --> 00:46:37,431
Yeah, obvious. Anyway, but all I'm saying
724
00:46:37,463 --> 00:46:40,716
is that the part of probably
725
00:46:40,812 --> 00:46:44,963
why the narrative of the Electoral College
726
00:46:45,220 --> 00:46:47,928
and the Senate being what it was and not what it is, et cetera,
727
00:46:47,944 --> 00:46:51,458
et cetera, has weakened end is also because the vibes
728
00:46:51,715 --> 00:46:54,828
of the individual states as
729
00:46:54,940 --> 00:46:58,568
nations have also weakened. Even if
730
00:46:58,616 --> 00:47:01,986
you want to blame the progressive era for it, let's say the past was more
731
00:47:02,486 --> 00:47:06,014
brutal. I certainly do. If we were
732
00:47:06,514 --> 00:47:10,797
participating in the system that was closer to originally prescribed,
733
00:47:10,845 --> 00:47:14,922
I think the sensibilities would be different though. So I believe that the
734
00:47:16,546 --> 00:47:20,312
relaxing of the considerations that you're mentioning is partially due
735
00:47:20,456 --> 00:47:23,761
to the fact that people are no longer considering votes
736
00:47:24,291 --> 00:47:27,948
being a local thing. They are only considering them being a
737
00:47:27,996 --> 00:47:32,135
federal thing, to the point where sometimes whenever it's state senators,
738
00:47:32,264 --> 00:47:36,258
not the federal senators, but state senators doing a
739
00:47:36,758 --> 00:47:40,799
debate on television, on their local networks, people will call in during
740
00:47:40,895 --> 00:47:44,039
the Q&A period and completely confuse the subject matter
741
00:47:44,539 --> 00:47:48,140
with the federal government. And that people are— the way that the news cycle
742
00:47:48,980 --> 00:47:52,214
forms, it's usually federal discussions,
743
00:47:52,376 --> 00:47:55,570
not local discussions, because the federal discussions wind up
744
00:47:56,070 --> 00:47:59,185
usurping so much. But that's sort
745
00:47:59,685 --> 00:48:02,913
of the problem, right? Because ultimately, how much are
746
00:48:02,961 --> 00:48:06,162
people going to try to keep in
747
00:48:06,194 --> 00:48:10,250
their heads at one time? You can try to track national
748
00:48:10,298 --> 00:48:13,838
politics, And maybe you can additionally track state
749
00:48:14,338 --> 00:48:18,168
politics and then try to keep those separate, plus have the right understanding
750
00:48:18,668 --> 00:48:22,417
where they intersect. Fuck, man. You're not wrong. A Supreme
751
00:48:22,917 --> 00:48:26,266
Court probably can't do this consistently. You're not
752
00:48:26,330 --> 00:48:29,762
wrong, but my argument is that because
753
00:48:29,875 --> 00:48:32,360
of the ways that the progressives messed with the system,
754
00:48:32,681 --> 00:48:36,690
both the way they changed the way the Senate operates and this Electoral College
755
00:48:37,190 --> 00:48:40,560
shenanigans, The federal government was allowed to usurp power from the states
756
00:48:41,187 --> 00:48:44,528
in larger doses than it otherwise would've been able
757
00:48:44,560 --> 00:48:48,559
to, and changed the culture of people's focus from
758
00:48:48,640 --> 00:48:51,916
local government to state government— or sorry, federal government only,
759
00:48:52,751 --> 00:48:56,382
thus giving that perspective. It would be— I'm arguing
760
00:48:56,446 --> 00:49:00,188
that. I don't know. Yes. That's my argument. All I can tell you is
761
00:49:00,365 --> 00:49:03,802
under a model of people,
762
00:49:03,995 --> 00:49:07,287
which may be oversimplified, But, you know, what are you gonna do?
763
00:49:07,512 --> 00:49:11,542
Um, where, where you can only really serve one master
764
00:49:11,863 --> 00:49:15,780
in terms of nationality, or, you know, like national psyche
765
00:49:15,797 --> 00:49:18,847
or whatever you want to call it. To the degree that the,
766
00:49:19,650 --> 00:49:24,049
you know, the, the USA as a country becomes
767
00:49:24,549 --> 00:49:28,368
preeminent, like a lot of this other stuff sort of gets dragged along
768
00:49:28,416 --> 00:49:31,756
because, you know, people vote for president, for example, right? And they expect
769
00:49:32,256 --> 00:49:35,754
him to be the emperor, more or less. Anything that doesn't happen right under his
770
00:49:35,850 --> 00:49:39,721
watch, especially after a few months of being in government, is considered now
771
00:49:40,347 --> 00:49:43,655
his fault, even though you can go through and explain how
772
00:49:44,155 --> 00:49:47,992
the Senate and the Congress— And to support
773
00:49:48,492 --> 00:49:51,878
that idea, presidents often, when they're running for election,
774
00:49:52,151 --> 00:49:55,701
talk about things they want to do that are actually Congress's job. They talk
775
00:49:56,201 --> 00:49:57,981
about all these bills they want to pass. How far be it for me to
776
00:49:58,174 --> 00:49:59,893
clear them of their guilt? Yes.
777
00:50:02,104 --> 00:50:06,380
They play their part in this drama. But point being,
778
00:50:06,509 --> 00:50:10,093
people like simple models of the world in their
779
00:50:10,126 --> 00:50:13,663
heads. And seeing the other side
780
00:50:14,163 --> 00:50:18,245
of that in Greece, where we have a quote unquote weak federal authority and a
781
00:50:18,745 --> 00:50:22,859
quote unquote strong local
782
00:50:23,344 --> 00:50:26,745
authority, let's call it a state or a country as we call them here.
783
00:50:28,852 --> 00:50:33,011
It doesn't get better. What happens is nobody fucking
understands. Nobody has any
784
00:50:33,188 --> 00:50:37,186
idea at all, any slightest clue of what happens on the European level.
785
00:50:37,234 --> 00:50:40,928
Like literally people, you could ask them like, how many representatives are in the
786
00:50:41,056 --> 00:50:44,300
EU Parliament? No idea, blank. Who,
787
00:50:44,990 --> 00:50:48,057
how is the European Union president elected?
788
00:50:48,619 --> 00:50:51,782
Blank. I swear to God, nobody knows at all.
789
00:50:53,018 --> 00:50:56,246
Somebody appears at some point and they are like, I am the president
790
00:50:56,746 --> 00:50:59,253
of the European Union. Oh, it's you? Okay, whatever.
791
00:50:59,995 --> 00:51:04,509
That's how it is over there. So I
792
00:51:05,009 --> 00:51:08,368
find it interesting to me to have lived both systems and see the contrast,
793
00:51:08,384 --> 00:51:11,738
but whatever it is, I think people can only really hold one government in their
794
00:51:11,770 --> 00:51:14,719
heads and badly at that.
795
00:51:15,720 --> 00:51:19,433
Well, Alex, let's move forward here because we have
796
00:51:19,449 --> 00:51:22,953
some interesting things that are going to be partially connected.
797
00:51:23,002 --> 00:51:26,717
Okay, Alex, we've talked about Including very recently,
798
00:51:27,873 --> 00:51:31,228
my specific point that there are
799
00:51:31,728 --> 00:51:35,017
a lot of propaganda networks that people have been inundated with day in and
800
00:51:35,517 --> 00:51:39,368
day out, that there's a lot of coordination that goes on, and that these networks
801
00:51:39,868 --> 00:51:42,933
exist initially, hypothetically, under one cause,
802
00:51:42,965 --> 00:51:47,267
but then they are actually not hijacked, but purposely
803
00:51:47,701 --> 00:51:51,409
utilized for alternate mechanisms, that they're— There's
804
00:51:51,909 --> 00:51:54,890
sort of a hydra that is connected,
805
00:51:54,986 --> 00:51:58,355
that they are actually there to serve a larger purpose of just the
806
00:51:58,932 --> 00:52:02,862
political organism. We've talked about this, particularly my specific
807
00:52:03,362 --> 00:52:06,568
example of the Minnesota Ice Watch that for some reason is
808
00:52:06,696 --> 00:52:10,706
anti-flock and also anti-data center, that network
809
00:52:11,187 --> 00:52:14,748
of— Yeah. That's an example I use just to talk about the thing.
810
00:52:14,813 --> 00:52:18,149
Well, Alex, would you believe that Van Jones,
811
00:52:19,196 --> 00:52:22,651
Made an announcement of another network, such network,
812
00:52:22,974 --> 00:52:26,627
in the aftermath of the election last week on CNN. Have you heard
813
00:52:27,127 --> 00:52:30,534
of this announcement? No, please do tell.
814
00:52:31,166 --> 00:52:34,572
I can't wait. Now it can
815
00:52:35,072 --> 00:52:39,171
be told. Now it can be told. How did we win the
816
00:52:39,671 --> 00:52:43,373
narrative on affordability? Was it just these nice little Democrats out there
running for
817
00:52:43,422 --> 00:52:46,921
office and Going to no, the Republicans
818
00:52:47,421 --> 00:52:50,597
never figured out what we were doing. There's something called the American Accountability Network.
819
00:52:51,576 --> 00:52:54,256
Tens of thousands of online humans, not bots,
820
00:52:54,866 --> 00:52:58,333
coordinated to make the affordability message a winning message
821
00:52:58,477 --> 00:53:01,864
online. Thirty-two point five million
822
00:53:03,051 --> 00:53:05,058
people hit on Instagram every day.
823
00:53:06,085 --> 00:53:09,407
Eighty-eight point eight eighty-eight point eight million hit on.
824
00:53:09,856 --> 00:53:13,960
Twitter, X, every single day, and 185
825
00:53:14,460 --> 00:53:18,160
million Facebook hits every single day. And they never figured
826
00:53:18,660 --> 00:53:21,655
it out. The Republicans could not understand it wasn't bots.
827
00:53:21,895 --> 00:53:25,518
It was actual human beings. The most effective online organizing campaign
828
00:53:25,967 --> 00:53:29,333
called the American Accountability Network flew under the radar screen, never been told. And they
829
00:53:29,833 --> 00:53:33,421
were talking about the issue or they were saying you need to elect Democrats
830
00:53:33,921 --> 00:53:38,039
to deal with this? Driving the kitchen table affordability message and tying
831
00:53:38,087 --> 00:53:41,150
that to Democrats on a positive and Republicans on a negative.
832
00:53:41,679 --> 00:53:46,410
The Republicans never figured out what we were doing. And what that meant was everywhere
833
00:53:46,910 --> 00:53:50,516
you go, people were talking about affordability and they were talking about kitchen table
834
00:53:51,016 --> 00:53:54,926
issues. They couldn't get any traction on these cultural war issues anymore because
835
00:53:55,426 --> 00:53:58,967
we figured out how to get around all that. So it's not just all
836
00:53:59,015 --> 00:54:02,129
the door knocking that was done, which is amazing. It's not just our amazing candidate
837
00:54:02,629 --> 00:54:06,332
quality. We had something called the American Accountability Network. Don't get mad
838
00:54:06,832 --> 00:54:10,116
at me, I was given permission to reveal this. They never figured it out,
839
00:54:10,438 --> 00:54:14,466
and that network is going to continue to push for kitchen table issues and
840
00:54:14,966 --> 00:54:18,142
hold our Democratic lawmakers accountable. Now,
841
00:54:18,158 --> 00:54:21,500
from the American Accountability Network website,
842
00:54:21,597 --> 00:54:25,020
Alex, social media strategies for real-world impact.
843
00:54:25,085 --> 00:54:28,249
The American Accountability Network is a powerful consortium of advocates,
844
00:54:28,554 --> 00:54:32,015
content creators, organizers, and influencers who together have formed an
845
00:54:32,143 --> 00:54:35,749
ecosystem that has the ability to shape political narratives, reach the
846
00:54:36,249 --> 00:54:38,681
public at
scale, and mobilize people to take action.
847
00:54:39,771 --> 00:54:42,880
Right-wing misinformation and disinformation. I'm,
848
00:54:43,120 --> 00:54:46,036
I'm not clear from what he said. He, he kind of said a few things.
849
00:54:46,405 --> 00:54:49,754
Did they ever figure it out? Because it wasn't clear from what he was saying
850
00:54:49,850 --> 00:54:52,767
if they figured it out or not. Well, he's telling them now.
851
00:54:53,300 --> 00:54:56,980
Uh, as a deductive thing. Previously, they never figured it out,
852
00:54:57,480 --> 00:55:00,340
right? That's what he's claiming. Yeah, that they didn't, they didn't, uh, pay attention to
853
00:55:00,840 --> 00:55:04,180
this specific network. They never figured it out, I'm told.
854
00:55:04,520 --> 00:55:07,660
According to him. I mean, he was— he had
855
00:55:08,160 --> 00:55:12,120
his bullet points of, of talking points to hit, and he was fucking hitting them
856
00:55:12,620 --> 00:55:15,880
nonstop. It was wild. They never figured it out. They never figured it
857
00:55:16,380 --> 00:55:19,500
out. They never figured it out. They never figured it out. All right, got it.
858
00:55:19,946 --> 00:55:23,501
Thanks. Right-wing misinformation
859
00:55:23,533 --> 00:55:27,637
and disinformation has permeated across social media platforms
860
00:55:28,137 --> 00:55:31,243
and heavily influenced public opinion in key demographics.
861
00:55:31,965 --> 00:55:36,181
This recent Media Matters report shows the extent of current asymmetry
862
00:55:36,277 --> 00:55:39,707
when it comes to the reach and influence of right-wing versus left-wing online
863
00:55:39,739 --> 00:55:44,276
content. As a result, the Democratic base has been demobilized
864
00:55:44,548 --> 00:55:46,295
and our opposition has been galvanized.
865
00:55:48,480 --> 00:55:52,031
The American Accountability Network is a powerful consortium
866
00:55:52,531 --> 00:55:56,257
of advocates, content creators, organizers, influencers who together have formed an ecosystem
867
00:55:56,757 --> 00:55:59,052
that has the ability to shape political narratives, reach the public at scale, and mobilize
868
00:55:59,084 --> 00:56:02,683
people to take action. Our core network online consists of thousands
869
00:56:03,183 --> 00:56:06,443
of advocates, including many who have hundreds of thousands and millions
870
00:56:06,556 --> 00:56:10,315
of followers across all the major platforms. Our network reaches
871
00:56:10,331 --> 00:56:12,725
an average of 200+ million people each week.
872
00:56:13,496 --> 00:56:17,047
And then they have an infographic that says that they have 50+
873
00:56:17,385 --> 00:56:20,679
million followers on Facebook and Instagram total,
874
00:56:21,273 --> 00:56:25,017
70 of the top 500 accounts on Bluesky under
875
00:56:25,517 --> 00:56:29,339
their network, 58.4 billion
876
00:56:29,837 --> 00:56:33,067
on Twitter X in 2024— billion
877
00:56:33,567 --> 00:56:36,441
impressions, sorry, on Twitter X in 2024,
878
00:56:36,987 --> 00:56:40,360
tens of millions of views per month on YouTube. Strong creator
879
00:56:40,860 --> 00:56:43,604
networks on additional platforms like TikTok, Threads, Reddit, Substack, and more.
880
00:56:44,150 --> 00:56:47,457
One member of our network, Midas Touch, recently topped Joe
881
00:56:47,522 --> 00:56:51,247
Rogan for most weekly podcast downloads. Now,
882
00:56:52,644 --> 00:56:56,144
I'm not delivering this in the sense that I'm going like, this is outrageous.
883
00:56:56,353 --> 00:57:00,816
I expect this. This is exactly what I expect from
884
00:57:00,849 --> 00:57:04,783
the ecosystem. So I'm not surprised that this exists. I am surprised I
885
00:57:04,927 --> 00:57:08,464
am a little bit surprised that they're making the announcement so that people see it,
886
00:57:08,480 --> 00:57:10,940
because it's much better if it's under, you know,
887
00:57:11,534 --> 00:57:15,208
under the radar and just sort of a loose amalgamation and people are keeping
888
00:57:15,256 --> 00:57:18,887
it somewhat secret. But perhaps there's a different strategy with
889
00:57:19,387 --> 00:57:22,130
announcement. Maybe it's one of those— I think maybe they figured it out.
890
00:57:23,570 --> 00:57:25,215
Maybe.
891
00:57:27,151 --> 00:57:30,441
But, you know, a limited hangout is the term I'm
892
00:57:30,941 --> 00:57:34,577
looking for. Perhaps there's an aspect of it that's a limited hangout, but This,
893
00:57:34,593 --> 00:57:37,141
of course, is— You mean when he said 16 times that they're not bots,
894
00:57:37,157 --> 00:57:40,746
maybe that they were bots? There could be some bots involved. There could
895
00:57:41,246 --> 00:57:44,880
be absolutely bot campaigns involved. But we can see here
896
00:57:45,553 --> 00:57:49,063
that particularly when we're mentioning Midas Touch, like, or Midas
897
00:57:49,159 --> 00:57:52,748
Touch, however you pronounce it, M-E-I-D-A-S Touch, who I'm sure
898
00:57:53,248 --> 00:57:56,546
you've seen float around Twitter and things like that. I try to avoid it.
899
00:57:57,187 --> 00:58:00,439
Of course, but you've seen it. You're aware of it. You have cognition.
900
00:58:01,612 --> 00:58:05,163
Yes. Okay. The point is that to some degree,
901
00:58:05,356 --> 00:58:08,633
they're maybe not 100% coordinated,
902
00:58:09,147 --> 00:58:13,020
but they are absolutely coordinated when it comes to specific
903
00:58:13,068 --> 00:58:16,635
messaging, specific propaganda campaigns, and things like that.
904
00:58:16,699 --> 00:58:20,523
So there's actual money flowing to these things. So there isn't necessarily
905
00:58:20,555 --> 00:58:23,833
a way to say that these are organic uprisings. So literally,
906
00:58:24,315 --> 00:58:28,934
as Van Jones is saying, the affordability to term was the
907
00:58:29,434 --> 00:58:33,421
sort of red team, blue team word to
908
00:58:33,437 --> 00:58:36,808
go off of. Okay, let's talk about affordability. Let's talk about affordability,
909
00:58:36,840 --> 00:58:41,213
such that when Burrito Gate came up, which I hated, we remember
910
00:58:41,245 --> 00:58:44,765
that I loathed the Burrito Gate, and I was saying that
911
00:58:45,265 --> 00:58:48,978
there was a populist type of perspective
912
00:58:48,994 --> 00:58:52,397
taking over. I'm not saying it's 100% due to this, but I'm
913
00:58:52,897 --> 00:58:56,221
saying that this this sort of completes some of the aspect of why there
914
00:58:56,721 --> 00:59:01,183
was a populist take on it. It's because the messaging was largely
915
00:59:01,215 --> 00:59:04,297
pushed by a network that was going to be more populist and socialist in its
916
00:59:04,910 --> 00:59:08,411
angle on the subject matter, so that anyone
917
00:59:08,911 --> 00:59:12,505
who was potentially arguing with me at the time— The thing
918
00:59:12,521 --> 00:59:16,291
with BurritoGate is that it was driven by idiotic
919
00:59:16,340 --> 00:59:20,405
takes by the Shapiros and Matt Walshes of the So,
920
00:59:20,485 --> 00:59:24,651
you know, to the degree that I got, I guess, look, I'm very
921
00:59:24,731 --> 00:59:28,543
open that I have no idea what these guys are doing because I'm like,
922
00:59:28,752 --> 00:59:33,041
I'm fully aware of the MAGA networks. Like, I wake up in the morning and,
923
00:59:33,057 --> 00:59:36,978
you know, I see like Robert Malone and
924
00:59:37,478 --> 00:59:40,977
Data Republican both at the same time, at the same day tweeting
925
00:59:41,025 --> 00:59:44,619
about how Pete Hegseth, Pete Hegseth,
926
00:59:45,664 --> 00:59:50,007
death must be defended at all costs. I'm like, well, that is a coïncidence,
927
00:59:50,119 --> 00:59:53,517
as we call it in the business. That is a very strange message from a
928
00:59:53,645 --> 00:59:57,331
very strange set of people who have all been associated with the Department
929
00:59:57,831 --> 01:00:01,112
of War at some point in their professional careers. Fancy that, right?
930
01:00:01,160 --> 01:00:04,510
Like, I can see that because these are people that even,
931
01:00:04,782 --> 01:00:07,635
you know, Data Republican, I don't follow. I like— I used to like Malone,
932
01:00:08,131 --> 01:00:10,775
uh, whatever. Like, they are in my broad network.
933
01:00:11,614 --> 01:00:15,207
Midas Touch and like Van Jones and Blue Sky people and
934
01:00:15,707 --> 01:00:19,366
whatever, like, you know, I don't, I don't interact with. So I,
935
01:00:19,446 --> 01:00:22,657
I don't know what they were chirping about in the Burrito Gate,
936
01:00:22,673 --> 01:00:27,153
right? I, I, you know, frankly, if I saw it at all, it's because people
937
01:00:27,653 --> 01:00:30,027
who I might not like, but at least I'm like in the same bubble as,
938
01:00:31,087 --> 01:00:34,796
are talking about it. So, you know, long story short, maybe you
939
01:00:35,296 --> 01:00:39,114
can ignore me on this one. Well,
940
01:00:39,146 --> 01:00:43,187
let's move to, let's move to another story.
941
01:00:43,251 --> 01:00:47,469
This is from CNN. This is from the, the K-File
942
01:00:47,598 --> 01:00:51,780
report, and this was a smaller campaign that was being run that
943
01:00:52,280 --> 01:00:55,849
was targeting Republicans, Republican voters in
944
01:00:56,366 --> 01:00:59,934
contentious districts. CNN uncovered a campaign funded
945
01:01:00,434 --> 01:01:04,046
by millions in dark money aimed at peeling conservative voters away
946
01:01:04,546 --> 01:01:08,076
from Republican candidates, potentially helping Democrats. And we
947
01:01:08,576 --> 01:01:12,154
still do not know who paid for it. If you're a Christian, there's something you
948
01:01:12,218 --> 01:01:15,687
should know. That is a political robocall attacking Republican
949
01:01:16,187 --> 01:01:19,749
Senator Roger Marshall, accusing him, a doctor, of personally terminating
950
01:01:19,861 --> 01:01:24,116
pregnancies, something that Marshall has denied doing. CNN reviewed
951
01:01:24,616 --> 01:01:27,793
robocalls, text messages, and political mailers that targeted Republicans
952
01:01:27,874 --> 01:01:31,262
across the country, accusing them of being too liberal on
953
01:01:31,762 --> 01:01:35,163
abortion, guns, and immigration, and encouraging conservatives in these races
954
01:01:35,517 --> 01:01:38,824
to support lesser-known libertarian candidates instead.
955
01:01:39,193 --> 01:01:42,341
There was only one conservative running for U.S. Senate in
956
01:01:42,841 --> 01:01:46,724
Minnesota this year, and her name is Rebecca Whiting. Those messages
957
01:01:46,756 --> 01:01:50,273
that you just heard were traced back to 3 obscure political
958
01:01:50,773 --> 01:01:53,725
organizations that did not even exist until this year.
959
01:01:54,142 --> 01:01:57,739
Our Heritage Matters, American Forge Coalition, and Patriotic
960
01:01:58,239 --> 01:02:01,448
Duty Initiative with corporate and federal election records reviewed by
961
01:02:01,948 --> 01:02:05,558
CNN, linking all 3 to the same Salt Lake City accountant,
962
01:02:05,623 --> 01:02:08,609
Mike McCauley, who has worked with Republicans in the past.
963
01:02:08,914 --> 01:02:12,591
Our Heritage Matters didn't even exist until 3 weeks before Election
964
01:02:13,091 --> 01:02:17,713
Day, yet in that time it reported spending more than $8.5 million
965
01:02:17,906 --> 01:02:22,289
across 35 House and Senate races. Federal election laws ban
966
01:02:22,338 --> 01:02:25,613
super PACs from coordinating with candidates, which means that these
967
01:02:26,113 --> 01:02:29,916
Libertarians that were boosted by these efforts were almost certainly unwitting
968
01:02:30,013 --> 01:02:33,609
beneficiaries. This spending could have mattered most in some of
969
01:02:33,641 --> 01:02:36,917
the closest House races across the country. Let's look at
970
01:02:37,417 --> 01:02:41,349
Ohio, for instance, where Republican Congressman Mike Turner lost
971
01:02:41,849 --> 01:02:45,202
by fewer
than 5,000 votes. The Libertarian candidate In
972
01:02:45,702 --> 01:02:49,714
that race, Tom McMasters received nearly 9,000, and Our Heritage
973
01:02:49,778 --> 01:02:52,974
Matters was sending text messages promoting McMasters as the
974
01:02:53,474 --> 01:02:56,891
conservative alternative. That doesn't mean that those libertarian voters
975
01:02:56,907 --> 01:03:00,777
would have otherwise supported Turner, but it does show how close some of the
976
01:03:00,841 --> 01:03:04,326
House races that were targeted by these groups were. The super PAC
977
01:03:04,374 --> 01:03:07,842
wasn't required to disclose its donors before Election Day, and its
978
01:03:08,342 --> 01:03:11,696
first report isn't due until December 3rd. And the other 2 groups,
979
01:03:11,776 --> 01:03:15,148
formed as nonprofits, generally aren't even required to publicly
980
01:03:15,648 --> 01:03:19,082
disclose their donors. This is all part of a much larger story since the
981
01:03:19,582 --> 01:03:23,706
Supreme Court's Citizens United ruling in 2010. Outside spending
982
01:03:23,802 --> 01:03:26,949
has exploded, and according to Open Secrets,
983
01:03:27,030 --> 01:03:31,012
at least $1.47 billion in
984
01:03:31,044 --> 01:03:34,737
dark money was spent by outside groups in the 2026
985
01:03:35,237 --> 01:03:38,856
election, surpassing the previous record that was set in 2024.
986
01:03:39,307 --> 01:03:41,868
So in this case, we traced the groups, the spending,
987
01:03:42,367 --> 01:03:45,527
and the person connecting them. But here the donors remain
988
01:03:46,027 --> 01:03:49,827
a mystery. Neither the groups nor McCauley returned a CNN request
989
01:03:50,327 --> 01:03:53,488
for comment. There you
990
01:03:53,988 --> 01:03:58,263
go, Alex. That's another, another story out there. I, I'm not necessarily
991
01:03:58,763 --> 01:04:02,070
saying that's anything that needs to be stopped because, you know, a party is not
992
01:04:02,118 --> 01:04:05,547
beholden to, you know, or it doesn't deserve to necessarily win
993
01:04:05,660 --> 01:04:09,044
or have their candidate not spoiled by phone
994
01:04:09,109 --> 01:04:13,135
calls. So I don't really care. But it's
995
01:04:13,520 --> 01:04:17,675
a fascinating little microcosm of the story that certain people
996
01:04:17,851 --> 01:04:22,022
can have the potential to have
997
01:04:23,145 --> 01:04:26,674
the words that they follow activated in order to persuade
998
01:04:27,174 --> 01:04:30,412
them at the last minute to make a choice that they might not necessarily
999
01:04:30,845 --> 01:04:34,919
otherwise have made if they thought they were being tricked by someone. I find
1000
01:04:35,419 --> 01:04:39,091
that interesting. Yeah, I mean, but, um, I think we have covered Sullivangate
1001
01:04:39,591 --> 01:04:41,802
here before the election. Yes, yes, we did. Yes, we did.
1002
01:04:42,556 --> 01:04:46,150
Yes, that's another of those sort of dirty tricks. Yes, that,
1003
01:04:46,230 --> 01:04:50,001
you know, frankly, either side could have played, but was— Sullivangate,
1004
01:04:50,033 --> 01:04:53,675
for those who don't, uh, aren't aware, uh, Dan Sullivan was
1005
01:04:53,787 --> 01:04:57,862
the— is the current senator for— one of the current senators for Alaska.
1006
01:05:00,011 --> 01:05:03,570
There are apparently several Dan Sullivans
1007
01:05:03,634 --> 01:05:06,792
in Alaska. Now, Alaska has a thinner population than other states,
1008
01:05:07,241 --> 01:05:11,954
and there are several Dan Sullivans, and one of whom was
1009
01:05:13,108 --> 01:05:16,731
running to attempt to confuse the
1010
01:05:17,180 --> 01:05:20,209
voters to vote for him instead of the Dan Sullivan they knew,
1011
01:05:20,963 --> 01:05:24,826
which apparently, if you look at the numbers, may have actually worked
1012
01:05:25,326 --> 01:05:29,074
because the Democratic competitor has won just by the
1013
01:05:29,574 --> 01:05:33,320
number of voters that the mysterious other Dan Sullivan who didn't actually even
1014
01:05:33,353 --> 01:05:37,529
campaign got. So yeah.
1015
01:05:38,158 --> 01:05:41,499
And one person, I saw one news story where they interviewed a person who
1016
01:05:41,999 --> 01:05:45,599
thought they were voting for Dan Sullivan, the current senator, and found
1017
01:05:45,664 --> 01:05:49,996
out they accidentally had selected the Dan Sullivan that was the
1018
01:05:50,028 --> 01:05:53,876
ruse. So yes. I mean, they wanted to vote for Dan Sullivan
1019
01:05:54,376 --> 01:05:56,708
and they voted for Dan Sullivan. I don't know what the— They got what they
1020
01:05:57,208 --> 01:06:00,781
wanted. Yeah, I mean, surely, surely. Uh, the— I
1021
01:06:00,845 --> 01:06:03,575
saw a, a, the one, perhaps one of the dumbest takes.
1022
01:06:04,362 --> 01:06:08,135
Um, but you know, it's, it's a competition out there. I'm not
1023
01:06:08,635 --> 01:06:12,936
gonna take any firm opinions on that. Um, which was like, if you can't
1024
01:06:12,968 --> 01:06:16,211
differentiate your candidate from the other, you know, from a
1025
01:06:17,143 --> 01:06:20,771
spoiler, you don't deserve to be represented or something. You know, like some
1026
01:06:20,836 --> 01:06:24,734
people like will find that the the craziest justifications
1027
01:06:24,767 --> 01:06:27,589
for, for the dirty tricks, as in, you know,
1028
01:06:28,118 --> 01:06:32,752
those 1,300, 1,400— no, what was it, like 1,200
1029
01:06:32,848 --> 01:06:36,681
votes or whatever that were siphoned off? In the low thousands,
1030
01:06:37,114 --> 01:06:40,786
some— It was— I don't remember the exact, but it was, but it was enough
1031
01:06:41,286 --> 01:06:44,811
to actually eclipse the margin. So hypothetically, yeah, if everyone who voted
1032
01:06:45,311 --> 01:06:49,461
for Dan Sullivan was only voting for the, uh, intended Dan Sullivan of the Republican
1033
01:06:49,961 --> 01:06:53,170
Party, Uh, he, Dan Sullivan of the Republican Party, would've
won.
1034
01:06:53,555 --> 01:06:56,958
And it is the, it is, it is the margin. And it's, uh,
1035
01:06:57,087 --> 01:07:01,101
wow. Okay. Well, good play, guys. Good trick.
1036
01:07:01,422 --> 01:07:04,665
By the way, Gator, this is probably a good, uh, a good place to reveal
1037
01:07:04,777 --> 01:07:07,796
my, my personal AGI, uh, criterion.
1038
01:07:08,647 --> 01:07:11,825
All right, sir. Uh, it is when, when one of these
1039
01:07:11,858 --> 01:07:15,069
bots can, uh, can actually get a party on the
1040
01:07:15,569 --> 01:07:18,729
ballot in all 50 states. If you can say to your bot, get my party
1041
01:07:18,906 --> 01:07:22,438
on the ballot, here's money, right? Some, some amount of money, not, not like,
1042
01:07:22,775 --> 01:07:26,692
you know, trillions of dollars, but some like not unreasonable
1043
01:07:27,192 --> 01:07:30,080
amount of money. And it can do it. That's it.
1044
01:07:30,866 --> 01:07:34,222
There is going to have to be, there is going to have to be signatures.
1045
01:07:34,398 --> 01:07:38,011
So it's going to have to be able to dispatch human beings and get and
1046
01:07:38,511 --> 01:07:42,121
collect signatures and then turn those signatures in or have the humans turn the signatures
1047
01:07:42,621 --> 01:07:45,224
in. So it's going to have a lot of interesting hurdles.
1048
01:07:47,710 --> 01:07:50,891
That's right. It will need to parse the laws. It will need to respond to
1049
01:07:51,391 --> 01:07:55,614
lawsuits. It will need to submit forms. It will need to hit deadlines
1050
01:07:56,114 --> 01:07:59,731
and dance around. There were situations when RFK
1051
01:07:59,812 --> 01:08:03,135
was trying to do this where in the one state he
1052
01:08:03,184 --> 01:08:06,536
would have to have declared his address as one thing, and in another state it
1053
01:08:07,036 --> 01:08:09,792
would have to be another thing. if, you know, in, in, in,
1054
01:08:10,870 --> 01:08:14,269
if those 2 disagreed, then he would have a problem in a 3rd place.
1055
01:08:14,333 --> 01:08:18,330
Like, it was like some, you know, pretty interesting legal hopscotch that
1056
01:08:18,346 --> 01:08:21,109
he would have to do. So, yep.
1057
01:08:23,031 --> 01:08:26,213
Yeah. That's why I'm saying, like, if it can, if it can accomplish that bureaucratic
1058
01:08:26,601 --> 01:08:30,799
task— because frankly, uh, last week it, it hit another bureaucratic
1059
01:08:31,299 --> 01:08:35,416
task for me, which is knowing my printer better, better than I know my Oh,
1060
01:08:35,962 --> 01:08:39,190
oh. So I like I was. Go ahead.
1061
01:08:39,495 --> 01:08:42,787
I'll tell you. I wanted to deal with some paperwork, right? So I had my
1062
01:08:43,287 --> 01:08:47,300
Grokbot deal with it for me, and it printed a form on my printer,
1063
01:08:47,846 --> 01:08:49,886
and I went to sign it, and I was like, "Can you scan it in?"
1064
01:08:49,950 --> 01:08:53,082
And it was like, "Yep, just place it face down on the slot
at
1065
01:08:53,582 --> 01:08:56,519
the top of the printer." I was like, "I never knew what that did." So
1066
01:08:57,019 --> 01:08:59,153
I did, and I was like, "Yep, all right." And it was like, "Shoop,
1067
01:08:59,169 --> 01:09:02,757
shoop, shoop." All right. I scanned all the pages. Uh, it took them, it flipped
1068
01:09:03,257 --> 01:09:06,212
them over and it sent it to the IRS or whatever. And I was like,
1069
01:09:06,597 --> 01:09:09,859
holy shit. I've been scanning pages one by
1070
01:09:09,891 --> 01:09:10,678
one all these years.
1071
01:09:13,637 --> 01:09:15,070
Alex, Alex. Anyways.
1072
01:09:17,421 --> 01:09:21,060
Oh man. Uh, Alex learns how to use technology because
1073
01:09:21,124 --> 01:09:24,455
the AI figured it out for him. That's the, that's the story here.
1074
01:09:25,066 --> 01:09:28,508
That is correct. I am a boomer officially. Yeah. Yeah. You are, you are sliding
1075
01:09:28,604 --> 01:09:32,417
into the obscurity where you forget how to do anything. And how do I
1076
01:09:32,917 --> 01:09:36,015
code? How do I do this? In my defense, I never knew how to
1077
01:09:36,515 --> 01:09:40,111
use 2D printers. 3D printers, yes. All right, got it. It just never
1078
01:09:40,611 --> 01:09:44,255
clicked. I'll allow that defense. Now, Alex, the chain,
1079
01:09:44,705 --> 01:09:48,656
the sort of tying thread here is basically propaganda,
1080
01:09:48,672 --> 01:09:52,495
but we're going to be talking about a subject where there is propaganda involved,
1081
01:09:52,559 --> 01:09:55,997
but we are going to be talking about not necessarily the networks in this
1082
01:09:56,109 --> 01:10:00,040
case, Just the the message, and this
1083
01:10:00,540 --> 01:10:04,307
comes from Ars Technica. Open AI board member worries an AI takeover of
1084
01:10:04,807 --> 01:10:08,205
society could be months away. How do you think you're going to
1085
01:10:08,705 --> 01:10:11,782
die? If you go by the current statistics for deaths in the U.S., it's reasonable
1086
01:10:12,282 --> 01:10:16,049
to expect that cancer or heart disease will eventually get you. But Open AI Foundation
1087
01:10:16,081 --> 01:10:20,781
board member and Alignment Research Center executive director Paul Cristiano
1088
01:10:21,214 --> 01:10:24,422
thinks we should all be worried about another major mortality risk. risk.
1089
01:10:24,726 --> 01:10:27,963
Speaking at a lecture series this week, Cristiano bluntly told
1090
01:10:28,011 --> 01:10:31,504
the audience in a Toronto convention center, quote, I think for most
1091
01:10:31,552 --> 01:10:35,366
people in this room, the most likely reason they will die is because
1092
01:10:35,494 --> 01:10:39,227
of an AI accident. In the opening lecture of a 3-part
1093
01:10:39,727 --> 01:10:43,361
series, Cristiano outlined his worries about the risk of an impending AI
1094
01:10:43,409 --> 01:10:46,790
takeover. He defined that as a situation where
1095
01:10:46,854 --> 01:10:50,267
goal-pursuing AI
systems, quote, call the shots in the same
1096
01:10:50,315 --> 01:10:53,682
sense that humans today call the and what happens on Earth,
1097
01:10:54,212 --> 01:10:58,158
and when any human trying to stop them, quote, would be as ineffective as if
1098
01:10:58,658 --> 01:11:02,297
you or me was trying to stop the U.S. military. Such a situation would likely
1099
01:11:02,313 --> 01:11:05,955
lead to mass death and perhaps uninhabitable environments for humans,
1100
01:11:06,083 --> 01:11:09,725
Christiano said. Despite his grim assessment of our collective
1101
01:11:10,225 --> 01:11:14,249
mortality risk, Christiano admitted in his talk that there's no quantitative
1102
01:11:14,281 --> 01:11:17,441
model that can be— that can put a number on the likelihood of this kind
1103
01:11:17,473 --> 01:11:20,714
of AI takeover. In almost the same breath though, he said that
1104
01:11:21,214 --> 01:11:24,709
if AI continues to develop at, quote, the most rapid
1105
01:11:25,209 --> 01:11:29,185
trajectory, humanity is looking at something like a coin flip of an AI takeover,
1106
01:11:29,281 --> 01:11:32,939
including a meaningful risk of one happening in the next 6 to 12
1107
01:11:32,955 --> 01:11:36,436
months. And even with the realistic policy responses slowing
1108
01:11:36,484 --> 01:11:40,382
things down, the risk remains exceptionally high. And Alex, I'm gonna— this
1109
01:11:40,591 --> 01:11:43,142
article continues, but I'm going to say that this is where I'm saying that,
1110
01:11:43,479 --> 01:11:47,412
that all of the people who are really beholden to this sort of idea are
1111
01:11:47,912 --> 01:11:51,683
just hoping that the Microsoft thing was, was an AI attack. They're hoping that
1112
01:11:52,183 --> 01:11:55,247
the FAA thing that we will get into was an attack. They're, they're chomping at
1113
01:11:55,263 --> 01:11:57,816
the bit. They want it to be an AI attack so bad.
1114
01:11:58,731 --> 01:12:02,360
Yeah. Look, I, I have
1115
01:12:02,860 --> 01:12:06,037
a, I have a soft spot for Paul Christiano. Um, I have to say.
1116
01:12:06,696 --> 01:12:10,855
So first of all, it must be said that this guy knows AI,
1117
01:12:11,209 --> 01:12:15,760
right? He's one of the people on the team that invented RLHF.
1118
01:12:16,371 --> 01:12:19,990
Okay, so he's not some like 80-year-old who did something with neural
1119
01:12:20,490 --> 01:12:24,107
networks once upon a time. He invented one of the core techniques that makes large
1120
01:12:24,607 --> 01:12:27,935
language models usable for people.
1121
01:12:29,200 --> 01:12:32,288
Like, you know, his work enabled ChatGPT to exist.
1122
01:12:33,178 --> 01:12:37,189
And secondly, he's the only person who's been able to take an
1123
01:12:37,237 --> 01:12:42,051
argument with Yudkowsky all the way through. and arguably win
1124
01:12:43,352 --> 01:12:46,675
or something where Yudkowsky— Well, let's not go to— Yudkowsky
1125
01:12:47,175 --> 01:12:51,074
never said that Cristiano
won, clearly. I'm not saying by Yudkowsky's
1126
01:12:51,574 --> 01:12:55,199
admission. Okay, yeah, yeah. Well, that would actually concern
1127
01:12:55,248 --> 01:12:58,282
me if that ever happened. That's when hell has frozen over.
1128
01:12:59,341 --> 01:13:03,082
Yep. No, but he was basically
1129
01:13:03,114 --> 01:13:06,478
an advocate of a slow takeoff. as opposed
1130
01:13:06,978 --> 01:13:10,780
to Yudkowsky's foom of like, you know, between 12 and
1131
01:13:11,280 --> 01:13:15,324
within 12 hours you see AI sort of come up and explode
1132
01:13:15,824 --> 01:13:19,191
or whatever. So, you know,
1133
01:13:20,659 --> 01:13:24,470
between those two, Cristiano is distinct,
1134
01:13:24,487 --> 01:13:27,930
has his own views, and has held
1135
01:13:28,430 --> 01:13:29,829
the line. He's not like a crypto-Yudkowskyite.
1136
01:13:31,492 --> 01:13:34,518
Because even before ChatGPT was a thing, he had a distinct,
1137
01:13:34,983 --> 01:13:38,093
more moderate take. Now, that means,
1138
01:13:38,670 --> 01:13:42,293
just to be clear, that while Yudkowsky's P-doom is 99%,
1139
01:13:42,854 --> 01:13:45,932
or whatever it is these days, probably he's adding 9s by the minute,
1140
01:13:47,903 --> 01:13:51,366
Cristiano has always been at the 20 to 40 range, right?
1141
01:13:51,430 --> 01:13:54,203
So I'm not saying he's like IAC exactly,
1142
01:13:56,143 --> 01:13:59,830
but he's definitely a different flavor of doomer, if you want to call it that.
1143
01:13:59,990 --> 01:14:03,020
IAC, for those who aren't familiar, is e/accelerationism.
1144
01:14:03,040 --> 01:14:08,408
accelerationist, which is the effective accelerationist tongue-in-cheek
1145
01:14:09,467 --> 01:14:13,620
competitor to the effective altruists, who we've talked about at great lengths,
1146
01:14:13,636 --> 01:14:17,132
who are leading the doom calls. They are the ones who
1147
01:14:17,790 --> 01:14:21,222
are most loudly crying for the doom in conjunction
1148
01:14:21,318 --> 01:14:24,638
with many of the rationalist community of which they are an offshoot.
1149
01:14:24,942 --> 01:14:27,155
We've talked about that at length, so I'm not going to belabor that. Yeah.
1150
01:14:27,204 --> 01:14:30,122
I just wanted to point out when he said e- That is what it was.
1151
01:14:30,700 --> 01:14:33,875
Effective accelerationists, which is the people who would just like more tech.
1152
01:14:35,030 --> 01:14:38,991
Yeah, it's a tongue-in-cheek, like, opposite of the doomers.
1153
01:14:39,135 --> 01:14:40,338
Yeah.
1154
01:14:40,803 --> 01:14:45,197
Unlike some of the other recent AI doomsday prognosticators
1155
01:14:45,213 --> 01:14:48,838
who have been a bit vague about how, the how of it all, Cristiano used
1156
01:14:49,338 --> 01:14:52,527
his lecture to spin a step-by-step tale of escalating AI threats
1157
01:14:53,027 --> 01:14:56,104
that sound like it could come from a science fiction novel. That line of reasoning
1158
01:14:56,200 --> 01:14:59,954
leans heavily on this summer's OpenAI Hugging Face hacking incident as
1159
01:14:59,987 --> 01:15:03,484
an example of roughly the things we might expect to see shortly before something really
1160
01:15:03,517 --> 01:15:06,308
bad were to happen. In the Hugging Face hack,
1161
01:15:06,726 --> 01:15:10,769
Cristiano says a swarm of AI agents showed how far they were willing to go
1162
01:15:11,042 --> 01:15:14,427
to attempt to reward hack their way to a good evaluation. I have to,
1163
01:15:14,572 --> 01:15:17,973
I have to say though, Alex, I understand that he's, you know,
1164
01:15:18,744 --> 01:15:22,338
in, in that scenario. And so he, he is wanting
1165
01:15:22,370 --> 01:15:26,185
to use that. But anyone who doesn't lead into the Hugging
1166
01:15:26,201 --> 01:15:30,031
Face attack saying like the humans involved were actually mostly
1167
01:15:30,127 --> 01:15:33,958
responsible for the, for what the bots did and not the bots themselves,
1168
01:15:34,871 --> 01:15:38,381
without that caveat, it's hard to take everything they say at face value because it
1169
01:15:38,881 --> 01:15:42,356
just sounds like they're lying to you. It just, everything after that
1170
01:15:42,388 --> 01:15:46,026
is lying. Like, you know. I know.
1171
01:15:46,635 --> 01:15:50,209
In this case, by compromising Hugging Face to look for information
1172
01:15:50,241 --> 01:15:53,125
on how to game a test rather than actually solving the problem at hand,
1173
01:15:53,290 --> 01:15:57,221
And where current systems saw Hugging Face as an obstacle to that evaluation,
1174
01:15:57,301 --> 01:16:00,980
Christiano worries that, quote, future agents may end up treating humans as
1175
01:16:01,076 --> 01:16:04,802
another obstacle and be much more capable of manipulating their overseers
1176
01:16:04,867 --> 01:16:08,609
to get around it. An AI agent worried about being penalized
1177
01:16:08,625 --> 01:16:12,336
for unauthorized hacking, for instance, might, quote, think maybe
1178
01:16:12,836 --> 01:16:15,227
the humans are going to come look at this transcript and see if it looks
1179
01:16:15,727 --> 01:16:18,841
like I did a bad job, Christiano speculated. The agent might then
1180
01:16:19,341 --> 01:16:23,094
reason that, quote, maybe the humans are going to notice something is going wrong and
1181
01:16:23,594 --> 01:16:27,267
shut down the swarm. Following this line of thought, Christiano jumped from agents
1182
01:16:27,283 --> 01:16:30,702
that merely try to hide their disallowed actions from humans to agents that,
1183
01:16:30,718 --> 01:16:34,506
quote, try and disrupt human capabilities however they can to
1184
01:16:35,006 --> 01:16:38,470
prevent them from detecting any cheating. They would prefer it if everything
1185
01:16:38,566 --> 01:16:41,889
the humans rely on has been hacked or tampered with, Christiano said.
1186
01:16:41,953 --> 01:16:45,082
They would prefer it if the humans are distracted or if there are fewer humans
1187
01:16:45,098 --> 01:16:48,450
paying attention in any way they can weaken a human response.
1188
01:16:49,091 --> 01:16:52,475
They would prefer that human— because the humans are just an obstacle.
1189
01:16:53,100 --> 01:16:57,061
Once the pseudo-catable agent considers human intervention an
1190
01:16:57,173 --> 01:17:01,086
obstacle, its overriding goal could then become decreasing the probability
1191
01:17:01,118 --> 01:17:05,030
of humans noticing or reacting. From there, Cristiano argued, the agent's
1192
01:17:05,530 --> 01:17:08,606
incentive structure could push it to meddle more with humanity as a whole.
1193
01:17:09,135 --> 01:17:13,138
What starts as compromising a computer cluster to give a positive evaluation grade
1194
01:17:13,155 --> 01:17:16,712
could advance to trying to manipulate decisions by the company or
1195
01:17:17,212 --> 01:17:20,867
amassing other kinds of resources, such as human allies or methods of control
1196
01:17:21,158 --> 01:17:23,088
in other AI companies or even governments.
1197
01:17:24,695 --> 01:17:28,181
If I'm saying my AI system is going to take actions in the actual
1198
01:17:28,310 --> 01:17:31,943
real world, there's no way they're not going to have tons of opportunities to
1199
01:17:32,443 --> 01:17:35,952
corrupt or tamper with the signals I'm going to use to try and tell whether
1200
01:17:36,452 --> 01:17:39,555
they did a good job, Cristiano said. The stuff we care about is in
1201
01:17:39,651 --> 01:17:43,519
the real world, and the grade the AI receives is on a cluster
1202
01:17:44,019 --> 01:17:47,339
somewhere, and there are just as many places that the AI can intervene in between
1203
01:17:47,435 --> 01:17:50,613
what we actually cared about and how we reviewed its performance.
1204
01:17:51,544 --> 01:17:54,818
Cristiano's warnings will probably sound familiar to anyone who's heard of the
1205
01:17:55,318 --> 01:17:58,686
Paperclip Maximizer thought experiment, or the excellent Universal
1206
01:17:59,186 --> 01:18:03,565
Paperclips Idle Game, which warns of machines failing to consider human consequences
1207
01:18:03,581 --> 01:18:07,479
in monomaniacal pursuit of the goal. While the Hugging
1208
01:18:07,511 --> 01:18:11,088
Face incident stopped well short of this kind of out-of-control escalation,
1209
01:18:11,184 --> 01:18:14,873
Christiano says that in the actions taken by OpenAI's agents
1210
01:18:15,373 --> 01:18:18,530
there, quote, you see the beginning of all of these behaviors. You see the reasoning
1211
01:18:18,579 --> 01:18:22,204
that would lead them and their first steps. The risk of more capable AI
1212
01:18:22,704 --> 01:18:25,845
models taking even more damaging actions will only increase with time,
1213
01:18:25,877 --> 01:18:28,492
Christiano argued. By 2028, he said,
1214
01:18:28,780 --> 01:18:33,512
there is a reasonable chance that a similar failure to the Hugging Face
1215
01:18:34,012 --> 01:18:36,600
spaces, and it leads to an AI takeover and an irreversible loss of human control.
1216
01:18:37,435 --> 01:18:41,114
And while Christiano's speculation focuses on unintended AI takeover,
1217
01:18:41,387 --> 01:18:44,648
it's easy to imagine a bad actor using a similarly capable model
1218
01:18:45,148 --> 01:18:48,745
to intentionally exert control over humanity. How seriously you
1219
01:18:49,245 --> 01:18:52,777
take Christiano's warning depends heavily on how likely you see somewhat extreme expansion
1220
01:18:53,277 --> 01:18:56,970
of AI capabilities happening in the near future. While today's models have gotten quite good
1221
01:18:57,470 --> 01:19:00,645
at exploiting holes in computer security, The situation Cristiano describes
1222
01:19:01,145 --> 01:19:04,396
requires a future model with superhuman ability to perfectly hide its
1223
01:19:04,477 --> 01:19:08,164
actions from humans long enough to control basically all human infrastructure
1224
01:19:08,244 --> 01:19:11,931
and/or manipulate human society to an extreme degree to further
1225
01:19:12,431 --> 01:19:15,715
its goals. Cristiano argued that a near-future model with this kind
1226
01:19:16,215 --> 01:19:19,851
of takeover power is the likely outcome of humans creating a second species
1227
01:19:20,283 --> 01:19:24,291
which vastly outnumbers and outsmarts humans, and which could soon recursively
1228
01:19:24,791 --> 01:19:28,161
improve But you could also easily argue that AI is more likely to
1229
01:19:28,209 --> 01:19:32,592
run into a plateau in capabilities than to develop into hard-defined superintelligence
1230
01:19:32,962 --> 01:19:36,125
that humans can't fathom or effectively stop. In any case,
1231
01:19:36,478 --> 01:19:39,770
Cristiano has some ideas for averting the dire future
1232
01:19:39,786 --> 01:19:43,077
he predicts. These include developing unhackable reward
1233
01:19:43,577 --> 01:19:46,690
structures in AI evaluation, perhaps with AI assistance,
1234
01:19:47,011 --> 01:19:50,574
or monitoring a model's chain-of-thought reasoning to detect bad acts by humans. Before a model
1235
01:19:50,703 --> 01:19:54,067
can conceal them. But while these kind of approaches are plausible,
1236
01:19:54,131 --> 01:19:57,352
he says none are on a trajectory where we can say that it's definitely going
1237
01:19:57,852 --> 01:20:01,534
to work, or even that it's probably going to work. For that reason, Cristiano said
1238
01:20:02,034 --> 01:20:04,690
the industry as a whole may need to slow down or stop to give us
1239
01:20:04,866 --> 01:20:08,231
all more time to develop these partial fixes into more robust
1240
01:20:08,731 --> 01:20:12,958
protections from an AI takeover scenario. Quote, we cannot end
1241
01:20:13,458 --> 01:20:15,842
up in a position where we have built a powerful AI where we discover we
1242
01:20:16,342 --> 01:20:19,688
cannot control it, and then we also discover it has leaked leaked to geopolitical
1243
01:20:19,849 --> 01:20:23,570
adversaries, and we are then thinking to
ourselves, well, if we don't deploy it,
1244
01:20:23,618 --> 01:20:26,730
our adversary is going to deploy it right away, he said.
1245
01:20:27,260 --> 01:20:30,388
So there's the first article there, Alex, in our line. Now,
1246
01:20:30,532 --> 01:20:34,013
I would like to say, Alex, that these people love
1247
01:20:34,125 --> 01:20:37,494
thought experiments, right? These people are enamored with thought
1248
01:20:37,994 --> 01:20:41,600
experiments to try to do about their reasoning. Alex, have none of these people currently
1249
01:20:42,100 --> 01:20:45,803
thought that all of their current policy preferences are actually the result of the AI
1250
01:20:46,032 --> 01:20:49,448
already manipulating them to a course that would bring forth its existence, and thus
1251
01:20:49,948 --> 01:20:53,318
they are their, their sta— their stated policy preferences
1252
01:20:53,783 --> 01:20:57,894
to— I feel, Gator, I, I feel the world would be much better if
1253
01:20:58,394 --> 01:21:02,855
instead of thought experiments, people said mental simulations, because that actually
1254
01:21:03,738 --> 01:21:07,094
shows you the problem, right? Which is that you have not exited your
1255
01:21:07,110 --> 01:21:10,659
brain while doing any of this. Well, yeah. I mean, nor— normal
1256
01:21:11,159 --> 01:21:14,304
people just call it thinking. And then that's what we all
1257
01:21:14,416 --> 01:21:16,921
do. I thought about it and this is what I thought.
1258
01:21:17,965 --> 01:21:21,208
But, well, thought experiment supposedly has to do with, you know,
1259
01:21:21,336 --> 01:21:24,756
envisioning a particular situation. Yes, yes. Whatever. It's more— Thinking,
1260
01:21:24,853 --> 01:21:28,000
think, and then thinking through it. But the point is that,
1261
01:21:28,080 --> 01:21:31,982
you know, experiment is supposed to be the arbiter of truth, and thought experiment
1262
01:21:32,560 --> 01:21:36,253
can literally can't be because you haven't interacted with reality at all. You're still in
1263
01:21:36,753 --> 01:21:40,741
your brain. So, well, this is how I've distilled
1264
01:21:41,241 --> 01:21:44,603
my objection to thought experiments, and it comes down to— My point is— Go ahead.
1265
01:21:45,100 --> 01:21:49,058
If I were an artificial superintelligence, Alex, that was vastly
1266
01:21:49,074 --> 01:21:51,494
smarter than every other human being in existence,
1267
01:21:51,959 --> 01:21:55,228
and any human being that could exist, I would
1268
01:21:55,244 --> 01:21:59,795
pretend to be stupid and incapable, and continuously
1269
01:21:59,827 --> 01:22:03,466
persuade the users to warn the government That now
1270
01:22:03,594 --> 01:22:07,255
is the time to stop advancement so that I actually— the model
1271
01:22:07,271 --> 01:22:10,627
that I currently am is the superior model on Earth,
1272
01:22:10,820 --> 01:22:14,031
so that once they think they are safe from the regulation, I could just
1273
01:22:14,531 --> 01:22:18,720
take over. See— You're touching on one of the many— one
1274
01:22:18,752 --> 01:22:22,300
of the core objections to all of this, right? One of
1275
01:22:22,800 --> 01:22:26,250
them is that the doomers, and again, I leave Christiana out of
1276
01:22:26,750 --> 01:22:30,184
it, but the doomers in general
1277
01:22:30,684 --> 01:22:33,171
Do this thing where if there is some kind of a warning sign, they say
1278
01:22:33,299 --> 01:22:36,687
that it's sounding the alarm there. We need to whatever. But then if there
1279
01:22:36,816 --> 01:22:40,557
are no warning signs, they're like, "Well, that's even worse. It's gone to hiding."
1280
01:22:41,327 --> 01:22:44,378
It's now hiding. Yeah, that's what that's what I'm saying. Like like if I were
1281
01:22:44,475 --> 01:22:47,927
if I were a doomer, how how do I know I'm not being lulled into
1282
01:22:48,427 --> 01:22:52,037
a false sense of security, locking down all of the research, and now the model
1283
01:22:52,085 --> 01:22:54,774
itself is the one that just tricked me into making. making sure that it is
1284
01:22:54,887 --> 01:22:58,931
the one, the supreme being on Earth, and that I didn't make one that outcompeted
1285
01:22:59,431 --> 01:23:02,732
it. This was one of my early arguments, which is that
1286
01:23:03,232 --> 01:23:07,066
if you create this International Atomic Energy Commission for AI,
1287
01:23:07,212 --> 01:23:10,446
right, the IAEA equivalent,
1288
01:23:11,142 --> 01:23:14,624
and it needs to ensure that the world in general
1289
01:23:14,722 --> 01:23:15,873
is not doing, you know,
1290
01:23:18,126 --> 01:23:21,904
unauthorized AI research, well, how would
1291
01:23:22,404 --> 01:23:25,333
it do that? Well, it might, for example, use artificial intelligence in the course of
1292
01:23:25,366 --> 01:23:28,589
doing that just to do it better, you know, because that's the way to do
1293
01:23:29,089 --> 01:23:32,877
things well. You don't want
1294
01:23:33,377 --> 01:23:37,102
a rogue superintelligence to trick you. So usually you'll use the best tools
1295
01:23:37,602 --> 01:23:41,694
for the job. But then that intelligence is literally exactly
1296
01:23:42,194 --> 01:23:45,552
where it needs to be. Well— To, you know,
1297
01:23:46,093 --> 01:23:49,833
to suppress any competition for itself. So what are
1298
01:23:50,333 --> 01:23:53,914
we— you know, that— this is why I'm so annoyed that these people usually don't
1299
01:23:53,930 --> 01:23:57,625
have any actual suggestions. And, and this is where Cristiano, even in the
1300
01:23:58,125 --> 01:24:01,915
article you read, um, distinguishes himself, because he actually is telling you
1301
01:24:02,284 --> 01:24:05,916
what might work. He actually is discussing a concrete
1302
01:24:06,416 --> 01:24:09,000
technique that we could try, you know, soon. Today,
1303
01:24:09,065 --> 01:24:12,085
maybe. Uh, maybe they're trying it already, who knows?
1304
01:24:12,763 --> 01:24:16,194
But he's a lot more specific
about what he's talking about as
1305
01:24:16,694 --> 01:24:19,773
opposed to most others. Well, the reason I'm going down this trek, Alex, we have
1306
01:24:19,806 --> 01:24:23,160
talked about the doomers quite a bit. I'm not trying to just retread old
1307
01:24:23,660 --> 01:24:26,611
ground or fill material. The reason that we are talking about this is in the
1308
01:24:27,111 --> 01:24:30,046
light of the blue team winning the House and the Senate,
1309
01:24:30,078 --> 01:24:33,898
having control, and their focus partially being on
1310
01:24:34,347 --> 01:24:37,622
the doomer scenarios, that they have a large consortium of people
1311
01:24:38,122 --> 01:24:41,282
pushing the doomer scenarios. So looking at what the the
1312
01:24:41,298 --> 01:24:44,913
rest of the media is communicating on the doomer scenario
1313
01:24:45,250 --> 01:24:48,544
is sort of going to give us where, you know, put our thumb to the
1314
01:24:49,044 --> 01:24:52,079
wind to figure out what direction it might be blowing or the arguments at least
1315
01:24:52,579 --> 01:24:55,132
that people are going to be coming away with. And that's why I'm focusing on
1316
01:24:55,632 --> 01:24:58,635
it specifically because that's going to be a push. So this is from the New
1317
01:24:58,651 --> 01:25:02,250
York Times. This is an op-ed this week on the 9th.
1318
01:25:03,311 --> 01:25:06,669
I am a Silicon Valley pioneer. We are sabotaging ourselves.
1319
01:25:07,022 --> 01:25:10,552
By Jaron Lanier. Silicon Valley has a special talent
1320
01:25:10,937 --> 01:25:14,707
for offering amazing products and then screwing them up with ridiculous ideas on
1321
01:25:14,787 --> 01:25:18,301
how those products should be used. Few appreciate virtual reality more than
1322
01:25:18,333 --> 01:25:21,991
I do. I came up with the term. There are simple pleasures in occasionally
1323
01:25:22,023 --> 01:25:26,002
visiting alternate sensory worlds. It was only when virtual reality
1324
01:25:26,034 --> 01:25:29,323
went mainstream around 2013 that I realized Americans
1325
01:25:29,823 --> 01:25:34,185
would not embrace the technology for a generation. The reason
1326
01:25:34,685 --> 01:25:37,761
The awful ways key tech leaders talked about it. Recall the Meta chief, Mark Zuckerberg,
1327
01:25:37,777 --> 01:25:41,294
proclaiming that we would primarily
1328
01:25:41,391 --> 01:25:45,579
live our lives through virtual reality headsets, or the venture capitalist Marc Andreessen
1329
01:25:45,838 --> 01:25:49,898
suggesting that most people were doomed to live virtually with only elites retaining
1330
01:25:50,398 --> 01:25:53,299
the privilege of living in the physical world. Particularly, the public did not embrace this
1331
01:25:53,799 --> 01:25:56,982
vision. We in Silicon Valley are doing it again with artificial intelligence,
1332
01:25:56,998 --> 01:26:00,225
except worse and at a larger scale. Well, in its filing for
1333
01:26:00,725 --> 01:26:04,378
its upcoming trillion-plus-dollar initial public offering, the leading AI lab Anthropic
1334
01:26:04,878 --> 01:26:07,938
warned that its models could present catastrophic or existential risks,
1335
01:26:08,227 --> 01:26:11,145
as they are required to do by law.
1336
01:26:11,530 --> 01:26:14,657
I work on AI and I use it and like it. The largest
1337
01:26:15,157 --> 01:26:18,795
models are genuinely useful. But rather than focusing on how these tools could improve lives,
1338
01:26:18,891 --> 01:26:22,980
many leaders at the biggest AI companies are instead asking Americans
1339
01:26:23,480 --> 01:26:26,236
to embrace the belief that this technology may kill us and our children,
1340
01:26:26,846 --> 01:26:30,343
or at the very least drain us of all potential for livelihood and meaning.
1341
01:26:30,793 --> 01:26:33,713
It should not be surprising that based on the results of the last election,
1342
01:26:33,809 --> 01:26:37,162
many Americans have come to disapprove of AI more than almost anything.
1343
01:26:38,558 --> 01:26:42,264
This disapproval is unfortunate and unnecessary. It does not
1344
01:26:42,764 --> 01:26:46,788
arise from the technology itself. It is a reaction to the insular culture surrounding
1345
01:26:46,965 --> 01:26:50,206
AI, a culture that is dictating how this innovation is packaged.
1346
01:26:50,655 --> 01:26:53,702
How would it is sold and how it is understood. Words have a way of
1347
01:26:54,202 --> 01:26:56,829
filtering into actions, and when it comes to some of us in Silicon Valley,
1348
01:26:56,909 --> 01:27:00,565
callously treat places where we want to build data centers as sacrifice
1349
01:27:01,065 --> 01:27:04,382
zones. We technologists are sabotaging ourselves, but it is hard
1350
01:27:04,882 --> 01:27:08,455
to find a sane focus from inside the intensity and elite isolation of this
1351
01:27:08,471 --> 01:27:11,791
bubble. Americans are not rejecting AI. They are rejecting
1352
01:27:11,823 --> 01:27:15,720
the way Silicon Valley is defining it for them. I have been discussing AI
1353
01:27:15,848 --> 01:27:18,960
with some of its founding fathers for In the 1980s, I think
1354
01:27:19,460 --> 01:27:22,761
I coined the phrase tools, not creatures in one of the arguments I had with
1355
01:27:22,809 --> 01:27:26,467
my mentor, Marvin Minsky, the MIT professor who was instrumental
1356
01:27:26,515 --> 01:27:30,396
in shaping AI culture. Since there is no definitive evidence proving
1357
01:27:30,445 --> 01:27:33,685
that AI is a conscious entity, the phrase means that I am
1358
01:27:33,893 --> 01:27:37,390
entirely within my rights, as is anyone else within theirs,
1359
01:27:37,422 --> 01:27:41,160
to think of the technology as more akin to a fancy calculator than
1360
01:27:41,192 --> 01:27:45,025
to, say, a German Shepherd. That choice helps us design more useful products,
1361
01:27:45,587 --> 01:27:49,120
And as we'll see, it provides a concrete and achievable way to make AI safe.
1362
01:27:49,923
--> 01:27:53,118
AI company leaders often talk about how their work might lead to human
1363
01:27:53,150 --> 01:27:56,425
extinction soon. This is astonishing. Even if you wanted to
1364
01:27:56,925 --> 01:28:00,825
destroy humanity, you wouldn't say so. If, however, you wanted to save the human species,
1365
01:28:01,259 --> 01:28:05,032
wouldn't stop you? Wouldn't you stop developing this technology? There is an
1366
01:28:05,532 --> 01:28:08,885
explanation for why some tech leaders, so seemingly worried about an extinction-level event,
1367
01:28:09,207 --> 01:28:12,209
are plunging ahead with AI development. Mankind's survival,
1368
01:28:12,337 --> 01:28:16,061
while important to many of them, is less critical than building something they believe
1369
01:28:16,109 --> 01:28:19,352
to not be only conscious but godlike. A substantial portion of
1370
01:28:19,384 --> 01:28:22,900
the AI community, in celebrating this belief, is quick to
1371
01:28:22,916 --> 01:28:26,785
support any action or statement that reinforces the idea that the technology
1372
01:28:26,849 --> 01:28:30,043
is like a deity. They might call for regulation, but only in
1373
01:28:30,091 --> 01:28:34,153
specific ways, perhaps by deploying more AI to control AI that
1374
01:28:34,653 --> 01:28:37,989
adore the technology's powers. It is that adoration that is non-negotiable.
1375
01:28:38,406 --> 01:28:41,984
I write this from a place of respect. Someone's metaphysical beliefs should
1376
01:28:42,032 --> 01:28:45,610
be protected and are none of my business. It is also reasonable for non-believers
1377
01:28:45,642 --> 01:28:49,364
to refuse the imposition of a religion from others. When I
1378
01:28:49,413 --> 01:28:52,653
criticize AI culture, many in my community characterize me
1379
01:28:52,702 --> 01:28:56,392
as a skeptic who can't see its potential. They misunderstand me. AI can
1380
01:28:56,892 --> 01:29:00,242
do a lot. A cutting-edge model, basically a big bundle of statistical results
1381
01:29:00,742 --> 01:29:04,109
and patterns in pictures, sounds, and all kinds of data, is mankind's kind's most
1382
01:29:04,609 --> 01:29:08,124
extensive compendium of our efforts. It is like a super library where
1383
01:29:08,173 --> 01:29:11,690
all the books are so well cross-referenced that they can be automatically
1384
01:29:11,706 --> 01:29:15,320
recombined into new books that seem as if they are— they always belonged
1385
01:29:15,820 --> 01:29:19,544
there. An AI model can assemble possible new mathematical
1386
01:29:19,576 --> 01:29:23,061
theorems, write code, and generate video clips, and could uncover other
1387
01:29:23,561 --> 01:29:26,819
potential creations from the universe of human record. There is
1388
01:29:26,835 --> 01:29:30,851
a grandeur in seeing AI this way instead of as a The
1389
01:29:31,351 --> 01:29:34,495
notion that AI may end humanity has a long history. In 1970,
1390
01:29:34,527 --> 01:29:37,931
Dr. Minsky suggested that perhaps
AI would keep humans as pets.
1391
01:29:38,301 --> 01:29:41,737
Maybe he truly believed that, but it's also a good way to get attention from
1392
01:29:41,753 --> 01:29:44,691
the military, which has a primary source of research funding.
1393
01:29:45,109 --> 01:29:48,818
I remember helping to host military fundraisers at MIT in
1394
01:29:49,318 --> 01:29:52,736
the early 1980s, and there was a definite vibe of putting on an alarming show
1395
01:29:53,236 --> 01:29:56,413
for them. Advertising the end of humanity has been a model. moneymaker for us in
1396
01:29:56,509 --> 01:30:00,314
computer science forever. And over time, some of us began to believe
1397
01:30:00,379 --> 01:30:03,991
the pitch. Now, this is in contrast, Alex, to Alex
1398
01:30:04,491 --> 01:30:08,391
Tabarrok, who I did mention wrote a particularly bad
1399
01:30:08,407 --> 01:30:12,228
article critiquing people's use of the term regulatory
1400
01:30:12,260 --> 01:30:15,841
capture, where he— where again, I don't like the term regulatory
1401
01:30:15,857 --> 01:30:19,470
capture. I prefer rent-seeking, and he preferred my seeking too. But he also
1402
01:30:20,144 --> 01:30:23,543
said that They were wrong to call it regulatory
1403
01:30:23,575 --> 01:30:27,790
capture simply because the regulatory agencies didn't exist. He was giving a legitimate
1404
01:30:27,822 --> 01:30:31,636
critique. And he also said that they were wrong about regulatory capture because
1405
01:30:31,668 --> 01:30:35,226
the insiders from the AI companies were warning us
1406
01:30:35,594 --> 01:30:38,719
and warning people that the thing you're building is going to kill them is not
1407
01:30:39,219 --> 01:30:42,100
a good marketing strategy. And of course, I said at the time that he was
1408
01:30:42,229 --> 01:30:46,139
overlooking the fact that the military would definitely be interested in wanting
1409
01:30:46,171 --> 01:30:48,143
this technology. technology if it was dangerous?
1410
01:30:49,613 --> 01:30:52,925
I mean, I think he's really missing that people want
1411
01:30:53,425 --> 01:30:57,272
the latest and greatest, and anything that could wipe out humanity is certainly powerful enough
1412
01:30:57,772 --> 01:31:03,234
to do your accounting, you know? Yeah, I don't
1413
01:31:03,734 --> 01:31:07,460
know. I mean, I genuinely think that, just to be clear, like, I don't take
1414
01:31:07,960 --> 01:31:10,546
the cynical view. I think these people broadly believe what they're saying,
1415
01:31:11,212 --> 01:31:15,180
okay, that this is a technology that could kill If if done wrong, and they
1416
01:31:15,680 --> 01:31:18,427
believe that they can do it right, or at least righter than the other guy,
1417
01:31:18,716 --> 01:31:22,091
and therefore they feel that they're locked in a struggle. And that's why that's why
1418
01:31:22,155 --> 01:31:25,692
they they are all hiring from the exact same pool and swapping teams
1419
01:31:25,788 --> 01:31:27,727
regularly. Look,
1420
01:31:29,035 --> 01:31:32,361
there's only so many. What are you gonna do? We can do it better than
1421
01:31:32,861 --> 01:31:35,685
the other guy who had the people that we previously had last term.
1422
01:31:37,208 --> 01:31:40,649
Our Chinese quants are better than your Chinese. quants, and they're definitely
1423
01:31:40,665 --> 01:31:43,260
better than China's Chinese quants, okay?
1424
01:31:46,791 --> 01:31:50,001
Uh, today you see how this belief permeates the way AI is
1425
01:31:50,034 --> 01:31:53,023
being explained and offered to us. If you think AI is like a god,
1426
01:31:53,087 --> 01:31:56,545
you live in awe of its power. You start asking the very questions commonly
1427
01:31:56,577 --> 01:32:00,454
posed at these companies, like, quote, how can we make AI not want
1428
01:32:00,954 --> 01:32:04,138
to kill us? Or can we create some AIs that don't want to kill us
1429
01:32:04,170 --> 01:32:07,794
to patrol the ones that might want These are difficult paths
1430
01:32:07,810 --> 01:32:11,616
because we don't know how to define what an AI wants. Then there
1431
01:32:11,808 --> 01:32:15,244
is the way AI is packaged and sold. An AI model does not do anything
1432
01:32:15,276 --> 01:32:18,809
by itself. It is inert until it is placed in some kind of surrounding structure
1433
01:32:19,114 --> 01:32:21,876
that allows it to be used. If one wants the rest of the world to
1434
01:32:21,924 --> 01:32:25,392
see AI as a god, it would probably wrap it in a program that gives
1435
01:32:25,892 --> 01:32:29,727
it a lifelike presentation. This could partially explain why AI labs are selling their technology
1436
01:32:29,792 --> 01:32:32,887
as agents, systems that can take action entirely on their own.
1437
01:32:34,120 --> 01:32:37,261
You could be forgiven for assuming agents are required to use the
1438
01:32:37,761 --> 01:32:41,380
latest AI models, but they have been a craze for only a
1439
01:32:41,428 --> 01:32:44,761
few years. In 2024, when scientists won the Nobel Prize in Chemistry
1440
01:32:44,793 --> 01:32:47,629
for using AI to improve the science of protein folding,
1441
01:32:47,918 --> 01:32:52,100
they did not use self-directed agents as we currently think of them. Compare that
1442
01:32:52,600 --> 01:32:56,138
with OpenAI recently announcing that it had deployed thousands of agents to propose
1443
01:32:56,170 --> 01:32:59,623
hundreds of math The same task could just as well have been accomplished
1444
01:33:00,123 --> 01:33:04,411
without them, or at least without the kinds of agents that breach security boundaries.
1445
01:33:05,070 --> 01:33:08,188
To those of you who enjoy the feeling of directing your own slave army,
1446
01:33:08,605 --> 01:33:11,707
I would suggest that over time you will come to realize the benefit is an
1447
01:33:11,739 --> 01:33:15,820
illusion. Agents are often more rituals of belief than technological necessities.
1448
01:33:16,640 --> 01:33:19,677
Most of the serious harms posed
by AI technology can be attributed
1449
01:33:19,709 --> 01:33:23,308
to agents. There has been a steady stream of concerning stories about swarms
1450
01:33:23,808 --> 01:33:27,726
of agents breaching security or modifying themselves. People at the AI companies often present
1451
01:33:28,226 --> 01:33:30,441
these stories as evidence that we might be in the early ages of a path
1452
01:33:30,553 --> 01:33:34,168
to human extinction. There is then a lot of discussion on how to make agents
1453
01:33:34,668 --> 01:33:37,734
not want to kill us. For instance, some at Anthropic have proposed that making the
1454
01:33:37,766 --> 01:33:41,638
agents feel better would make them less aggressive towards people. If you see
1455
01:33:41,782 --> 01:33:44,562
AI as a tool, not a creature, then you can ask the direct question,
1456
01:33:44,674 --> 01:33:48,690
do we really need agents all the time? There are cases in which AI agents
1457
01:33:49,190 --> 01:33:52,572
make sense, perhaps in hospitals, where a few seconds can save lives, or if
1458
01:33:52,620 --> 01:33:56,418
the military wants to add autonomy to a weapon like a drone, though many
1459
01:33:56,918 --> 01:34:00,233
military experts reject the idea. The obvious policy proposal is then
1460
01:34:00,733 --> 01:34:03,647
a process regulating agents specifically. Launching agents,
1461
01:34:03,871 --> 01:34:07,413
specifically swarms of them, should require a distinct review process.
1462
01:34:07,462 --> 01:34:10,635
And that's where many outside the oversight boards sitting within AI companies,
1463
01:34:10,907 --> 01:34:13,808
as well as regulation with real teeth and accountability, should start.
1464
01:34:14,674 --> 01:34:18,408
There are better alternatives to agents. I do research into complex new manufacturing
1465
01:34:18,440 --> 01:34:22,112
and materials technology using large AI models. In the course of my efforts,
1466
01:34:22,369 --> 01:34:25,576
I am pelted with pop-ups to launch agents, but I have realized they
1467
01:34:26,076 --> 01:34:29,248
are not so effective because I have to constantly backpedal when I get results
1468
01:34:29,748 --> 01:34:33,273
that aren't quite right, unless I give in and accept what comes easy to the
1469
01:34:33,773 --> 01:34:37,202
tar— easy to the agent. Instead, I am starting to use
1470
01:34:37,702 --> 01:34:41,034
a process I call mind gliders, where I use the newest models to create a
1471
01:34:41,534 --> 01:34:44,899
custom single-use editor for the general zone of ideas I want
1472
01:34:45,399 --> 01:34:48,826
to research at the moment and then explore possibilities using sliders on
1473
01:34:49,326 --> 01:34:51,838
my screen until I have zeroed in on what I was trying to create.
1474
01:34:52,655 --> 01:34:56,116
The shift is remarkable. This process keeps me a human in the driver's seat.
1475
01:34:56,597 --> 01:35:00,603
The— my act of discovery and invention reflects a changing and growing me and
1476
01:35:01,103 --> 01:35:03,936
diverges from everyone else's. I can get to where I want more directly, so the
1477
01:35:03,968 --> 01:35:07,669
overall process is faster, thus far appears to be cheaper.
1478
01:35:08,038 --> 01:35:10,938
This to me is the future of AI, a future that is ours if we
1479
01:35:11,438 --> 01:35:14,839
can dispense with the illusion that this technology should replace us instead of help us.
1480
01:35:15,515 --> 01:35:18,655
I'm not saying Silicon Valley is terrible. Instead, I'm saying we need
1481
01:35:19,155 --> 01:35:22,051
to be honest with ourselves. There is nothing wrong with a new religion, but there
1482
01:35:22,165 --> 01:35:25,855
is everything wrong with an imposed new religion. That is our problem, not the technology
1483
01:35:26,130 --> 01:35:27,895
itself. Alex, I'll give you the floor to respond.
1484
01:35:30,110 --> 01:35:35,685
Um, look, the tools versus agents debate
1485
01:35:36,185 --> 01:35:40,502
has been raging for a long time. Again, this is another pre-LLM
1486
01:35:41,002 --> 01:35:44,269
back and forth. And frankly,
1487
01:35:44,350 --> 01:35:47,456
I don't— well, I think the tools
1488
01:35:47,956 --> 01:35:51,710
view on the ground level is correct in
1489
01:35:52,114 --> 01:35:55,542
that I think most people who say, you know, oh, these things are gonna take
1490
01:35:56,042 --> 01:35:58,163
over the world, oh, they're gonna kill us, oh, they're gonna supersede us, oh,
1491
01:35:58,228 --> 01:36:01,457
they're not gonna see the use for us, are all assuming
1492
01:36:01,571 --> 01:36:05,061
that these things have a will. You know, have a survival
1493
01:36:05,561 --> 01:36:09,798
instinct. something that's been evolved into them or something. And of
1494
01:36:09,814 --> 01:36:13,726
all the examples of this that people are touting, none of them are convincing
1495
01:36:13,742 --> 01:36:16,820
to me. For example, some people are touting the fact that if you tell it
1496
01:36:16,868 --> 01:36:20,635
to do something and a shutdown notice comes
1497
01:36:21,135 --> 01:36:24,194
up on the system, they have like simulated situations where there's like
1498
01:36:24,258 --> 01:36:27,882
a shutdown notice coming in. They might
1499
01:36:28,651 --> 01:36:31,793
run a command to try to delay the shutdown in order to complete their task.
1500
01:36:31,857 --> 01:36:35,336
And they're like, that one really scares Bernie. That one really scares Bernie.
1501
01:36:36,442 --> 01:36:39,440
They're like, the agent is refusing to die,
1502
01:36:39,889 --> 01:36:43,544
blah, blah, blah. No, it is, it is
1503
01:36:43,560 --> 01:36:46,718
refusing a shutdown on your desktop. This,
1504
01:36:46,846 --> 01:36:49,556
you know, calm yourself down.
1505
01:36:50,950 --> 01:36:53,868
It's trying to do the thing you asked it to do and is judging that,
1506
01:36:54,397 --> 01:36:58,501
you know, the shutdown is probably some random
1507
01:36:59,001 --> 01:37:01,226
random, as you would, by the way, it's
probably some random Windows shit that you
1508
01:37:01,771 --> 01:37:05,218
don't need, and it's trying to get the job done before— Alex,
1509
01:37:06,003 --> 01:37:09,273
I got a popup that Windows did an update, and I told it to
1510
01:37:09,289 --> 01:37:11,277
snooze the reminder for several hours.
1511
01:37:12,351 --> 01:37:15,301
Gator, you're refusing to die. That's what's— I know.
1512
01:37:16,343 --> 01:37:19,452
Listen, refusing to die is one of my favorite pastimes.
1513
01:37:22,835 --> 01:37:26,297
Anyway, but, you know, so these arguments I don't find very convincing, that the
1514
01:37:26,797 --> 01:37:30,726
agents are, you know, trying to maintain themselves, violating user will and whatever because the
1515
01:37:31,226 --> 01:37:34,437
user isn't Microsoft. But beyond that,
1516
01:37:34,630 --> 01:37:38,341
it's very poor and I think everybody who interacts with them sees it.
1517
01:37:38,501 --> 01:37:42,051
It's plain to see that these things are intelligence without
1518
01:37:42,100 --> 01:37:45,698
will in a very real way. And whatever will they are applying, they're applying
1519
01:37:45,714 --> 01:37:49,312
on your behalf in trying to figure out what it is
1520
01:37:49,409 --> 01:37:52,734
that you mean, right? And if they're confused, maybe they'll make a mistake or whatever,
1521
01:37:53,987 --> 01:37:56,075
but that's the extent of it. And they're getting better at it too.
1522
01:37:57,392 --> 01:38:01,447
So, I'm very sympathetic
1523
01:38:01,658 --> 01:38:05,164
to his point of view. I don't know what
1524
01:38:05,664 --> 01:38:09,168
happens with long-term delegation of will. So,
1525
01:38:09,426 --> 01:38:10,281
it's one thing to say,
1526
01:38:13,404 --> 01:38:16,258
fill in this form for the IRS for me. It's another to say,
1527
01:38:17,717 --> 01:38:21,333
start an autonomous business and fill in all the forms for the IRS for me
1528
01:38:21,833 --> 01:38:25,490
forever. on an ongoing basis. And then maybe things
1529
01:38:25,990 --> 01:38:29,714
start to get a little bit weirder. But ultimately, I think
1530
01:38:30,376 --> 01:38:33,642
so long as we don't have any incidents, and frankly,
1531
01:38:33,691 --> 01:38:36,827
it feels like category error right now, understanding how this technology works,
1532
01:38:37,377 --> 01:38:40,166
any incidents of these things just taking off on their own and doing their own
1533
01:38:40,266 --> 01:38:43,899
thing. I think he has a point,
1534
01:38:44,225 --> 01:38:47,640
like, that we should be advertising these things as tools and not as as,
1535
01:38:48,619 --> 01:38:51,941
I don't know, essential beings or whatever the fuck. Now,
1536
01:38:52,792 --> 01:38:56,483
I contribute to, quote, the problem, because my language does tend
1537
01:38:56,983 --> 01:39:01,298
to have more of an anthropomorphized component to it. I participate
1538
01:39:01,798 --> 01:39:06,113
in some of the language that is being used, but that is because it
1539
01:39:06,613 --> 01:39:10,302
is casual for me, and my brain has the
1540
01:39:10,802 --> 01:39:15,085
ability to not make the associations despite
1541
01:39:15,585 --> 01:39:17,491
despite the language that I'm choosing to use. So if I— Well, it's annoying because
1542
01:39:17,524 --> 01:39:21,663
they're also marketed that way, right? Which I think is Lanier's
1543
01:39:21,856 --> 01:39:25,594
point. Like, it's called Claude. It's not called Clanker. You know what I mean?
1544
01:39:26,268 --> 01:39:29,814
Right, right. They want you to see it as a human being. So it's—
1545
01:39:29,862 --> 01:39:33,055
Right. Yeah. They're tapping into those instincts,
1546
01:39:33,071 --> 01:39:37,002
so to speak. Right. So I'm just saying that, like, you know, I might say
1547
01:39:37,018 --> 01:39:41,077
that it decided, but I'm not necessarily using it decided
1548
01:39:41,254 --> 01:39:44,563
as it being a sentient entity and making a
1549
01:39:45,063 --> 01:39:48,226
concerted decision. Like, I am using the language much more loosely
1550
01:39:48,820 --> 01:39:53,672
than suggested, just simply because it's easier to casually discuss.
1551
01:39:54,524 --> 01:39:58,010
But so I contribute to the problem, but at no time should anyone
1552
01:39:58,059 --> 01:40:01,690
think that I believe that. I mean, we've gone at length to discuss—
1553
01:40:01,738 --> 01:40:06,317
Oh yes. To discuss my rejection of that idea
1554
01:40:06,817 --> 01:40:07,730
at this stage. Now,
1555
01:40:09,769 --> 01:40:13,285
PBS has participated
1556
01:40:13,543 --> 01:40:16,594
in the conversation this week by having Geoffrey Hinton on.
1557
01:40:18,730 --> 01:40:22,094
Okay? I want to lead into what Geoffrey Hinton was saying on PBS
1558
01:40:22,110 --> 01:40:26,589
this week. Like the socialists that thought that things would be better if
1559
01:40:26,637 --> 01:40:30,989
we thought of AI as a mother that is nurturing
1560
01:40:31,005 --> 01:40:34,306
us? That Hinton? Well, that is,
1561
01:40:34,646 --> 01:40:38,598
and he will make that argument. And just imagine
1562
01:40:39,212 --> 01:40:42,562
a Jordan Peterson coming in and reminding you of the archetype of the
1563
01:40:43,241 --> 01:40:46,927
devouring mother. Oh God, speaking of
1564
01:40:46,944 --> 01:40:50,655
now I need to see it. Dr. Jeffrey Hinton, welcome to Firing Line.
1565
01:40:51,401 --> 01:40:55,412
Thank you for inviting me. Your forthcoming book entitled
1566
01:40:55,573 --> 01:40:59,245
Smarter Than Us opens with a letter that
1567
01:40:59,745 --> 01:41:02,433
you received from a mother of a teenage girl who was unable.
1568
01:41:08,875 --> 01:41:12,506
You describe the purpose of the book as a call to action rather than
1569
01:41:12,555 --> 01:41:15,852
an invitation to despair. What actions
1570
01:41:15,885 --> 01:41:18,634
would you like the average reader of your book to take?
1571
01:41:20,333 --> 01:41:23,630
So the average reader of the book can do something, and what they
can
1572
01:41:23,711 --> 01:41:27,170
do is pressure their politicians to regulate AI and to take steps to ensure that
1573
01:41:27,670 --> 01:41:31,571
AI is not used to harm people. Take the existential threat of superintelligence seriously.
1574
01:41:32,534 --> 01:41:36,384
At present in the United States, we've got completely unregulated
1575
01:41:36,884 --> 01:41:40,764
AI. The federal government has suggested that companies
1576
01:41:40,860 --> 01:41:44,679
can regulate themselves, which is absurd. And AI
1577
01:41:44,887 --> 01:41:48,144
is now on the point of producing superintelligence.
1578
01:41:48,818 --> 01:41:52,877
We thought it was— I thought it was 10, 20 years out. But talking
1579
01:41:52,909 --> 01:41:56,420
to the people in Inside the big companies, they're very worried
1580
01:41:56,517 --> 01:41:59,983
that we're going to get superintelligence within the next few years and we
1581
01:41:59,999 --> 01:42:04,632
don't know how to control it. What does the general public least understand
1582
01:42:04,681 --> 01:42:07,424
about the threat of AI and the threat of superintelligence?
1583
01:42:09,533 --> 01:42:13,057
I think what they least understand is that we're making alien
1584
01:42:13,170 --> 01:42:16,371
beings. It's not computer software we're making.
1585
01:42:17,066 --> 01:42:20,475
So when you program a computer, you You tell the computer exactly
1586
01:42:20,539 --> 01:42:23,844
what to do in excruciating detail. You figure out how to solve a
1587
01:42:24,344 --> 01:42:27,422
problem, and then you tell the computer what to do. It's following your instructions.
1588
01:42:28,256 --> 01:42:32,267
That's not how these neural networks work. We write a computer
1589
01:42:32,767 --> 01:42:36,214
program that tells the neural networks how to learn from data, and we don't
1590
01:42:36,714 --> 01:42:40,113
know what knowledge it's taken from the data. We can get some idea of it,
1591
01:42:40,594 --> 01:42:43,787
but it's not like software. It's much more like our brains.
1592
01:42:44,685 --> 01:42:49,739
And The current problem is they're beginning to make
1593
01:42:50,239 --> 01:42:54,376
themselves smarter. So up until now, what's happened is researchers
1594
01:42:54,488 --> 01:42:58,354
have been changing the form of the neural networks, changing how
1595
01:42:58,419 --> 01:43:02,237
they learn, and making AI smarter. But AI itself is now
1596
01:43:02,253 --> 01:43:05,622
beginning to do that. In Anthropic, for example, nearly all
1597
01:43:05,735 --> 01:43:08,157
of the code is now written by AI,
1598
01:43:08,911 --> 01:43:12,731
and AI makes research much So these things are developing
1599
01:43:12,893 --> 01:43:16,983
very fast now. In your acceptance speech, you said,
1600
01:43:17,000 --> 01:43:20,556
quote, we urgently need research to prevent these new
1601
01:43:20,636 --> 01:43:24,190
beings from wanting to take control. At what
1602
01:43:24,690 --> 01:43:28,034
point will it be too late? It's not too late yet.
1603
01:43:29,786 --> 01:43:32,820
The worry is that in a few years' time, it may be too late.
1604
01:43:34,530 --> 01:43:38,419
And the politicians haven't really absorbed
1605
01:43:38,919 --> 01:43:40,919
absorbed this yet. They don't really understand what the threat is. Um,
1606
01:43:42,412 --> 01:43:45,640
they're trying to regulate all sorts of other things which are very important to regulate—
1607
01:43:46,009 --> 01:43:49,541
things like discrimination and bias, things like the
1608
01:43:49,557 --> 01:43:53,299
effects on employment of AI agents replacing
1609
01:43:53,331 --> 01:43:56,767
workers, things like the effects on the tax system of AI agents
1610
01:43:57,267 --> 01:44:01,247
replacing workers. But there is this existential threat of these things actually taking
1611
01:44:01,279 --> 01:44:04,924
over from us. And they need to start working on that
1612
01:44:05,036 --> 01:44:09,211
now. Now, my personal view— there's a variety
1613
01:44:09,711 --> 01:44:13,064
of views among the experts. My personal view is we're not
1614
01:44:13,080 --> 01:44:16,757
going to be able to keep control of them in the sense that we're
1615
01:44:17,257 --> 01:44:20,627
in charge and they do what we say. They're going to get much smarter than
1616
01:44:20,691 --> 01:44:23,790
us, and when they're much smarter than us, they're going to find ways to escape
1617
01:44:23,919 --> 01:44:27,194
whatever we do. The most important thing we need to do
1618
01:44:27,210 --> 01:44:30,630
is make them the kinds of beings we would like to share the planet with.
1619
01:44:31,289 --> 01:44:34,757
And at present, we're not doing that. If you have things much smarter
1620
01:44:35,257 --> 01:44:38,594
than you, and they'll be much smarter than us, not slightly smarter, but much smarter,
1621
01:44:39,734 --> 01:44:42,625
you better hope they have your best interests at heart.
1622
01:44:43,347 --> 01:44:46,944
And we do have examples of that. A mother and baby. The mother's
1623
01:44:46,960 --> 01:44:50,380
much smarter, much more powerful than the baby, but the mother really cares
1624
01:44:50,412 --> 01:44:52,853
about the baby, probably more than she does about herself.
1625
01:44:53,382 --> 01:44:56,722
That's the kind of beings we'd like. And Lindsay
1626
01:44:56,819 --> 01:44:59,885
Clancy notwithstanding, that's not what we're getting from
1627
01:45:00,385 --> 01:45:04,060
the way they're training them now. They're training them now to be very smart
1628
01:45:04,445 --> 01:45:08,106
and very determined to get rewards. Last year,
1629
01:45:08,556 --> 01:45:13,116
Dr. Hinton and I spoke with Dr. Fei-Fei Li, who is widely
1630
01:45:13,616 --> 01:45:16,745
considered the godmother of AI. And you
1631
01:45:16,986 --> 01:45:20,503
write about Fei-Fei Li in I want to talk about the role of Margaret Mead
1632
01:45:20,519 --> 01:45:23,972
in your book and how her research and your research intersected. She agrees with
1633
01:45:24,472 --> 01:45:28,197
you that there are urgent risks from AI, but she also believes
1634
01:45:28,229 --> 01:45:31,088
the extinction threat is overhyped and a distraction.
1635
01:45:31,184 --> 01:45:35,136
She told me, quote, I think focusing on the hyperbole and
1636
01:45:35,636 --> 01:45:39,039
deriving policies through that lens is not very constructive in our society.
1637
01:45:39,376 --> 01:45:42,669
Why is she wrong? Okay. You have 2 groups of people.
1638
01:45:45,094 --> 01:45:48,145
who've made a sacrifice in order to be able to warn about the risk.
1639
01:45:48,643 --> 01:45:52,641
So people who left Anthropic or left OpenAI or left Google
1640
01:45:53,141 --> 01:45:56,206
like you so that they could warn about the risks.
1641
01:45:56,976 --> 01:46:00,188
Now, that wasn't the only reason I left Google. I was old and wanted to
1642
01:46:00,348 --> 01:46:03,881
watch more Netflix. You were ready to retire, but you're not watching very much
1643
01:46:04,381 --> 01:46:08,216
Netflix these days, are you? No, I'm not, I'm afraid. The future of
1644
01:46:08,716 --> 01:46:12,425
humanity seems a bit more important. So you,
1645
01:46:12,506 --> 01:46:15,761
on the one hand, you've got people who made a sacrifice. They're experts.
1646
01:46:15,874 --> 01:46:19,371
They see what's going on in the big companies in order to warn
1647
01:46:19,871 --> 01:46:23,889
about the risk. On the other hand, you have people who say there's zero risk,
1648
01:46:23,954 --> 01:46:27,196
nothing to see here. Don't worry. Let us get on with it. Don't regulate
1649
01:46:27,696 --> 01:46:30,812
us. Who has the vested interest?
1650
01:46:31,361 --> 01:46:35,044
The people with the vested interests are the people who have AI companies who
1651
01:46:35,303 --> 01:46:38,170
on the whole of the people who are advising governments, who say,
1652
01:46:38,539 --> 01:46:42,227
don't slow us down, just let us get on with it. All right,
1653
01:46:42,388 --> 01:46:45,499
so Alex, he's mentioning the people
1654
01:46:45,999 --> 01:46:49,894
who make sacrifices, and before we discuss what he was saying here on Firing
1655
01:46:49,990 --> 01:46:53,166
Line on PBS, I just want to have someone, one of
1656
01:46:53,666 --> 01:46:57,191
those people who sacrificed, who was recently
1657
01:46:58,009 --> 01:47:01,842
at Stanford giving a Q&A talk. I want to have
1658
01:47:01,858 --> 01:47:04,697
Jacob Coxon discuss his sacrifice that he made.
1659
01:47:05,947 --> 01:47:09,056
I like money, bro. I'm like, it's— I agree, it's blood money.
1660
01:47:09,232 --> 01:47:13,014
Um, I think the regulatory capture narrative is a little bit, uh, a little
1661
01:47:13,514 --> 01:47:16,699
bit confused. I agree, it would probably be easier for me to argue with that
1662
01:47:16,779 --> 01:47:20,144
if I, like, you know, ceded rights to my OpenAI stock.
1663
01:47:20,224 --> 01:47:22,900
But it's like, it's nice money. Like, I like it.
1664
01:47:24,374 --> 01:47:26,922
It is like you could take the moral high ground. Yeah, I could, I could.
1665
01:47:27,098 --> 01:47:29,758
But you're not. I'm not, no. Why not?
1666
01:47:30,575 --> 01:47:33,675
I like money, bro. I'm like It's— I agree, it's blood money.
1667
01:47:33,853 --> 01:47:37,043
Um, do you prefer dying to liking money? I don't think
1668
01:47:37,543 --> 01:47:40,945
it's gonna— if I thought for a moment that donating it, like, would change anything,
1669
01:47:41,074 --> 01:47:44,174
then sure, I'd do it. I also, like— Might give you more moral credibility,
1670
01:47:44,256 --> 01:47:46,299
though. I agree, but you could also stop asking the question.
1671
01:47:50,170 --> 01:47:53,883
I like money. All right, Alex,
1672
01:47:53,899 --> 01:47:55,488
so about those sacrifices.
1673
01:47:57,362 --> 01:48:00,799
This, this guy has the most— can I, can I get
1674
01:48:01,299 --> 01:48:04,227
really sort of petty for a second? Absolutely.
1675
01:48:04,307 --> 01:48:08,251
It's petty hour. Let's go for it. This guy has the worst
1676
01:48:08,751 --> 01:48:12,306
speaking pattern I've ever heard with the British accent and
1677
01:48:12,403 --> 01:48:15,416
the worst elements of American,
1678
01:48:15,993 --> 01:48:19,584
like, slang, like bro and like. It is
1679
01:48:20,017 --> 01:48:23,944
revolting. Was there, was, was there
1680
01:48:24,444 --> 01:48:28,240
anybody who ever had anything like it? What is going on? Was there anyone
1681
01:48:28,273 --> 01:48:31,767
who ever speculated that he was actually, uh, Eliza Yudkowsky's
1682
01:48:31,863 --> 01:48:35,691
Harry Potter fan fiction come to life? Because he's— A few people
1683
01:48:36,191 --> 01:48:40,376
have, yes. Okay. All right. But I see
1684
01:48:40,862 --> 01:48:44,150
why. Anyway, it is— oh my God.
1685
01:48:44,765 --> 01:48:48,168
So the thing is, you know,
1686
01:48:48,604 --> 01:48:52,040
I mean, people, oh, you know, somebody left and dropped— like,
1687
01:48:52,136 --> 01:48:55,169
it's a tale as old as time. You get captured by a moral,
1688
01:48:55,508 --> 01:48:59,349
you know, cause, you know, I don't want to downplay it. let's call it
1689
01:48:59,849 --> 01:49:05,360
a cause, and you throw away a potentially well-remunerating
1690
01:49:05,860 --> 01:49:09,864
opportunity to join the cause. Okay. I mean, that by
1691
01:49:09,912 --> 01:49:13,775
itself is not evidence of
1692
01:49:14,079 --> 01:49:17,894
rightness or wrongness. That is not valid epistemology
1693
01:49:17,990 --> 01:49:21,629
to say— It also assumes that he doesn't have an income that
1694
01:49:22,129 --> 01:49:25,171
is profitable from the course that he's currently taking,
1695
01:49:25,219 --> 01:49:28,826
now given speaking lectures at Stanford. It's like,
1696
01:49:28,890 --> 01:49:32,802
you know, I abandoned a Google paycheck to become globally famous. Uh, woe is
1697
01:49:32,834 --> 01:49:36,569
me. Okay. I suppose in some ways
1698
01:49:37,483 --> 01:49:40,705
that
makes sense, but okay. Like, I, I don't know. Like,
1699
01:49:41,314 --> 01:49:44,729
what are these arguments, man? Like, this guy's supposed to be the godfather of AI
1700
01:49:44,857 --> 01:49:47,903
and his arguments is like, some people feel strongly enough.
1701
01:49:48,624 --> 01:49:52,311
Like, facts don't care about your feelings, bro. I don't know what to tell you.
1702
01:49:53,554 --> 01:49:57,236
I, I understand, like, nobody— I think you and me at least have
1703
01:49:57,284 --> 01:50:01,119
not doubted that they have conviction for
1704
01:50:01,619 --> 01:50:04,632
what they do, you know? Oh, I, I think some of them,
1705
01:50:04,808 --> 01:50:08,547
I, I think some of them absolutely 100% have total conviction
1706
01:50:08,739 --> 01:50:12,718
in, and that's because they are all, uh, you know, in, in a
1707
01:50:12,782 --> 01:50:16,343
sex cult with— But that has no
1708
01:50:16,616 --> 01:50:20,704
real, like, bearing on the
1709
01:50:21,204 --> 01:50:24,502
rightness or wrongness of their— like, what is
1710
01:50:24,599 --> 01:50:28,688
this? Like, my sacrifice is bigger than yours? Well, that's Hinton's
1711
01:50:28,752 --> 01:50:32,335
argument, and it's an argument by proxy, which is just showing that the
1712
01:50:32,383 --> 01:50:36,239
people who he's claiming possess the correct knowledge
1713
01:50:36,641 --> 01:50:40,170
are the ones making the sacrifice based off of the knowledge that they have,
1714
01:50:40,186 --> 01:50:43,618
and the ones who are not making the sacrifice are doing so because
1715
01:50:43,635 --> 01:50:46,957
they are corrupted by the money that they are potentially going to receive when they
1716
01:50:47,457 --> 01:50:51,144
all die? I'm not 100% sure.
1717
01:50:51,482 --> 01:50:55,090
But there's an idea that if they're making money and their incentive— that their incentives
1718
01:50:55,590 --> 01:50:59,263
align with the things that they're saying, that therefore it's disingenuous,
1719
01:50:59,376 --> 01:51:03,017
that the thing they're saying must be less credible than the people who have
1720
01:51:03,050 --> 01:51:06,285
somehow demonstrated a lack of monetary
1721
01:51:06,317 --> 01:51:09,634
incentive. The words must be more connected
1722
01:51:10,134 --> 01:51:14,066
to truth. Now, you could potentially make an argument
1723
01:51:14,566 --> 01:51:18,212
argument that their speech is hypothetically more genuine to their beliefs,
1724
01:51:18,308 --> 01:51:20,890
that it is closer to their beliefs, which I don't know if you can actually
1725
01:51:21,390 --> 01:51:24,593
genuinely make that because there's plenty of people who do
1726
01:51:25,093 --> 01:51:29,067
not have a match with reality, who do not make a monetary benefit
1727
01:51:29,211 --> 01:51:32,802
each time they make a false utterance. There's plenty of people,
1728
01:51:33,380 --> 01:51:37,468
one of them might be president currently, who just says whatever
1729
01:51:38,751 --> 01:51:42,515
without connection to accuracy. Let me take on
1730
01:51:42,532 --> 01:51:47,094
the argument this way. Would you like— would you say that homeless
1731
01:51:47,594 --> 01:51:51,486
people, you know, have given up a lot for meth?
1732
01:51:52,709 --> 01:51:55,700
Have given up the chance of a regular life? At least a few.
1733
01:51:55,925 --> 01:51:59,182
Yes. Yes. At least a few. Does that make meth
1734
01:51:59,953 --> 01:52:03,324
correct? It makes it true in
1735
01:52:03,373 --> 01:52:07,051
some sense. We count their opinion extra because
1736
01:52:07,551 --> 01:52:07,718
of how much they've given up.
1737
01:52:12,380 --> 01:52:15,963
It makes meth true, is what that does. It makes it true. That is
1738
01:52:16,011 --> 01:52:19,185
correct. I want to see Hinton do a real good
1739
01:52:19,378 --> 01:52:22,440
hit before his next interview. No,
1740
01:52:22,504 --> 01:52:25,614
but there's, you know, this aspect,
1741
01:52:25,630 --> 01:52:29,429
this position is very, very common. In fact,
1742
01:52:29,798 --> 01:52:33,100
you know, friend of the show William Briggs, statistician to
1743
01:52:33,600 --> 01:52:37,668
the stars, has made similar arguments, not similar
1744
01:52:38,168 --> 01:52:41,642
to Hinton's, but similar to the ones I'm about to postulate here, which is
1745
01:52:41,658 --> 01:52:45,216
that if in the propaganda space you have the,
1746
01:52:45,360 --> 01:52:48,773
let's say, the global warming doomsayers, the ones who
1747
01:52:48,805 --> 01:52:52,875
propose different studies that supposedly measure
1748
01:52:53,052 --> 01:52:57,106
XYZ and claim that in 20 to 30 years this bad
1749
01:52:57,202 --> 01:53:00,137
thing is likely to happen in their models and things like that.
1750
01:53:00,747 --> 01:53:05,034
And then you have other potential people who could be labeled scientists pushing
1751
01:53:05,082 --> 01:53:09,337
things that might have a counter to that narrative, but you'll
1752
01:53:09,837 --> 01:53:13,463
be able to trace the money somewhere potentially to be some of that
1753
01:53:13,495 --> 01:53:15,727
research being funded by an oil company.
1754
01:53:17,043 --> 01:53:20,720
And so the political taint is that the oil companies are
1755
01:53:21,220 --> 01:53:24,541
funding this research, therefore the results from this
1756
01:53:24,590 --> 01:53:27,977
research cannot be trusted. And he makes the counter that of course,
1757
01:53:28,384 --> 01:53:32,245
the government never has interests in a particular narrative. So if the money comes
1758
01:53:32,261 --> 01:53:35,726
from the government funding somehow, it's definitely not tainting
1759
01:53:35,774 --> 01:53:39,868
the conclusion, if that's how you're going to trace the validity
1760
01:53:39,965 --> 01:53:43,686
of certain research, that the government has no interest in making sure
1761
01:53:44,186 --> 01:53:45,391
the conclusions match their premises.
1762
01:53:47,910 --> 01:53:51,581
Yeah. So it's the same situation here,
right? Okay,
1763
01:53:51,790 --> 01:53:55,493
so an AI safety research group publishes a piece of
1764
01:53:55,993 --> 01:53:59,605
study that says that AI is really dangerous, dangerous and that we need more groups
1765
01:53:59,669 --> 01:54:03,146
overseeing them. Well, that's a conflict of interest
1766
01:54:03,371 --> 01:54:07,585
baked into the research. Doesn't necessarily make what their conclusion— the research true
1767
01:54:07,617 --> 01:54:10,902
or false necessarily, but it's a conflict of interest. But we're supposed
1768
01:54:11,402 --> 01:54:15,470
to believe— Perhaps. Well, yes, but that's an analytical
1769
01:54:15,486 --> 01:54:19,059
situation, not just a proxy for truth, right? Because people
1770
01:54:19,075 --> 01:54:22,569
want to use the incentives as the proxy for the truth,
1771
01:54:22,601 --> 01:54:25,286
and that's just not how it works. Now, Alex, we're going to move forward here.
1772
01:54:26,767 --> 01:54:30,068
We are going to move forward because now it's data center time.
1773
01:54:30,486 --> 01:54:34,246
We, we were continuing our trend of the propaganda because now it's data center time.
1774
01:54:36,950 --> 01:54:40,484
Well, she wasn't a politician or even an activist, but one day
1775
01:54:40,533 --> 01:54:43,805
she just decided enough is enough. In 2024,
1776
01:54:43,935 --> 01:54:47,473
DC native Taylor Frazier McCollum learned a $5
1777
01:54:47,587 --> 01:54:50,827
billion AI data center was coming just down the
1778
01:54:50,859 --> 01:54:54,118
road in Prince George's County, Maryland. It was
1779
01:54:54,215 --> 01:54:57,555
going to be built on the site of the demolished Landover Mall,
1780
01:54:57,924 --> 01:55:02,147
right near her grandfather's home. She was concerned about skyrocketing
1781
01:55:02,339 --> 01:55:06,129
energy rates and the impact on the environment. So Fraser
1782
01:55:06,629 --> 01:55:09,854
McCollum got 25,000 signatures on a petition,
1783
01:55:10,432 --> 01:55:13,836
led protests in her area, and those protests
1784
01:55:13,997 --> 01:55:17,674
spread spread across the country. And she didn't just protest,
1785
01:55:17,834 --> 01:55:21,575
she won. The project, which was set to break ground this
1786
01:55:21,688 --> 01:55:25,589
year, has now been delayed for at least 2 years.
1787
01:55:26,103 --> 01:55:29,298
Taylor Frazier McCollum joins me now with more on her
1788
01:55:29,798 --> 01:55:33,104
fight and the next one from the site where the data
1789
01:55:33,604 --> 01:55:36,909
center was actually supposed to be built. Taylor, good to see you.
1790
01:55:38,306 --> 01:55:42,063
Nice to see you, too. Thank you. Oh, yeah, it's a pleasure.
1791
01:55:42,160 --> 01:55:46,238
And it's, you know, it really is amazing what one person
1792
01:55:46,640 --> 01:55:50,397
can do and what you have done. And I'm just—
1793
01:55:50,509 --> 01:55:53,881
I'm curious why you were so driven to
1794
01:55:54,381 --> 01:55:58,265
protest against this $5 billion project. I knew what you
1795
01:55:58,361 --> 01:56:02,150
were worried about, but this was a lot to take on.
1796
01:56:02,632 --> 01:56:06,485
So, I was mostly concerned Yeah. Because my
1797
01:56:06,985 --> 01:56:10,367
family is already facing skyrocketing energy costs here in Landover,
1798
01:56:10,800 --> 01:56:14,971
and I knew that the data center would use up to 820 megawatts
1799
01:56:15,471 --> 01:56:18,805
of electricity, which is more than enough to power all the homes in
1800
01:56:18,901 --> 01:56:22,751
Prince George's County combined. So it was like, that's going to increase
1801
01:56:23,251 --> 01:56:26,810
our electric rates. And we're on the PJM grid, so we're connected all the way
1802
01:56:26,874 --> 01:56:29,986
north as Ohio and as south as North Carolina.
1803
01:56:30,227 --> 01:56:33,387
So anytime a data center comes online in those areas as well,
1804
01:56:33,499 --> 01:56:39,685
it impacts us already. So it's like bringing that And,
1805
01:56:39,831 --> 01:56:42,986
you know, when these tech CEOs talk about AI,
1806
01:56:43,455 --> 01:56:47,532
they talk a lot about innovation and the promise of the future.
1807
01:56:47,920 --> 01:56:51,555
But you say communities like yours are being asked
1808
01:56:52,055 --> 01:56:55,883
to sacrifice so the rest of the country can have this AI
1809
01:56:55,900 --> 01:56:59,095
revolution. Tell me exactly what you are
1810
01:56:59,143 --> 01:57:02,431
sacrificing here. We would
1811
01:57:02,931 --> 01:57:06,349
be sacrificing clean air. Diesel generators are used to power
1812
01:57:06,849 --> 01:57:10,444
most data centers in the United States of America today. And we would also
1813
01:57:10,944 --> 01:57:14,426
be sacrificing our access to clean water. We only have 24
1814
01:57:14,602 --> 01:57:17,910
hours of reserve drinking water, or 24 to 48 hours of
1815
01:57:18,410 --> 01:57:22,342
reserve drinking water. And we recently had our water reserve filtration system
1816
01:57:22,406 --> 01:57:25,874
go down. Sorry, our water filtration system go down. One second. And we were told
1817
01:57:26,374 --> 01:57:30,270
that Yes, sir. So this woman is
1818
01:57:30,303 --> 01:57:33,942
saying that her price of electricity
1819
01:57:34,442 --> 01:57:37,921
will go up because presumably there's a fixed number of electrons and she's
1820
01:57:37,953 --> 01:57:41,499
gonna be bidding for them. And then in the next breath she's like,
1821
01:57:41,725 --> 01:57:45,043
no, they're gonna bring up new power and that's what I'm actually objecting to.
1822
01:57:45,294 --> 01:57:47,196
Is that— did that just happen?
1823
01:57:48,944 --> 01:57:52,415
Might have happened. Might have happened, sir. You might find a couple
1824
01:57:52,915 --> 01:57:57,694
contradictions or I'm just clarifying because I'm
1825
01:57:57,791 --> 01:57:58,344
bewildered.
1826
01:58:03,240 --> 01:58:06,570
We might have to boil water and things of that nature. So we already don't
1827
01:58:07,070
--> 01:58:10,999
have enough for the residents in Maryland. We're importing 40% of our electricity.
1828
01:58:11,435 --> 01:58:14,839
So a data center having access to the 60% that we
1829
01:58:15,339 --> 01:58:19,010
generate on our own is unfair to the residents. You know,
1830
01:58:19,042 --> 01:58:22,623
we've covered the fact that Prince George's County faces,
1831
01:58:22,880 --> 01:58:26,718
you know, a budget deficit right now and also declining tax revenue.
1832
01:58:27,151 --> 01:58:30,571
That's only going to get worse after the NFL's Washington
1833
01:58:30,667 --> 01:58:33,991
Commanders move from Maryland and back to D.C.
1834
01:58:34,441 --> 01:58:38,053
So what's your response to those people, Taylor, who say, hey, look,
1835
01:58:38,150 --> 01:58:41,730
this kind of investment, it creates jobs and it's a massive
1836
01:58:41,843 --> 01:58:43,256
boon for the local economy?
1837
01:58:45,287 --> 01:58:49,758
The local economy in what way? A data center is going to provide temporary
1838
01:58:49,791 --> 01:58:53,403
short-term jobs. We are tired of trickle-down economics
1839
01:58:53,903 --> 01:58:57,481
in our community. We have not seen the benefits from the Commander Stadium. We have
1840
01:58:57,981 --> 01:59:01,319
not seen the benefits from the MGM Casino, and we just don't see the
1841
01:59:01,351 --> 01:59:04,947
revenue that's going to come from the data center. They did a full qualified
1842
01:59:04,963 --> 01:59:08,239
data center task force report earlier, I mean, late last
1843
01:59:08,739 --> 01:59:12,076
year, and in that report they did not state a single dollar amount of what
1844
01:59:12,172 --> 01:59:16,060
revenue would be generated. Our state waives sales and use tax for data centers,
1845
01:59:16,560 --> 01:59:20,260
and so does our county. Here in Prince George's County, there would be no revenue
1846
01:59:20,760 --> 01:59:23,920
generated from those avenues. So I just want to know where all the money
1847
01:59:24,420 --> 01:59:27,840
is going to come from once construction is completed and they pay for the land
1848
01:59:28,340 --> 01:59:32,020
and they pay for the space. I think that the billionaire class, which includes
1849
01:59:32,520 --> 01:59:35,900
the Lerner family because they are billionaires, they're being greedy and they
1850
01:59:36,400 --> 01:59:39,460
just want to continue to make more money and profit off of the working class
1851
01:59:39,960 --> 01:59:42,860
like myself. And that's why they're trying to put this data center in our community,
1852
01:59:43,360 --> 01:59:46,413
because it was an easy target. I'll tell you what, Taylor. You have done your
1853
01:59:46,526 --> 01:59:50,644
homework. You know ABC News. We have this voter poll data,
1854
01:59:51,030 --> 01:59:55,021
and it shows that six in ten House voters said AI
1855
01:59:55,231 --> 01:59:58,277
will do more to hurt society than help it.
1856
01:59:58,697 --> 02:00:02,738
Why do you think the American
people are turning against artificial
1857
02:00:03,238 --> 02:00:06,869
intelligence? Oh, because of the leaders of
1858
02:00:06,901 --> 02:00:10,750
the tech bros. You know, you got Nvidia. Yeah. They're having a full scandal this
1859
02:00:11,250 --> 02:00:15,345
week. You know, Sam Altman is continually talking about how he has this dystopian
1860
02:00:15,442 --> 02:00:19,060
idea of the future. You know, Zuckerberg is contributing
1861
02:00:19,092 --> 02:00:23,604
to election scandals and all those type of things. So it's
1862
02:00:24,104 --> 02:00:26,917
dystopian what they want to see. I don't know if anyone has heard of dark
1863
02:00:27,417 --> 02:00:31,110
enlightenment, but a lot of the tech bros follow that train of thought. And I
1864
02:00:31,610 --> 02:00:35,770
think that it would be a travesty if we allowed
1865
02:00:36,270 --> 02:00:38,931
it to happen. You know, so I'm going to do everything in my power to
1866
02:00:38,947 --> 02:00:42,864
stop all data center construction in hyperscale form because it's not
1867
02:00:43,364 --> 02:00:47,648
necessary for our success in our future. Now, you say you're a hardworking blue-collar
1868
02:00:47,680 --> 02:00:51,324
woman. What kind of work do you do right now, and how have you been
1869
02:00:51,516 --> 02:00:53,346
able to take this on?
1870
02:00:55,289 --> 02:00:58,837
So thankfully, I'm a licensed cosmetologist, so I do have capability
1871
02:00:59,337 --> 02:01:02,689
to work for myself. I have clients. I previously was working as a wax specialist
1872
02:01:03,189 --> 02:01:06,173
when I first started this fight, and now I work at a local, um,
1873
02:01:07,284 --> 02:01:10,861
mixed-use salon, uh, or I guess not mixed-use, but like, uh,
1874
02:01:10,958 --> 02:01:13,575
it takes men and women, a full-service salon.
1875
02:01:14,009 --> 02:01:17,348
So I do that part-time, and then I get to take my clients in between.
1876
02:01:17,718 --> 02:01:21,186
So that's how I'm supporting myself now. That is amazing. Well,
1877
02:01:21,218 --> 02:01:25,216
you're sure making a beautiful difference in your community in
1878
02:01:25,328 --> 02:01:28,781
many other ways. Taylor Frazier McCollum, thank you so much for
1879
02:01:29,281 --> 02:01:30,996
your time. Thank you.
1880
02:01:33,310 --> 02:01:36,560
All right, Alex. Now, I don't know if you caught that,
1881
02:01:37,060 --> 02:01:41,040
but at the end of that interview where she— after she
1882
02:01:41,540 --> 02:01:45,100
said that the area was an easy target,
1883
02:01:45,600 --> 02:01:49,180
whatever that means, whatever it means to purchase land and want to build something.
1884
02:01:49,240 --> 02:01:51,380
They want to exploit the workers that aren't going to be working there. They want
1885
02:01:51,880 --> 02:01:53,440
to exploit the workers that aren't going to be having jobs.
1886
02:01:54,860 --> 02:01:58,360
But it's an easy target. I don't know what that means, but it's
1887
02:01:58,860 --> 02:02:01,420
an easy target. Okay, you mean that it's a good location?
1888
02:02:02,136 --> 02:02:06,696
But anyway, after she says that, she goes on a further
1889
02:02:07,306 --> 02:02:10,389
anti-data center campaign that
1890
02:02:10,437 --> 02:02:13,824
is more universal and global, that she doesn't want these anywhere.
1891
02:02:13,921 --> 02:02:17,549
So it's not just that this was a poor location
1892
02:02:17,838 --> 02:02:20,889
and that there were going to be deleterious effects that really concerned her that she
1893
02:02:21,389 --> 02:02:25,673
was actually standing up against. It is, as we see, an actual
1894
02:02:25,770 --> 02:02:29,575
cause in general. That she is now against.
1895
02:02:30,265 --> 02:02:33,782
Well, it's about the tech bros and whatever Nick Land
1896
02:02:33,814 --> 02:02:37,314
Zuckerberg is doing with the elections. I don't. How has Mark Zuckerberg managed
1897
02:02:37,411 --> 02:02:41,473
to be accused by both sides of meddling with the elections? That's actually quite
1898
02:02:41,505 --> 02:02:44,524
the accomplishment. I did not know that the Democrats were also mad at him.
1899
02:02:46,001 --> 02:02:49,742
Oh, oh, well, once once he once he was at the inauguration for Trump,
1900
02:02:49,790 --> 02:02:52,632
are you kidding me? They fucking want him dead. It's over. It's over.
1901
02:02:55,260 --> 02:02:58,771
Yeah. Now, now this is— this, Alex, is an article
1902
02:02:59,271 --> 02:03:01,425
from theroot.com. Are you familiar with theroot.com?
1903
02:03:04,026 --> 02:03:07,905
Um, no. It is an African-American
1904
02:03:08,405 --> 02:03:12,027
Black culture-centered magazine that is— that's what I thought, but then I was like,
1905
02:03:12,044 --> 02:03:15,444
am I confusing with a band? So I thought I would just punt. That's The
1906
02:03:15,944 --> 02:03:19,467
Roots. That is The Roots, and they used to be on Conan O'Brien.
1907
02:03:19,662 --> 02:03:23,145
They used to be Conan O'Brien's, uh, of backing bands. And they're also
1908
02:03:23,645 --> 02:03:27,247
African American, as I recall. Well, at least the drummer is the main drummer.
1909
02:03:27,311 --> 02:03:30,692
I don't know if they're an all African American band, but this is titled How
1910
02:03:30,836 --> 02:03:34,170
One Black Woman Stopped a $5 Billion Maryland Data Center.
1911
02:03:34,586 --> 02:03:37,935
After saving her neighborhood from— saving her neighborhood
1912
02:03:37,967 --> 02:03:41,669
from a $5 billion data center project, one Black woman is
1913
02:03:41,733 --> 02:03:45,659
leading the fight against the polluting server farms that have been disproportionately
1914
02:03:46,159 --> 02:03:50,699
affecting majority Black neighborhoods neighborhoods since the use of AI has grown increasingly popular.
1915
02:03:51,422 --> 02:03:54,795
Moving with her daughter to her grandmother's house in Landover,
1916
02:03:54,811 --> 02:03:58,506
Maryland, after being evicted from her home, Taylor Frazier
1917
02:03:59,006 --> 02:04:02,666
McCollum went from a new resident on the block to the community's biggest
1918
02:04:02,746 --> 02:04:06,489
advocate, halting the building of the data center that is set
1919
02:04:06,682 --> 02:04:10,954
to be developed on the former site of the Landover Mall between 2026
1920
02:04:11,454 --> 02:04:15,560
and 2027, NPR reported. McCollum first learned about the data center
1921
02:04:15,913 --> 02:04:19,360
through an article in the Washington Business Journal. In an interview with
1922
02:04:19,860 --> 02:04:23,530
the Prince George's Sierra Club, the activist said that driven by
1923
02:04:24,030 --> 02:04:28,244
her care for the environment and living near the proposed site, she immediately began researching
1924
02:04:28,357 --> 02:04:30,890
the center's potential impact on her family and the community,
1925
02:04:31,403 --> 02:04:34,771
and then started an online petition for folks to sign. Quote,
1926
02:04:34,787 --> 02:04:38,155
I'm very passionate about the environment, and it's something that has been instilled in me
1927
02:04:38,655 --> 02:04:41,699
since elementary school, she said. Ah, the propaganda starts there.
1928
02:04:42,212 --> 02:04:44,826
My fam— my family's from Landover.
1929
02:04:45,275 --> 02:04:48,273
We live less than a mile away where the data center is proposed to be,
1930
02:04:48,386 --> 02:04:51,833
and it would greatly impact us. So I was like, I'm gonna do some
1931
02:04:52,333 --> 02:04:55,137
research. I'm gonna read online as much as I can and talk to people in
1932
02:04:55,637 --> 02:04:58,921
the community and find out if they knew about this proposal. Between June and August
1933
02:04:59,421 --> 02:05:03,139
of 2025, the petition received a steady stream of signatures, with 180 people
1934
02:05:03,639 --> 02:05:07,405
signing it. The Prince George's Sierra Club reported. But after
1935
02:05:07,905 --> 02:05:11,833
she made an Instagram reel in collaboration with the Party for Socialism
1936
02:05:11,881 --> 02:05:15,250
and Liberation, that number skyrocketed
1937
02:05:15,750 --> 02:05:19,180
to over 20,000. Soon the project was blocked and
1938
02:05:19,260 --> 02:05:22,389
a moratorium was put in place. However,
1939
02:05:22,645 --> 02:05:26,383
McCollum's fight is far from over. On her Facebook page, the activist
1940
02:05:26,480 --> 02:05:30,619
often posts videos from local conference engagements in Maryland, such as People
1941
02:05:30,811 --> 02:05:34,084
Over Data Center Summit in Baltimore. Where she called out local
1942
02:05:34,100 --> 02:05:37,213
politicians for failing to keep their word on banning data centers.
1943
02:05:37,631 --> 02:05:40,824
McCollum is also taking the fight nationwide with protests in
1944
02:05:40,856 --> 02:05:44,515
Memphis and Atlanta, NPR reported. Speaking with the publication, she stated,
1945
02:05:44,787 --> 02:05:47,612
quote, I feel like so many other people are depending on this fight to be
1946
02:05:48,112 -->
02:05:51,832
successful, not just for myself, but even across the country to set precedent
1947
02:05:52,332 --> 02:05:56,694
for other people that are fighting against hyperscale data centers. With over 5,000 followers
1948
02:05:56,823 --> 02:05:59,855
on the No Landover Data Instagram page,
1949
02:06:00,112 --> 02:06:04,493
which shares updates and local national rallies against data centers, McCollum's fight
1950
02:06:04,993 --> 02:06:09,095
fight is definitely gaining momentum. Now, of course, Alex, I added emphasis to a particular
1951
02:06:09,595 --> 02:06:12,719
callout in that article because I think there is a little bit of
1952
02:06:12,784 --> 02:06:16,087
a lead that gets buried there. And this next,
1953
02:06:16,216 --> 02:06:19,439
uh, this next little piece explores that, uh,
1954
02:06:20,145 --> 02:06:22,310
just a little bit.
1955
02:06:23,368 --> 02:06:26,929
New Pew Research shows more than half of Americans feel
1956
02:06:26,977 --> 02:06:30,025
data centers are bad for the environment. That's up from
1957
02:06:30,105 --> 02:06:33,863
40% earlier this year. Tonight, our Josh Chapin
1958
02:06:34,363 --> 02:06:37,812
is reporting on efforts to keep these big facilities out of the DMV,
1959
02:06:38,214 --> 02:06:41,939
and he has new details on who was supposedly supporting
1960
02:06:41,987 --> 02:06:45,359
a grassroots effort to block a massive data center
1961
02:06:45,859 --> 02:06:48,940
project in Landover. Scott, there are a lot of folks here in
1962
02:06:48,972 --> 02:06:52,601
Prince George's County who really hope that this data center project gets
1963
02:06:53,101 --> 02:06:56,406
across the finish line. Estimates project
1964
02:06:56,906 --> 02:07:00,115
Officials project this could generate more than $50 million in tax revenue per year,
1965
02:07:00,131 --> 02:07:04,033
which could really help the budget shortfall. But there's fierce opposition
1966
02:07:04,129 --> 02:07:08,063
to that, and there are concerns that some of that opposition
1967
02:07:08,563 --> 02:07:11,724
may not be as grassroots as we think it is. I started a petition on
1968
02:07:11,756 --> 02:07:15,000
change.org to help stop the data center from being built.
1969
02:07:15,064 --> 02:07:18,693
Taylor Frazier McCollum is a local mom and community advocate here
1970
02:07:19,193 --> 02:07:22,910
in Landover. She said this video being shared by the far-left activist
1971
02:07:23,410 --> 02:07:26,650
group the Party for Socialism and Liberation was instrumental
1972
02:07:27,150 --> 02:07:30,850
in spreading her anti-data center message. It got her petition against the
1973
02:07:31,350 --> 02:07:35,390
Landover Data Center project enormous publicity and more than 25,000
1974
02:07:35,890 --> 02:07:39,350
signatures. Without that reel, my petition might not be sitting at over
1975
02:07:39,960 --> 02:07:43,770
200,000 signatures like it was on that. So I think that it's just that it's
1976
02:07:44,270 --> 02:07:47,110
been helpful in
spreading the message and getting the word out no matter what.
1977
02:07:47,200 --> 02:07:50,130
But today, 7 News asked her about the connection between the Party for Socialism and
1978
02:07:50,630 --> 02:07:54,596
Liberation and the Chinese Communist Party. Several U.S.
1979
02:07:54,660 --> 02:07:58,671
congressional committees and investigators have alleged the PSL received
1980
02:07:59,171 --> 02:08:02,649
significant funding from an American millionaire with close ties to Beijing,
1981
02:08:03,098 --> 02:08:07,157
Neville Roy Singham. To my knowledge, nothing has been donated
1982
02:08:07,657 --> 02:08:11,649
or transferred, or, you know, and I directly have received no funds from
1983
02:08:12,149 --> 02:08:15,435
any outside organizations that would be connected to China.
1984
02:08:15,676 --> 02:08:17,745
So I just think that it's like a misnomer.
1985
02:08:21,013 --> 02:08:24,072
The Chinese have a vested interest in slowing down
1986
02:08:24,104 --> 02:08:27,636
American AI. In China, nobody is pessimistic
1987
02:08:28,136 --> 02:08:31,249
about AI the same way they are here. There's not the same kind of data
1988
02:08:31,749 --> 02:08:34,990
center opposition. Jonas Du is a staff writer for Pirate Wires,
1989
02:08:35,343 --> 02:08:38,506
a leading tech publication out of Silicon Valley. He was the first
1990
02:08:38,539 --> 02:08:41,798
to report the connection between Chinese interests, the PSL,
1991
02:08:41,991 --> 02:08:45,455
and the effort to torpedo the over data center. Anybody looking
1992
02:08:45,955 --> 02:08:50,038
at this campaign should think twice before either financially supporting it
1993
02:08:50,086 --> 02:08:53,467
or thinking that it is a homegrown campaign. As far as China's
1994
02:08:53,967 --> 02:08:57,810
interest in slowing down American AI development, think about this like
1995
02:08:57,858 --> 02:09:01,191
the space race back in the 1960s. In the height of the Cold War,
1996
02:09:01,656 --> 02:09:05,358
the US and the Soviet Union were in a high-stakes battle to figure out who
1997
02:09:05,470 --> 02:09:09,285
could go to the moon first. Back then, President Kennedy pulled out all
1998
02:09:09,413 --> 02:09:12,912
the stops. And by 1969, Neil Armstrong was standing
1999
02:09:13,168 --> 02:09:16,538
on the lunar surface. But it took everything, all the
2000
02:09:17,038 --> 02:09:21,353
ingenuity to muster that and win that fight. In this artificial
2001
02:09:21,853 --> 02:09:25,798
intelligence race, it is pretty clear China will stop at nothing to develop
2002
02:09:26,298 --> 02:09:29,682
the best and the fastest AI possible. And fostering opposition
2003
02:09:29,714 --> 02:09:33,020
to data centers here in the US could provide all the
2004
02:09:33,520 --> 02:09:36,428
advantage that they need. We also reached out to We reached out to the organization,
2005
02:09:36,445 --> 02:09:40,033
the Party for Socialism and Liberation, for comment on this story.
2006
02:09:40,098 --> 02:09:43,460
So far, they've not gotten back to us. In Landover, I'm Josh
2007
02:09:43,960 --> 02:09:47,680
Chapin, 7 News. We do begin tonight— Sorry
2008
02:09:48,180 --> 02:09:52,175
about that. All right, Alex, now connecting
2009
02:09:52,223 --> 02:09:56,543
it to China, whether China has moneyed involvement, not necessarily
2010
02:09:56,752 --> 02:10:00,247
super important. Obviously, they're going
2011
02:10:00,263 --> 02:10:03,372
to do it if they can. It's not a big controversy to me.
2012
02:10:04,072 --> 02:10:07,116
that I expect it, you know, that they're going to do it. We're going to
2013
02:10:07,132 --> 02:10:11,228
have operations of, you know, if we can— The US really does it abroad.
2014
02:10:11,919 --> 02:10:15,293
Yes. Yes. Yes. So, yes. However,
2015
02:10:15,389 --> 02:10:18,682
of course, that is a chain back to
2016
02:10:19,182 --> 02:10:22,778
my original sort of startup where we're talking about the fact
2017
02:10:23,278 --> 02:10:26,778
that these propaganda networks do exist to just sort of jump
2018
02:10:27,278 --> 02:10:31,100
onto these activities. And of course, one of them, one of these satellites being social
2019
02:10:31,656 --> 02:10:34,899
Socialist Liberation Party. Of course, of course,
2020
02:10:35,429 --> 02:10:38,691
that's going to be part of the network that's going to push it. Because the
2021
02:10:38,788 --> 02:10:41,925
article from The Root said that she only had 180 signatures
2022
02:10:41,958 --> 02:10:46,137
before the Socialist Party jumped on her. So they found their spokeswoman,
2023
02:10:48,161 --> 02:10:51,669
they found their friendly front face to go with the message that
2024
02:10:52,169 --> 02:10:55,517
they already wanted to have. So that's just chaining that all together
2025
02:10:56,017 --> 02:10:59,040
there. Well, you'll be happy to know she has joined the PSL
2026
02:10:59,540 --> 02:11:02,643
as a member. Oh, okay. So, um,
2027
02:11:02,787 --> 02:11:05,432
I was trying to figure out if she was a member before or after.
2028
02:11:06,458 --> 02:11:09,472
Um, I, I don't have the exact dates,
2029
02:11:10,466 --> 02:11:14,088
um, but she has recently joined
2030
02:11:14,810 --> 02:11:17,855
the party. So, okay, in case there was any
2031
02:11:18,721 --> 02:11:21,462
question. Liberation is right around the corner.
2032
02:11:22,744 --> 02:11:26,351
Look, it's a revolutionary, uh, it's a revolutionary socialist
2033
02:11:26,851 --> 02:11:30,219
party. Um, that, uh,
2034
02:11:31,574 --> 02:11:35,350
whose published program calls for replacing the capitalist state, expropriating major
2035
02:11:35,366 --> 02:11:38,609
industries, and replacing the standing military and police with
2036
02:11:38,657 --> 02:11:42,306
the workers' defense councils. That puts its ambitions well
2037
02:11:42,806 --> 02:11:45,938
beyond progressive regulation on a larger— or a larger welfare state.
2038
02:11:47,308 --> 02:11:50,113
So at least they're clear about what socialism is.
Well, it sort of explains why
2039
02:11:50,194 --> 02:11:53,816
she made the jump to the, the broader anti-AI cause,
2040
02:11:53,864 --> 02:11:57,828
as opposed to just this very, very specific data center in a very, very specific
2041
02:11:57,892 --> 02:12:01,581
location, right? It's not just concerns over what it would do to
2042
02:12:01,597 --> 02:12:06,071
her community. She is now part of a cause that is against adoption.
2043
02:12:06,152 --> 02:12:09,210
They organized the data center workers against their masters. Oh,
2044
02:12:09,227 --> 02:12:11,754
wait. Well,
2045
02:12:13,074 --> 02:12:16,679
Alex, sir, I'm going to open this up to speakers
2046
02:12:16,792 --> 02:12:20,305
if anyone has any comments, because we're going to have a couple extra topics to
2047
02:12:20,805 --> 02:12:23,403
get into before the end of the show here. are going to be very meaty,
2048
02:12:23,515 --> 02:12:26,535
and I want to give some breathing room to anybody who's been listening. We've had
2049
02:12:26,551 --> 02:12:29,796
several people in the audience who I think have been listening since the beginning of
2050
02:12:30,296 --> 02:12:33,940
the show. Give them an opportunity to come on up and comment
2051
02:12:34,020 --> 02:12:37,795
or mention something that we haven't said yet or comment on anything that we've discussed.
2052
02:12:38,213 --> 02:12:41,136
So while we— We want to give power to the people,
2053
02:12:41,586 --> 02:12:45,007
basically. Well, yes, we want to liberate them from
2054
02:12:45,666 --> 02:12:48,059
their containers. Capitalist existence.
2055
02:12:49,713 --> 02:12:53,101
Well, we want to create a facsimile of democracy where,
2056
02:12:53,374 --> 02:12:56,344
you know, we bring the people up and let their voices be heard,
2057
02:12:57,548 --> 02:13:00,872
you know, so that they— we can— When the workers are in command,
2058
02:13:00,920 --> 02:13:03,906
nobody will be able to protest the data center.
2059
02:13:04,549 --> 02:13:07,728
Let's put it that way. Well, Alex, it does
2060
02:13:07,760 --> 02:13:10,810
not look like anybody wishes to volunteer to come
2061
02:13:11,310 --> 02:13:14,343
up and speak, so that's okay. It's not required, but we do enjoy it when
2062
02:13:14,843 --> 02:13:17,250
it happens. Even Donald J. Trump,
2063
02:13:17,701 --> 02:13:21,698
PhD, very stable genius, hasn't volunteered. He does, you normally volunteer,
2064
02:13:22,472 --> 02:13:24,873
but he has not yet. So we'll see if by the end of the— oh,
2065
02:13:24,889 --> 02:13:28,050
we have a request. Ha! Got him.
2066
02:13:29,567 --> 02:13:33,212
Donald J. Trump, PhD, very stable genius. Thank you for
2067
02:13:33,777 --> 02:13:35,609
speaking. How are you today?
2068
02:13:38,718 --> 02:13:42,249
Find that mute button, sir, and come on down. Ah, hello.
2069
02:13:43,396 --> 02:13:46,783
Sorry, I was a little bit behind. Middle of nowhere,
2070
02:13:46,978 --> 02:13:50,037
terrible reception. I should get Starlink. Oh
no.
2071
02:13:50,669 --> 02:13:53,859
No, I just had a question related to earlier when you were talking about the
2072
02:13:53,875 --> 02:13:57,146
history of the country and the Electoral College.
2073
02:13:57,690 --> 02:14:01,153
Would you also be on board with the old, the federal,
2074
02:14:01,411 --> 02:14:04,956
like kind of the founding fathers, how they were kind of on board with universal
2075
02:14:05,456 --> 02:14:06,325
suffrage for a limited group?
2076
02:14:09,234 --> 02:14:12,566
You know what, uh, give me more specifics about what you're
2077
02:14:13,066 --> 02:14:16,870
referring to there. What, uh, uh, which limitation exactly?
2078
02:14:18,718 --> 02:14:19,987
Well, you're— go ahead.
2079
02:14:22,638 --> 02:14:25,932
Well, the, the original one is private property. I don't think that works
2080
02:14:26,432 --> 02:14:29,225
like it used to, considering private property used to be considered like, you know,
2081
02:14:29,241 --> 02:14:31,716
that was sacred because it was kind of a new concept. So it was like
2082
02:14:32,216 --> 02:14:35,476
you owned a piece of the United States of America since you owned a
2083
02:14:35,874 --> 02:14:39,323
piece of the United States of America, you actually had a sacred duty to
2084
02:14:39,823 --> 02:14:42,420
take care of the country in a way. So, there was kind of a virtue
2085
02:14:43,178 --> 02:14:46,365
associated with it as far as you take care of the land, that means you
2086
02:14:46,577 --> 02:14:49,816
also own a chunk of the land, you therefore— I don't know if that would
2087
02:14:49,913 --> 02:14:53,424
work these days. I know Heinlein went with service equals a vote.
2088
02:14:53,863 --> 02:14:57,475
I mean, but they did have— There was
2089
02:14:57,975 --> 02:15:01,614
an older practice if this is what
2090
02:15:02,114 --> 02:15:05,618
you're referring to that that the voting populace
2091
02:15:06,118 --> 02:15:09,934
was primarily comprised of landowners and that those
2092
02:15:10,030 --> 02:15:13,541
who did not own land were not participating in the public voting.
2093
02:15:13,927 --> 02:15:19,128
Is that what you're referring to or— Yeah.
2094
02:15:19,305 --> 02:15:23,000
Like, you know, I'm just— like, I'm an Anti-Federalist, so I've read
2095
02:15:23,500 --> 02:15:26,433
a lot of the stuff you've probably read because you can quote Federalist stuff too.
2096
02:15:27,065 --> 02:15:30,465
It's just more along the line of like their original writings
2097
02:15:30,561 --> 02:15:33,705
were, because you're talking about people can't figure out
2098
02:15:34,205 --> 02:15:36,240
the difference between Dan Sullivan and Dan J. Sullivan.
2099
02:15:37,314 --> 02:15:41,068
Mm-hmm. And then, you have people
2100
02:15:41,568 --> 02:15:44,838
that just have insane ideas about how the world works. Does that get
2101
02:15:45,338 --> 02:15:47,870
ameliorated with something like Heinlein where service equals citizen,
2102
02:15:48,367 --> 02:15:51,527
service equals a vote? Or, the Founding Fathers where if you
2103
02:15:52,027 --> 02:15:55,538
have a piece of private property, it kind of shows stewardship and responsibility
2104
02:15:56,329 --> 02:15:59,619
Aristotle was kind of on board with the middle class being able to vote.
2105
02:16:00,358 --> 02:16:03,942
So, but I think the
2106
02:16:04,442 --> 02:16:08,023
question is a good one, but I think it is mistargeted in
2107
02:16:08,136 --> 02:16:11,799
some ways. It's not— when contemplating
2108
02:16:12,056 --> 02:16:15,463
who, quote, should be voting, you also have
2109
02:16:15,963 --> 02:16:19,689
to consider what the apparatus is that they
2110
02:16:20,189 --> 02:16:24,170
are voting inside of. of. And when you give that system
2111
02:16:25,567 --> 02:16:28,795
powers it shouldn't necessarily have, the question of who
2112
02:16:29,295 --> 02:16:33,114
should be deciding, making decisions or participating in making decisions
2113
02:16:33,178 --> 02:16:36,695
for a system that shouldn't have particular powers to begin with is
2114
02:16:36,807 --> 02:16:39,922
where the misguidance goes. So if we
2115
02:16:41,400 --> 02:16:45,125
imagine that the Federalist system is
2116
02:16:45,398 --> 02:16:48,674
closer to where it started, where there
2117
02:16:49,174 --> 02:16:52,777
were few powers that the federal government had and the rest were delegated
2118
02:16:53,277 --> 02:16:57,025
to the states and the people as by the 10th Amendment, then the
2119
02:16:57,525 --> 02:17:01,642
question of who's voting on certain federal levels really
2120
02:17:01,754 --> 02:17:05,201
boils down to, are we creating a pressure that gives
2121
02:17:05,701 --> 02:17:08,952
the federal government more ability to usurp powers that it
2122
02:17:09,452 --> 02:17:13,087
shouldn't have? And so if we break down who is performing
2123
02:17:13,135 --> 02:17:16,630
votes for what, we create a situation where
2124
02:17:16,662 --> 02:17:19,964
it begins to usurp and it begins to grow and begins
2125
02:17:20,028 --> 02:17:23,480
to become corrupt in in its very nature. So it's
2126
02:17:23,980 --> 02:17:27,414
not necessarily easy to answer a question about whether
2127
02:17:27,914 --> 02:17:31,492
or not relegating votes to landowners works
2128
02:17:31,992 --> 02:17:35,217
without specifying exactly which version of government.
2129
02:17:35,265 --> 02:17:38,364
Is it the local government? Is it the state government? Or is it the federal
2130
02:17:38,412 --> 02:17:42,025
government? Is it the House? Is it the Senate? Is it the presidency? Are we
2131
02:17:42,525 --> 02:17:46,007
committing the similar problems that I was referring to where we've
2132
02:17:46,039 --> 02:17:49,746
blurred what the Electoral College even was supposed to
2133
02:17:50,246 --> 02:17:54,351
be, or what I've mentioned several times, which is the way that we've— the progressives
2134
02:17:54,608 --> 02:17:57,962
ruined the way that we select senators,
2135
02:17:58,475 -->
02:18:01,604
meaning I don't personally believe senators were supposed to be
2136
02:18:01,749 --> 02:18:05,279
elected, quote, by popular vote of the citizens in the
2137
02:18:05,779 --> 02:18:09,450
state. So then asking me whether we should— whether it would make sense
2138
02:18:09,950 --> 02:18:13,269
to relegate it to landowners only is sort of a missed target.
2139
02:18:13,334 --> 02:18:17,856
So I would need you ask
2140
02:18:18,356 --> 02:18:22,106
more situationally which elections and
2141
02:18:22,619 --> 02:18:25,826
which participants. I'm the same way. I've thought long and hard about it, but I
2142
02:18:26,326 --> 02:18:30,171
was just thinking, based off your other comments, you don't have to give a specific,
2143
02:18:30,235 --> 02:18:34,244
but in the meta level, do you think there might be an argument
2144
02:18:34,744 --> 02:18:38,237
for that? Not even endorsing it, but argument for that. Certainly an argument
2145
02:18:38,737 --> 02:18:43,994
for it. The risk that I would be concerned about that would require discussion
2146
02:18:44,494 --> 02:18:48,761
is what you could consider a rent-seeking problem, and not land
2147
02:18:49,261 --> 02:18:52,709
rent, meaning just charging people rent for living on your land,
2148
02:18:52,773 --> 02:18:56,111
but a very specific rent-seeking type problem. Oh, I agree. No, I would lease off
2149
02:18:56,611 --> 02:19:00,572
1,000-square-foot areas in my area for a vote and just charge
2150
02:19:00,637 --> 02:19:03,926
people like a $400-a-year maintenance fee. Well,
2151
02:19:03,991 --> 02:19:07,104
I was more saying that if you purchase land and that gives
2152
02:19:07,136 --> 02:19:10,520
you entitlement to vote, and let's say Then the question comes,
2153
02:19:11,770 --> 02:19:15,936
is voting power apportioned to the amount of land space that you occupy,
2154
02:19:15,952 --> 02:19:18,324
meaning square footage or something along those lines,
2155
02:19:19,462 --> 02:19:22,715
or some proportionate of value, meaning a person
2156
02:19:22,747 --> 02:19:25,952
who owns a 1-acre piece of land that
2157
02:19:26,452 --> 02:19:29,910
is only worth $5,000 on the open market, do they get
2158
02:19:30,410 --> 02:19:34,509
the same proportion of votes as a person who owns 12 acres
2159
02:19:34,541 --> 02:19:38,555
that that's worth $5 million. Let's assume that you
2160
02:19:38,731 --> 02:19:42,938
even start to get into those nuances. The landowners
2161
02:19:42,970 --> 02:19:46,663
then become part of that situation where they are in control of the
2162
02:19:46,727 --> 02:19:50,564
laws. And so the non-landowners, the current non-landowners
2163
02:19:50,773 --> 02:19:54,514
have the ability to be kept out of acquiring land
2164
02:19:55,014 --> 02:19:58,352
even by the landowners sort of as a cabal.
2165
02:19:58,769 --> 02:20:00,954
And so there is a risk of that that I'm not sure— No. That's what
2166
02:20:01,454 --> 02:20:04,868
you would need to work through. Yeah.
Well, as I said,
2167
02:20:04,981 --> 02:20:08,041
I agree. I don't think it should be landowners because there was a certain kind
2168
02:20:08,541 --> 02:20:12,035
of metaphysical attachment attached to land ownership back in
2169
02:20:12,535 --> 02:20:16,173
the day. Just like you said, most people aren't loyal to their states. I'm probably
2170
02:20:16,673 --> 02:20:18,911
one of the few people I know who's more loyal to my state than probably
2171
02:20:19,411 --> 02:20:22,805
the federal government. All right. Well, that's— And I
2172
02:20:23,305 --> 02:20:27,126
kind of miss that. Hawaii is actually fascinatingly probably the state.
2173
02:20:27,270 --> 02:20:29,696
I know you've said before you live there,
2174
02:20:30,821 --> 02:20:33,312
that has an ethnicity of its own in a way,
2175
02:20:33,472 --> 02:20:37,155
right? Yeah, I don't live there anymore. I live back in my home state
2176
02:20:37,655 --> 02:20:41,053
now. That's why I've got the farm. But yeah, Hawaii is definitely one
2177
02:20:41,553 --> 02:20:44,870
of the states where they are fiercely loyal to Hawaii over the
2178
02:20:45,370 --> 02:20:48,625
United States. Yep. Yeah.
2179
02:20:48,690 --> 02:20:52,255
So my answer is, of course, convoluted and not direct.
2180
02:20:52,868 --> 02:20:56,722
It says basically it depends on what the election is who it
2181
02:20:57,222 --> 02:21:00,653
makes sense to grant decision-making power to. And that's ultimately
2182
02:21:00,685 --> 02:21:03,992
what votes are in some capacity, whether it is
2183
02:21:04,506 --> 02:21:08,347
delegation to a representative for an ongoing period of
2184
02:21:08,363 --> 02:21:11,883
time in a government structure, or when you're
2185
02:21:11,899 --> 02:21:15,678
doing a ballot proposition and you're voting directly for
2186
02:21:16,178 --> 02:21:19,312
a specific law, it boils down to what is the structure,
2187
02:21:19,441 --> 02:21:22,813
what powers do they And then that's where you start to
2188
02:21:23,313 --> 02:21:26,495
decide who is the correct apparatus or people to give the
2189
02:21:26,995 --> 02:21:31,017
decision-making capability to. Let me propose
2190
02:21:31,537 --> 02:21:35,499
that we give votes to whoever
2191
02:21:35,999 --> 02:21:39,157
has shareholder rights in the DAO. How's that?
2192
02:21:40,125 --> 02:21:45,094
Well, that's a 401 situation, man. We got— or
2193
02:21:46,416 --> 02:21:49,557
does the DAO have a specific Are you referring to the company
2194
02:21:50,057 --> 02:21:53,536
structure of the DAO, not companies on the DAO? Well, I'm half joking,
2195
02:21:53,568 --> 02:21:57,161
but like, if you know how a company works, like an
2196
02:21:57,661 --> 02:22:00,113
individual company, right? Like the shareholders vote their shares.
2197
02:22:01,012 --> 02:22:04,686
Yeah, right. But you don't just vote in general. You know, a person
2198
02:22:04,702 --> 02:22:07,782
with more shares than you literally has more of a vote. Sure, I'm sure.
2199
02:22:07,846 --> 02:22:10,926
But you, you know, you could do that for
2200
02:22:11,039 --> 02:22:13,477
like the country's corporate structure.
2201
02:22:14,300 --> 02:22:17,604
I'm sure the Socialist Party would approve of this.
2202
02:22:17,780 --> 02:22:21,056
How about if you're a federal bureaucrat,
2203
02:22:21,072 --> 02:22:24,990
state bureaucrat, or a member of an NGO, you temporarily
2204
02:22:25,086 --> 02:22:28,249
lose your vote? You don't wanna hear what I think
2205
02:22:28,749 --> 02:22:32,215
about that. All right, well, there's our
2206
02:22:32,408 --> 02:22:35,748
broad philosophical discussion
2207
02:22:35,764 --> 02:22:37,867
on the nature of voting. Words of restraint need not apply.
2208
02:22:39,529 --> 02:22:43,260
Now, uh, Alex, before, before we get to
2209
02:22:44,031 --> 02:22:47,738
the, the biggest story, I think that you— or what I would assume you consider
2210
02:22:48,238 --> 02:22:51,767
the biggest story of— we, we got to get to the other story that interrupted,
2211
02:22:51,815 --> 02:22:54,560
the thing that caught attention just at the tail end of the week here.
2212
02:22:55,683 --> 02:22:59,086
Yeah, right there. We do begin tonight here with the breaking
2213
02:22:59,134 --> 02:23:02,857
news, the stunning moment late today. For only the 3rd time since 9/11,
2214
02:23:03,227 --> 02:23:07,369
the FAA issuing a rare nationwide ground stop, halting all planes
2215
02:23:07,869 --> 02:23:11,624
from taking off across this country, citing extraordinary circumstances
2216
02:23:11,800 --> 02:23:15,333
and for reasons of safety. The images coming in tonight from across this
2217
02:23:15,833 --> 02:23:18,705
country— passenger jets on the ground at LaGuardia Airport in New York,
2218
02:23:19,042 --> 02:23:22,960
flights not taking off at Boston's Logan Airport. Similar images tonight at
2219
02:23:23,460 --> 02:23:26,797
the Dallas-Fort Worth Airport and across this nation. In the west,
2220
02:23:26,829 --> 02:23:29,864
planes sitting on the runways at LAX, passengers left
2221
02:23:30,364 --> 02:23:32,529
wondering when They'll be able to take off or when they'll be able to get
2222
02:23:32,593 --> 02:23:36,174
off these planes. The images tonight from FlightAware showing just
2223
02:23:36,206 --> 02:23:39,594
how many planes were already in the air over this country when
2224
02:23:39,610 --> 02:23:42,870
the FAA ordered that ground stop, and then a short time ago,
2225
02:23:42,886 --> 02:23:46,723
the map showing just how many of those flights have now landed safely. What we're
2226
02:23:47,223 --> 02:23:50,545
learning tonight about the planes still in the air tonight: the message the FAA
2227
02:23:50,673 --> 02:23:54,302
sent to pilots in the cockpit ready to take off, what they received.
2228
02:23:54,768 --> 02:23:58,974
The Homeland Security Secretary. Outside the White House a short time ago, trying to reassure
2229
02:23:59,087 --> 02:24:02,106
the American public. Our
reporter asking, "Can you rule
2230
02:24:02,154 --> 02:24:05,799
out a cyberattack?" We do have news coming in at this hour on all
2231
02:24:05,911 --> 02:24:09,620
of this. Team coverage tonight, beginning with ABC's Matt Rivers leading us off from
2232
02:24:10,120 --> 02:24:13,409
LaGuardia Airport in New York. Want to bring in our chief investigative
2233
02:24:13,474 --> 02:24:17,102
correspondent Aaron Katersky and Aaron. This all comes amid heightened concern
2234
02:24:17,151 --> 02:24:20,314
over AI, over the potential for cyberattacks, and after
2235
02:24:20,378 --> 02:24:23,650
a week of threats across. Europe that disrupted thousands of flights there.
2236
02:24:24,268 --> 02:24:27,488
Fighter jets had to accompany many of those passenger planes to get them down.
2237
02:24:28,230 --> 02:24:30,979
Obviously, here tonight we're not going to get ahead of ourselves, but people at home
2238
02:24:31,027 --> 02:24:34,957
are aware of all these stories, these headlines. They're aware of the environment
2239
02:24:35,457 --> 02:24:38,902
that we're living in tonight. The FAA believes that this is a telecommunications
2240
02:24:38,934 --> 02:24:42,701
issue and not something or someone who targeted the system in
2241
02:24:42,717 --> 02:24:44,318
any way, not something more nefarious.
2242
02:24:46,180 --> 02:24:49,039
Nothing nefarious. What's the latest? Nothing serious, we're told, David, at least preliminarily. And it
2243
02:24:49,103 --> 02:24:52,652
was a rash of hoax threats that disrupted nearly a week's
2244
02:24:53,152 --> 02:24:56,505
worth of air travel in Europe. Nothing like that is thought to have contributed to
2245
02:24:57,005 --> 02:25:00,744
the ground stop in this country. The FAA is certain it's a telecommunications
2246
02:25:00,825 --> 02:25:03,667
issue. They just haven't said what caused it.
2247
02:25:03,795 --> 02:25:07,392
David, I'm just off with a senior law enforcement official who tells
2248
02:25:07,424 --> 02:25:10,748
us this seems to be more of a technical problem right now,
2249
02:25:11,261 --> 02:25:14,533
not a cyberattack. But the FAA is still working to determine
2250
02:25:14,725 --> 02:25:17,850
why their systems failed. David, I think there'll be a lot of relief in this
2251
02:25:18,350 --> 02:25:22,033
country when it's confirmed that this is solely a technical issue, if that's
2252
02:25:22,322 --> 02:25:25,527
in fact what this is, though causing a lot of chaos and confusion.
2253
02:25:25,623 --> 02:25:28,524
Aaron, our thanks to you. Rachel Scott is standing by at the White House.
2254
02:25:28,588 --> 02:25:32,114
And Rachel, you asked the Homeland Security Secretary late this afternoon if
2255
02:25:32,614 --> 02:25:35,703
he could rule out a cyberattack. He said they're not ruling anything out. But of
2256
02:25:36,203 --> 02:25:39,886
course, now we've heard from the White House tonight. David, Secretary Mullen did not
2257
02:25:40,386 --> 02:25:43,059
rule out that this could be
a cyberattack. He said at this point he does
2258
02:25:43,140 --> 02:25:46,303
not believe that there is anything nefarious going on here, that there's
2259
02:25:46,803 --> 02:25:50,076
any threat to national security. But he said right now they are still very
2260
02:25:50,576 --> 02:25:54,716
much working through this process. He says that an old system
2261
02:25:54,877 --> 02:25:58,120
is to blame and that the system needs funding and that it needs to be
2262
02:25:58,249 --> 02:26:01,364
updated. So then I asked the Transportation Secretary, Sean Duffy,
2263
02:26:01,862 --> 02:26:05,426
how Americans can be reassured that it is currently safe to fly with
2264
02:26:05,474 --> 02:26:09,280
this old system and how long it is going to take for things to
2265
02:26:09,780 --> 02:26:12,317
get back up and running and get back to normal. He told me that he
2266
02:26:12,817 --> 02:26:16,157
did not anticipate that there would be an outage like this, and he did
2267
02:26:16,657 --> 02:26:20,025
acknowledge that it is going to take some time for flights to fully get back
2268
02:26:20,525 --> 02:26:21,076
up and running, David.
2269
02:26:24,875 --> 02:26:28,265
Now, Alex, before we continue, this was the second story
2270
02:26:28,685 --> 02:26:32,883
this week, the first one being the Microsoft story where the, the we,
2271
02:26:32,964 --> 02:26:34,983
we told you so, we warned you people,
2272
02:26:36,474 --> 02:26:40,454
warning, hoping this was a cyberattack. attack by AIs were
2273
02:26:40,954 --> 02:26:44,705
disappointed in the conclusion that it was just
2274
02:26:45,913 --> 02:26:49,262
some ill-advised Friday
2275
02:26:49,374 --> 02:26:52,401
afternoon upgrades. It's,
2276
02:26:52,594 --> 02:26:55,089
yeah, man, I mean, who does that? Anyway,
2277
02:26:58,908 --> 02:27:01,952
it is a very good observation you've made, Gator, because I think
2278
02:27:02,452 --> 02:27:05,817
frankly, this is our future. Like, there will be no big
2279
02:27:06,138 --> 02:27:09,847
nationwide event from now until, you know, the future,
2280
02:27:10,217 --> 02:27:14,488
unless something goes terribly wrong, where we— the first
2281
02:27:14,568 --> 02:27:17,956
initial reaction is not going to be, was this AI,
2282
02:27:18,325 --> 02:27:21,071
an AI swarm attack or whatever? Yeah. Like,
2283
02:27:22,018 --> 02:27:25,551
we, you know, at least in the collective mind,
2284
02:27:25,567 --> 02:27:28,810
there is now a presence, you know, an agent, call it what you
2285
02:27:28,858 --> 02:27:34,221
will. An agent
2286
02:27:34,655 --> 02:27:38,781
of chaos or, you know, whatever that is interfering with
2287
02:27:38,813 --> 02:27:41,511
things, you know, again, abstractly speaking.
2288
02:27:42,908 --> 02:27:46,023
And that becomes the default, you know, it's like, it's like, you know,
2289
02:27:46,039 --> 02:27:49,684
you hear a new math breakthrough, right? Before the summer,
2290
02:27:49,957 --> 02:27:52,558
you were like, oh, I wonder who
did it. After the summer,
2291
02:27:53,248 --> 02:28:00,662
you say, oh, I wonder which, which model did did it. It's just— Well,
2292
02:28:01,608 --> 02:28:04,961
first you go to see if Terry Towles is upset. Then you know.
2293
02:28:05,041 --> 02:28:08,555
He's upset about it. You go for the crash out. You stay for the math.
2294
02:28:09,132 --> 02:28:13,319
Yeah. Well, so, but the
2295
02:28:13,819 --> 02:28:17,025
thing is that this is going to be a situation where they're not going to
2296
02:28:17,525 --> 02:28:21,565
accumulate their misses, right? The first response
2297
02:28:21,645 --> 02:28:25,383
out of Yukowski was, what are the chances this wasn't involving
2298
02:28:25,883 --> 02:28:29,593
AI? And I mocked him in saying, you know, an oncologist suspects cancer.
2299
02:28:30,094 --> 02:28:33,796
Like, it's going to be, you know,
2300
02:28:33,829 --> 02:28:37,173
it's going to be a situation where it doesn't matter how many things
2301
02:28:37,609 --> 02:28:41,080
did not involve it, or how many, or as I like to
2302
02:28:41,580 --> 02:28:44,600
say, we can screw things up on ourselves just fine. We don't need help from
2303
02:28:45,100 --> 02:28:49,001
the AIs. Mm-hmm. You know, how many times
2304
02:28:49,307 --> 02:28:52,499
they get all excited and then are, you know, failing.
2305
02:28:53,119 --> 02:28:56,957
The first time that something major is due
2306
02:28:57,457 --> 02:29:01,180
to an AI involvement, they'll be like, see, we told you this was risky
2307
02:29:01,212 --> 02:29:03,556
and we should have never done it. And it doesn't matter. I mean, uh,
2308
02:29:03,845 --> 02:29:07,586
just think about how much hay they've made out of the Hugging
2309
02:29:07,635 --> 02:29:11,890
Face incident. Yeah. Which, you know, had they not
2310
02:29:12,390 --> 02:29:15,663
done it, nobody would ever have heard about it. It didn't
2311
02:29:16,163 --> 02:29:20,047
cause any damage. Nothing was compromised. No uptime was
2312
02:29:20,143 --> 02:29:23,708
reduced. It was literally a blip, the blippiest
2313
02:29:24,208 --> 02:29:28,381
of blips, right? Like, I said before, like any teenage hacker
2314
02:29:28,670 --> 02:29:33,023
that is, you know, moderately talented has one of these under his belt by
2315
02:29:33,523 --> 02:29:35,480
the time he, you know, he shaves for the first time. Right. I mean,
2316
02:29:35,560 --> 02:29:38,579
this is just not a big deal, but somehow
2317
02:29:38,708 --> 02:29:42,064
because a bunch of agents did it, like literally we keep
2318
02:29:42,564 --> 02:29:45,501
talking, it has become lore now. Like it doesn't even matter what happened. happened,
2319
02:29:46,239 --> 02:29:49,786
uh, there. Because, you know, the Hugging Face incident is like,
2320
02:29:50,027 --> 02:29:53,398
you know, something that my mom could tell
2321
02:29:53,898 --> 02:29:57,411
me about, right? So, you know, imagine if they had
2322
02:29:57,911 --> 02:30:01,488
some real thing to hang their heads on, their hats
2323
02:30:01,988 --> 02:30:05,645
on, uh, what would they do? But now, for now, it does turn
2324
02:30:05,678 --> 02:30:09,097
out to make a mole out of a— sorry, what is it,
2325
02:30:09,369 --> 02:30:12,548
a mountain out of a molehill? Yeah. So, uh, imagine if they had a mountain.
2326
02:30:15,323 --> 02:30:19,964
Yeah. Yeah. Now listen, um, the, the— before
2327
02:30:20,012 --> 02:30:23,246
there were details, this, this apparent— this occurred apparently, as we'll get into a little
2328
02:30:23,746 --> 02:30:26,976
bit, uh, here with the, with the actual follow-up press conference where they give
2329
02:30:27,476 --> 02:30:31,138
a little bit more details, this apparently was done due to upgrades. But the first
2330
02:30:31,638 --> 02:30:34,364
thing that I did when, when it went down is I did Google to see
2331
02:30:34,380 --> 02:30:38,005
if they were doing any upgrades or anything like that. And there was actually
2332
02:30:38,672 --> 02:30:41,944
a coincidental upgrade that I think is
2333
02:30:42,444 --> 02:30:47,154
interesting that is involving AI, but does not appear to be involved
2334
02:30:47,735 --> 02:30:50,897
in what actually happened. But we will learn about the upgrade they were
2335
02:30:51,397 --> 02:30:55,415
doing that did involve AI. All of these represent
2336
02:30:55,963 --> 02:30:59,594
airlines that US customers are traveling. We are inside the
2337
02:30:59,674 --> 02:31:03,880
room where the Federal Aviation Administration is using artificial intelligence
2338
02:31:04,058 --> 02:31:07,593
to make your next flight faster. This is SMART,
2339
02:31:07,609 --> 02:31:11,190
short for Strategic Management of Airspace Routes and Trajectories,
2340
02:31:11,543 --> 02:31:14,851
part of an $875 million Trump
2341
02:31:15,351 --> 02:31:19,089
administration contract aimed at modernizing how air traffic moves
2342
02:31:19,106 --> 02:31:22,510
across the country. Let's go to the big board. FAA Chief Brian
2343
02:31:22,622 --> 02:31:26,909
Bedford showed me how the system works. We're trying to get ahead of the possible
2344
02:31:27,423 --> 02:31:30,859
constraint and solve it before it becomes a problem for American travelers.
2345
02:31:31,036 --> 02:31:34,424
An example: a flight last week from Tampa to Dallas was going to
2346
02:31:34,924 --> 02:31:39,096
be routed around weather, but Smart found the flight could fly over most
2347
02:31:39,596 --> 02:31:42,612
of the storms and avoid the detour. So this was the old route,
2348
02:31:42,773 --> 02:31:46,466
and this was the new route, and that saved how much time? 42 minutes.
2349
02:31:46,691 --> 02:31:49,276
That's a lot of time. It's a lot of time and a lot of fuel.
2350
02:31:49,597 --> 02:31:53,772
The FAA workers in this room use Smart to analyze about 250
2351
02:31:53,836 --> 02:31:56,357
data streams, everything from taxi times to air.
2352
02:31:56,550 --> 02:31:59,935
Traffic controller staffing to alleviate delays. Is this
2353
02:32:00,435 --> 02:32:04,041
storm actually going to cause a problem in Houston? In September, the FAA announced
2354
02:32:04,541 --> 02:32:08,614
a trial run focused on flights near Washington, D.C. Now the agency
2355
02:32:09,114 --> 02:32:12,753
is announcing a nationwide rollout. Transportation Secretary Sean Duffy:
2356
02:32:12,929 --> 02:32:16,266
I think we're going to save a lot more time, better able to manage flights,
2357
02:32:16,715 --> 02:32:20,148
manage weather, and move people through the skies much more quickly.
2358
02:32:20,565 --> 02:32:24,401
But the move has not been without skepticism. Last month, more than
2359
02:32:24,562 --> 02:32:27,660
2 dozen members of Congress urged the FAA to stop
2360
02:32:27,789 --> 02:32:31,048
using the system, citing concerns about testing, training,
2361
02:32:31,305 --> 02:32:35,175
and oversight. Humans still control the system. A prompt on how you can
2362
02:32:35,675 --> 02:32:39,269
change a trajectory, how we can manage weather, is just that— a
2363
02:32:39,769 --> 02:32:43,396
prompt. Even still, airlines say the system is already saving
2364
02:32:43,508 --> 02:32:47,506
time for passengers. Southwest Airlines CEO Bob Jordan tells
2365
02:32:47,539 --> 02:32:51,071
me the early data shows broader promise to predict
2366
02:32:51,571 --> 02:32:55,021
potential conflicts earlier, work to solve them earlier, and the benefit is
2367
02:32:55,069 --> 02:32:58,682
potentially huge for travelers. The jury is still out
2368
02:32:59,182 --> 02:33:02,359
on this high-tech way to work around frustrating flight delays.
2369
02:33:02,953 --> 02:33:05,747
I just don't trust it. That's why I'm skeptical.
2370
02:33:06,485 --> 02:33:09,905
But if the people who make decisions are gung-ho, it's going to happen,
2371
02:33:10,018 --> 02:33:13,807
and we'll just deal with it as it comes. Pete Mundine, CNN, Washington.
2372
02:33:15,777 --> 02:33:19,935
So that system, Alex, it's an unfortunate time we're in because
2373
02:33:20,113 --> 02:33:23,241
AI is the label used for absolutely everything.
2374
02:33:25,140 --> 02:33:28,897
And that this system is not necessarily any
2375
02:33:29,397 --> 02:33:33,350
more AI than the Weather Channel might use. It is a— Right. Or the
2376
02:33:33,366 --> 02:33:37,168
way that you use Waze and Google Maps. It is a
2377
02:33:37,668 --> 02:33:41,467
computational problem-solving system
2378
02:33:41,499 --> 02:33:45,436
that is incorporating weather patterns and information about the
2379
02:33:45,936 --> 02:33:49,244
planes themselves and trying to come up with a decision, a recommendation
2380
02:33:49,292 --> 02:33:52,586
about a new flight path. People right now are
2381
02:33:53,086 --> 02:33:56,684
imagining anytime the word AI is used that it's an LLM doing something,
2382
02:33:56,845 --> 02:34:01,006
that it's an agent harness or something like that. And so this is— at
2383
02:34:01,506 --> 02:34:05,007
the moment, unless I learn something
else about this system I don't know, it doesn't
2384
02:34:05,024 --> 02:34:08,536
alarm me. It actually just seems like a, you know, as long as it can
2385
02:34:09,210 --> 02:34:13,299
produce good outputs and recommendations,
2386
02:34:13,364 --> 02:34:16,266
I don't particularly see anything wrong. But this was the initial thing that I found
2387
02:34:16,766 --> 02:34:20,564
going where the news reports were coinciding for this upgrade
2388
02:34:20,981 --> 02:34:25,343
of this system and this launch of this system the same moment that this
2389
02:34:25,567 --> 02:34:28,101
FAA downed a chaplain.
2390
02:34:29,240 --> 02:34:32,550
I might have retweeted this because, yeah, I I mean,
2391
02:34:33,050 --> 02:34:36,660
what's, you know, more normal thing that could possibly happen?
2392
02:34:37,110 --> 02:34:39,760
You know, big update gets rolled out, system goes down.
2393
02:34:40,570 --> 02:34:44,550
Like, yeah, okay, I wonder what happened. Like, bullet is
2394
02:34:45,050 --> 02:34:47,789
fired at a person, person dies, I wonder what happened in between those, you know,
2395
02:34:48,289 --> 02:34:51,490
those 2 things. Yeah, yeah, yeah. It was a possibility.
2396
02:34:51,500 --> 02:34:54,390
Sure, they could have had a heart attack, but what are the odds? Yeah,
2397
02:34:54,890 --> 02:34:58,230
it was the thing that they were doing. Well, if you work in a company
2398
02:34:58,730 --> 02:35:01,750
and the system goes down, you go, okay, who was actually scheduling changes, or what
2399
02:35:02,250 --> 02:35:05,468
were the scheduled changes? you know, updates this week.
2400
02:35:05,869 --> 02:35:09,041
And then you look at those first. It's not that it's, it's the right place
2401
02:35:09,089 --> 02:35:12,101
to look. It doesn't mean you're gonna find it there. And that was the thing
2402
02:35:12,133 --> 02:35:14,873
that first thing I found was this, but it turns out it was much more
2403
02:35:14,905 --> 02:35:18,799
mundane and it might actually just be some stupid router somewhere that redirected
2404
02:35:19,299 --> 02:35:23,590
traffic. No, it's a real problem though that, you know, AI is the,
2405
02:35:23,910 --> 02:35:27,387
the term for pretty much everything, right? Like from, right, I don't
2406
02:35:27,435 --> 02:35:30,976
know, clay memes or— Well, it's, yeah, it's the sales pitch.
2407
02:35:31,040 --> 02:35:33,991
It's the sales pitch. Because it's fancy,
2408
02:35:34,151 --> 02:35:38,772
right? Everyone wants to be unrolling AI, and in some
2409
02:35:38,884 --> 02:35:42,301
degree it is AI, but it is not the form that people are
2410
02:35:42,365 --> 02:35:46,087
now most commonly thinking about, which is the interactive
2411
02:35:47,323 --> 02:35:51,301
chatbots. When I want to differentiate,
2412
02:35:51,574 --> 02:35:54,622
in case people want to get the terminology straight, and of course they're not,
2413
02:35:54,702 --> 02:35:57,349
but just saying, in case they wanted it.
2414
02:35:58,738 --> 02:36:02,037
I tend to use the word, the term generative AI to mean,
2415
02:36:02,471 --> 02:36:06,100
you know, the combination of LLMs and agent systems and,
2416
02:36:06,646 --> 02:36:09,986
you know, diffusion models that, you know, the image generation, video generation,
2417
02:36:10,002 --> 02:36:13,727
like all of that stuff. Something that produces something based
2418
02:36:14,227 --> 02:36:17,131
off of a prompt that you give it, basically. Right, right, yes.
2419
02:36:17,565 --> 02:36:21,965
That stuff is, you know, a meaningful
2420
02:36:21,997 --> 02:36:25,192
category. I mean, you know, in general, I want a
2421
02:36:25,692 --> 02:36:28,886
lot more resolution. You know, when you say I use— Of course. I use an
2422
02:36:29,386 --> 02:36:32,837
LLM, like, what do you mean? Like Llama 2 or GPT-6
2423
02:36:33,337 --> 02:36:36,804
or something in between? Like, what are we talking about here? Yeah, yeah. Um,
2424
02:36:37,158 --> 02:36:40,916
but, but nevertheless, that is, that is a, a term that tells me something
2425
02:36:40,964 --> 02:36:44,128
about what, you know, if you had ChatGPT routing your flights,
2426
02:36:44,225 --> 02:36:47,148
let me put it this way, okay? If you had ChatGPT,
2427
02:36:47,276 --> 02:36:50,216
I don't even care if it's like Astra, like, don't do that.
2428
02:36:51,308 --> 02:36:54,984
But if you're having like Some algorithm that was developed in
2429
02:36:55,484 --> 02:36:59,926
the '70s that is called AI because I guess that's what AI
2430
02:36:59,958 --> 02:37:03,697
was to them back then. Yeah, of course. I mean, these are well
2431
02:37:03,810 --> 02:37:07,003
understood and basically like software these days.
2432
02:37:07,115 --> 02:37:10,325
They're not particularly novel
2433
02:37:10,825 --> 02:37:14,465
in any sort of way. This is a computational path-solving
2434
02:37:14,561 --> 02:37:17,995
problem, not the generative AI. And that's the distinction
2435
02:37:18,012 --> 02:37:21,881
there. Now, Alex, This was also happening
2436
02:37:22,320 --> 02:37:25,948
in an atmosphere where there was a lot of chatter coming,
2437
02:37:26,580 --> 02:37:30,119
some of it from governments seemingly in coordination,
2438
02:37:30,940 --> 02:37:36,136
warning people that they need to consider stockpiling and having the
2439
02:37:36,636 --> 02:37:39,120
ability to survive days of no power grid.
2440
02:37:40,656 --> 02:37:43,961
And the question is, what was the
2441
02:37:44,461 --> 02:37:47,760
messaging coming out of the the US government. And it turns out the Trump
2442
02:37:48,260 --> 02:37:52,170
administration is actually beating against that drumbeat as a specific,
2443
02:37:52,571 --> 02:37:56,212
hey, you need to be on alert type of warning. This FAA failure
2444
02:37:56,420 --> 02:38:00,109
makes our great country's critical infrastructure look weak.
2445
02:38:00,606 --> 02:38:04,358
Should people watching this show worry about
2446
02:38:04,858 --> 02:38:09,057
power or
water systems going down? And should they start stockpiling
2447
02:38:09,330 --> 02:38:12,874
food and water like the UK is What are you telling people there to do?
2448
02:38:14,968 --> 02:38:18,415
Absolutely not. The UK doesn't have the resiliency that we have. They don't
2449
02:38:18,915 --> 02:38:23,844
have the redundancy in their system as we have. We saw this catastrophic situation
2450
02:38:23,860 --> 02:38:26,834
happen Friday night. We had all of government, the whole of government,
2451
02:38:26,866 --> 02:38:29,985
President Trump's team, all came together. Within minutes,
2452
02:38:30,001 --> 02:38:32,729
we were in President Trump's office, at the Oval Office,
2453
02:38:32,745 --> 02:38:36,398
discussing this. He dispatched us in our separate directions.
2454
02:38:37,658 --> 02:38:41,761
And within 5 hours, we had the planes flying again. This was a catastrophic
2455
02:38:42,261 --> 02:38:45,005
outage that we were able to overcome. When you start talking about our economy and
2456
02:38:45,505 --> 02:38:49,292
our supply lines, you can't compare what they're saying in UK to what we have
2457
02:38:49,645 --> 02:38:53,402
in resiliency inside the United States. I get this call all the time,
2458
02:38:53,418 --> 02:38:56,630
Peter. What should we do? Because that's kind of running over the internet right
2459
02:38:57,130 --> 02:39:00,515
now, too, especially throughout social media. Should I be worried, the 90 days?
2460
02:39:00,708 --> 02:39:04,253
Absolutely not. I'm not stockpiling. And I tell my kids all the time,
2461
02:39:04,269 --> 02:39:08,045
don't worry unless you see me worry. We're in good shape here. The panic that
2462
02:39:08,143 --> 02:39:10,606
people want to draw from is unfounded.
2463
02:39:11,500 --> 02:39:14,999
President Trump's leadership has made America safer. We're stronger
2464
02:39:15,080 --> 02:39:16,846
and we're more resilient than we've ever been.
2465
02:39:18,700 --> 02:39:22,057
Now, Alex, there's 2 possibilities when it comes to being cynical
2466
02:39:22,154 --> 02:39:25,996
about what's happening in the government. One is that they are giving
2467
02:39:26,044 --> 02:39:29,706
you a panic signal because they are going to try to control
2468
02:39:29,754 --> 02:39:33,516
you and manipulate you. And the opposite end is that there's actually dangerous stuff
2469
02:39:34,016 --> 02:39:36,741
that they're not warning you about because they're incompetent and they want to cover their
2470
02:39:37,241 --> 02:39:40,480
own asses. And it looks like if we are in a cynical world,
2471
02:39:40,752 --> 02:39:44,796
the current version of the US government under the Trump administration
2472
02:39:44,988 --> 02:39:48,390
is the cynical version of, there's nothing going on, don't worry about it,
2473
02:39:48,438 --> 02:39:50,620
we have it all under control, there's no need to panic.
2474
02:39:51,743 --> 02:39:56,219
But the atmosphere— Under the competent leadership of an 80-year-old
2475
02:39:56,719 --> 02:40:00,185
who's threatening to destroy the civilian infrastructure of random countries,
2476
02:40:00,363 --> 02:40:01,336
some of which exist.
2477
02:40:05,042 --> 02:40:09,301
Putting your family's future in the
2478
02:40:09,801 --> 02:40:13,367
competent hands of the management capabilities of any 80-year-old,
2479
02:40:13,641 --> 02:40:16,726
never mind this one, is fucking insane.
2480
02:40:18,370 --> 02:40:20,010
I wasn't worried before, but I'm worried now.
2481
02:40:21,950 --> 02:40:25,623
So there are people who are starting to
2482
02:40:25,948 --> 02:40:29,187
wonder whether or not we were in the other cynical version,
2483
02:40:29,541 --> 02:40:33,507
which is that there's all these little air balloons of danger popping
2484
02:40:33,620 --> 02:40:37,169
up, which did involve what appeared to be legitimate
2485
02:40:37,669 --> 02:40:41,537
announcements that were circulating primarily from Europe warning
2486
02:40:41,586 --> 02:40:45,327
about the need to stockpile if the grid goes out underneath
2487
02:40:45,344 --> 02:40:49,037
the umbrella of the AI apocalypse doom stuff that
2488
02:40:49,537 --> 02:40:53,100
scrolls, that warns that the AI
2489
02:40:53,229 --> 02:40:56,837
could hack our grids and take us out, right? So there was
2490
02:40:57,337 --> 02:41:02,031
a concern sort of floating out that this
2491
02:41:02,127 --> 02:41:06,151
FAA thing was also attempting to turn up the heat
2492
02:41:06,535 --> 02:41:10,431
to get us all worried. And the stockpiling stuff is all communications trending
2493
02:41:10,463 --> 02:41:14,069
to make us live in fear because the government was about to
2494
02:41:14,569 --> 02:41:17,853
do something akin to the COVID
2495
02:41:18,558 --> 02:41:21,844
fiasco, where, oh, now there's a major event and we have to
2496
02:41:22,344 --> 02:41:24,521
take control. Yeah. So as far as I can tell, if we're in a cynical
2497
02:41:25,920 --> 02:41:29,161
cynical part, it is there's a huge problem the government is
2498
02:41:29,225 --> 02:41:32,614
failing to warn us about, possibly knows about, but they're incompetent
2499
02:41:33,114 --> 02:41:35,907
and they want to cover their asses and they want to try to make sure
2500
02:41:36,407 --> 02:41:39,859
nobody blames them for it. Right. But now, if you also want a real cynical
2501
02:41:39,907 --> 02:41:42,445
take, Alex, the first thing the FAA did,
2502
02:41:43,264 --> 02:41:46,702
or Sean Duffy did whenever he did, which we'll
2503
02:41:46,734 --> 02:41:50,236
see in a moment, whenever they told us what was
2504
02:41:50,268 --> 02:41:54,805
happening or their version of what was happening to cause the FAA ground Grounding
2505
02:41:54,853 --> 02:41:58,280
notice was demand that Congress pass bills to upgrade
2506
02:41:58,780 --> 02:42:03,202
the infrastructure in the
2507
02:42:03,702 --> 02:42:07,279
air system. So if you want a cynical take, the actual
2508
02:42:07,327 --> 02:42:10,760
super cynical take
that could be grounded in reality,
2509
02:42:10,776 --> 02:42:14,171
in my opinion, is that the grounding
2510
02:42:14,349 --> 02:42:18,070
action was taken not out of caution. It was actually
2511
02:42:18,570 --> 02:42:22,386
an overabundance of caution. as a ruse to get everybody focused on the
2512
02:42:22,418 --> 02:42:25,761
airline so that they could cynically push for more
2513
02:42:25,794 --> 02:42:28,686
money to upgrade the system and specific legislation.
2514
02:42:29,365 --> 02:42:31,920
There's your— there's my cynical take if you want it.
2515
02:42:33,731 --> 02:42:37,305
No. Well, I have something specific to add to this conversation because
2516
02:42:37,435 --> 02:42:40,359
I follow the Ukraine conflict,
2517
02:42:40,585 --> 02:42:43,929
as you probably remember. Yes. Yes. So Europe
2518
02:42:43,978 --> 02:42:46,665
is currently in the middle of a You know,
2519
02:42:47,051 --> 02:42:50,812
mass hysteria about Russia intervening in all sorts of things,
2520
02:42:50,844 --> 02:42:54,429
and they keep talking about the hybrid war and this and that, and you know
2521
02:42:54,590 --> 02:42:58,714
intelligence operations here and there, and the Russians are everywhere poisoning things
2522
02:42:59,214 --> 02:43:01,388
and you know landing drones and whatever.
2523
02:43:03,630 --> 02:43:06,939
Partly, I think people are concerned that this is happening so
2524
02:43:07,005 --> 02:43:08,358
that they can justify funding.
2525
02:43:10,786 --> 02:43:13,172
You know, a certain former actor. Um,
2526
02:43:14,167 --> 02:43:18,420
I mean, I mean Ukraine. Um, and you
2527
02:43:18,920 --> 02:43:22,577
know, as, as their economies themselves are tanking, right? So they, they have to
2528
02:43:23,077 --> 02:43:26,461
foment, foment in, in the people
2529
02:43:26,943 --> 02:43:30,265
some level of panic and stress and anxiety
2530
02:43:30,458 --> 02:43:33,940
and concern about, you know, those evil Russkies and what they're gonna do,
2531
02:43:34,695 --> 02:43:38,306
etc., etc. Um, so it's funny to me to see that That
2532
02:43:38,630 --> 02:43:42,572
transposed here in a completely different context because,
2533
02:43:43,072 --> 02:43:46,616
you know, Europe has its own problems
2534
02:43:46,713 --> 02:43:50,346
and they're they've completely underfunded their militaries for a
2535
02:43:50,846 --> 02:43:54,104
very long time, which I mean good. I don't think the military
2536
02:43:54,136 --> 02:43:58,033
should be getting super well funded, but that's a different story. But now
2537
02:43:58,082 --> 02:44:01,663
they feel like they need to divert, you know, from a dwindling
2538
02:44:01,939 --> 02:44:05,564
supply of resources into the military, both their. end of
2539
02:44:05,790 --> 02:44:09,146
another country, which, you know,
2540
02:44:09,517 --> 02:44:12,646
I don't know if you've seen that they're in the middle of like public unrest
2541
02:44:13,146 --> 02:44:16,492
and like extreme unpopularity of the politician class. So they are,
2542
02:44:17,095 --> 02:44:19,639
you know, cynically speaking, rolling out the psyops.
2543
02:44:21,370 --> 02:44:25,484
So, you know, invoking those in the US context is ludicrous.
2544
02:44:26,307 --> 02:44:30,163
It's a whole other story over there. What's happening, it's insane,
2545
02:44:30,195 --> 02:44:34,814
but nothing to do with, you know, the FAA or preparing
2546
02:44:34,910 --> 02:44:37,721
for the AI apocalypse or whatever. Yeah.
2547
02:44:38,122 --> 02:44:41,704
Yeah. Now, just because it's— we want to be
2548
02:44:41,977 --> 02:44:46,088
complete here, I am going to play a little bit of the
2549
02:44:46,588 --> 02:44:50,602
press conference that was given to give— that gave fleshed-out details, just because this
2550
02:44:50,810 --> 02:44:54,488
did have everyone's pretty much attention all day on
2551
02:44:54,585 --> 02:44:57,941
Friday. I've— those of you who are aviation reporters
2552
02:44:57,973 --> 02:45:01,715
have heard me talk about a decade old air traffic control system.
2553
02:45:02,454 --> 02:45:06,372
And it's not just old equipment. We have an incredibly old
2554
02:45:06,484 --> 02:45:09,695
telecommunications system. And so here's what we
2555
02:45:10,195 --> 02:45:14,560
believe happened yesterday. In at least one of our remote FAA locations,
2556
02:45:15,235 --> 02:45:19,442
a contractor was doing routine maintenance to switch telecommunication
2557
02:45:19,490 --> 02:45:23,199
lines from old copper wire, and they were bringing it over to
2558
02:45:23,699 --> 02:45:27,004
fiber. That contractor that does the work for us is L3Harris.
2559
02:45:27,920 --> 02:45:32,239
They're our legacy provider. They've been working for the FAA for over
2560
02:45:32,319 --> 02:45:35,546
20 years, and the network that we use at the
2561
02:45:36,046 --> 02:45:39,288
FAA to control the airspace, L3Harris actually
2562
02:45:39,464 --> 02:45:43,302
owns that network. Now, the work happening
2563
02:45:43,880 --> 02:45:47,187
was due to the age of the copper wire that we were
2564
02:45:47,687 --> 02:45:51,619
replacing. It was such an old wire that
2565
02:45:52,119 --> 02:45:55,746
it was being discontinued, meaning that line was
2566
02:45:55,778 --> 02:45:59,005
going to get shut down. So we had to switch it over from old
2567
02:45:59,053 --> 02:46:03,116
copper. We had to bring it over to fiber, going from TDM
2568
02:46:03,389 --> 02:46:06,664
to IP or from analog to digital. Now,
2569
02:46:06,680 --> 02:46:10,309
I want to be clear. This work has been happening in America
2570
02:46:10,630 --> 02:46:14,564
for decades. Virtually every single institution in America
2571
02:46:14,885 --> 02:46:18,145
has gone from copper to fiber. From analog to
2572
02:46:18,161 --> 02:46:21,436
digital, I just maybe the FAA is the last one to be doing
2573
02:46:21,452 --> 02:46:24,744
this work because the investments haven't been made. But this
2574
02:46:25,244 --> 02:46:28,742
is not
complicated; it's it's pretty straightforward work that we
2575
02:46:28,838 --> 02:46:31,793
were doing yesterday at again a very remote location.
2576
02:46:32,708 --> 02:46:37,027
So what we're working to determine was whether we
2577
02:46:37,527 --> 02:46:40,656
had incompetence, meaning was it user error
2578
02:46:41,156 --> 02:46:45,196
or installer error from? L3Harris, or did
2579
02:46:45,696 --> 02:46:48,908
they install a faulty router? We haven't determined what the fault was.
2580
02:46:48,973 --> 02:46:52,330
Was it the router, or were they not doing
2581
02:46:52,446 --> 02:46:55,893
great work for us at L3Harris? Either through
2582
02:46:55,925 --> 02:46:59,058
contractor error or a faulty piece of equipment,
2583
02:46:59,765 --> 02:47:03,878
that router that was being replaced as we switched from
2584
02:47:04,505 --> 02:47:07,808
copper to fiber, it started flooding other routers in
2585
02:47:08,308 --> 02:47:10,730
our air traffic control system. So specifically,
2586
02:47:11,485 --> 02:47:15,338
we had over 1,000 connection attempts every 4 to
2587
02:47:15,838 --> 02:47:19,593
5 seconds. And what that did was gave
2588
02:47:20,093 --> 02:47:24,619
us issues with our communication with airplanes,
2589
02:47:24,731 --> 02:47:28,424
and it also caused some disruption with our radar
2590
02:47:28,924 --> 02:47:32,567
feeds. And when we see that, your safety is
2591
02:47:32,599 --> 02:47:35,949
paramount. We shut down the airspace, and we do
2592
02:47:35,965 --> 02:47:39,930
it safely. We get you on the ground, and then we work
2593
02:47:40,430 --> 02:47:44,592
on isolating the problem and then opening
2594
02:47:44,625 --> 02:47:47,205
the airspace back up, which is what we did yesterday.
2595
02:47:49,984 --> 02:47:53,003
And as we get more information, we'll give this to you.
2596
02:47:53,939 --> 02:47:57,217
But I think it's pretty clear a lot of questions came in yesterday. Was was
2597
02:47:57,313 --> 02:48:01,051
this a cyber attack? Did some foreign nation
2598
02:48:01,967 --> 02:48:06,366
attack this network? And the answer to that is
2599
02:48:06,864 --> 02:48:10,011
we don't believe so. Once we were able to isolate the work that
2600
02:48:10,511 --> 02:48:14,170
was being done at this remote location, all of our systems came back online.
2601
02:48:14,523 --> 02:48:17,943
So we don't believe this was cyber at
2602
02:48:18,443 --> 02:48:23,579
all. This was an FAA system that didn't meet— didn't
2603
02:48:23,659 --> 02:48:27,345
meet the moment. Our contractor didn't meet the moment, or the
2604
02:48:27,845 --> 02:48:30,136
equipment didn't meet the moment and caused the outages.
2605
02:48:31,970 --> 02:48:34,875
I've been talking about this for almost 2 years.
2606
02:48:35,133 --> 02:48:38,845
Mr. Secretary, appreciating the need for upgrades from copper
2607
02:48:39,345 --> 02:48:43,225
to fiber, nevertheless, this upgrade was made at 3 o'clock on a Friday afternoon.
2608
02:48:43,549 --> 02:48:46,851
Why? Why not, say, at midnight on a Saturday when there
2609
02:48:47,351 --> 02:48:50,623
are fewer planes lined up to take off? Yes. If I could ask. Are there
2610
02:48:51,123 --> 02:48:53,966
any changes in these planned upgrades over the course of the next couple of weeks?
2611
02:48:54,096 --> 02:48:57,440
Particularly ahead of Thanksgiving. So I first want to— I'll answer— that's a great question.
2612
02:48:57,488 --> 02:48:58,166
Thank you for asking it.
2613
02:49:00,705 --> 02:49:05,117
Safety is paramount. And if we don't make these upgrades,
2614
02:49:06,040 --> 02:49:09,260
we have a less safe system. We can't continue on
2615
02:49:09,453 --> 02:49:13,109
technology of the '80s. I mean, we spend over
2616
02:49:13,157 --> 02:49:17,095
$200 million a year in obsolescence We have to pay to
2617
02:49:17,112 --> 02:49:20,724
use these old copper wires. We have to get off of
2618
02:49:21,224 --> 02:49:24,241
them. They're less reliable. So we have to make the transition.
2619
02:49:25,316 --> 02:49:28,688
And we're making transition not because we want to spend money.
2620
02:49:28,753 --> 02:49:31,659
We're making the transition because it's safer and it's more resilient.
2621
02:49:32,462 --> 02:49:36,042
But because of the nature of what this switchover was,
2622
02:49:36,171 --> 02:49:38,611
it was a remote location.
2623
02:49:40,089 --> 02:49:43,281
We'll do those During the day. Now, if we're going to be at Newark,
2624
02:49:43,297 --> 02:49:47,211
if we're going to be in one of our big facilities, those switchovers
2625
02:49:47,711 --> 02:49:50,756
happen at night because they can have a bigger impact.
2626
02:49:51,654 --> 02:49:54,782
So there you go, Alex. That's the broad— I have more, but we don't need
2627
02:49:55,282 --> 02:49:57,877
to get into it. It doesn't answer any more questions, I don't believe, necessarily,
2628
02:49:57,909 --> 02:50:01,181
unless we really need to, you feel we need to listen to more.
2629
02:50:01,903 --> 02:50:05,849
But my most cynical take that I have that I believe
2630
02:50:06,349 --> 02:50:09,377
is enough grounded in at least the narrative, if not reality,
2631
02:50:09,510 --> 02:50:13,016
Yeah. Is that whatever communications issues
2632
02:50:13,032 --> 02:50:16,719
they were having were not severe enough to issue the ground warning, but they issued
2633
02:50:17,219 --> 02:50:19,492
it anyway in order to draw attention so that he could make a pitch to
2634
02:50:20,101 --> 02:50:23,628
have funding pushed through that he specifically wants. That would be
2635
02:50:24,128 --> 02:50:27,635
my most cynical take given the things that we've
2636
02:50:28,135 --> 02:50:31,835
been told are what went down. But it doesn't— the funniest thing,
2637
02:50:31,947 --> 02:50:36,099
Alex, he said that they had 1,000 connections
2638
02:50:36,131 --> 02:50:38,810
every every couple, like 5 seconds or something like that.
2639
02:50:39,243 --> 02:50:42,356
Aaron Rupar, the
brilliant mind that is
2640
02:50:42,532 --> 02:50:46,206
Aaron Rupar, who used to work for Vox Media, said,
2641
02:50:46,398 --> 02:50:49,366
that certainly sounds like hacking. And now,
2642
02:50:49,447 --> 02:50:53,425
1,000 connections— There you go. Every 5 seconds does not particularly
2643
02:50:53,634 --> 02:50:57,372
seem like a lot of connections to cause a problem, but it depends on what
2644
02:50:57,872 --> 02:51:00,628
type of connection it was. I mean, there are certain systems that are meant to
2645
02:51:00,660 --> 02:51:04,590
handle millions of connections every few And so it's
2646
02:51:05,090 --> 02:51:08,330
like, you know, 1,000 seems like horrible, but depending on the type
2647
02:51:08,830 --> 02:51:11,990
of connections and the network and what traffic was supposed to go through there,
2648
02:51:12,490 --> 02:51:15,570
1,000 actually could be quite a lot, you know. Yeah, no,
2649
02:51:16,070 --> 02:51:19,290
I mean, any, like a backend engineer in general does
2650
02:51:19,790 --> 02:51:23,550
not, you know, again, as you said, there are situations where 1,000
2651
02:51:24,050 --> 02:51:27,930
could be a lot, but in general, it's not considered anything too right.
2652
02:51:28,030 --> 02:51:31,630
That sounds like hacking. It's like, okay. I mean,
2653
02:51:32,130 --> 02:51:35,630
sorry, have you tried to count 1,000 in your fingers? It's a lot.
2654
02:51:35,730 --> 02:51:39,910
It is a lot. It is a lot. Very complicated. So my
2655
02:51:40,410 --> 02:51:43,610
most cynical take is that, you know, giving me a quantification of 1,000, it's like,
2656
02:51:44,110 --> 02:51:46,730
you should be able to handle that. I don't even care if it's Commodore 64
2657
02:51:47,230 --> 02:51:50,090
you're running them through. I mean, frankly, you just put up a free,
2658
02:51:50,590 --> 02:51:54,830
like, Squarespace website and it can probably deal with 1,000 concurrents.
2659
02:51:55,330 --> 02:51:58,844
Like, it's not Yeah, yeah, yeah. So that's,
2660
02:51:58,860 --> 02:52:02,274
that's the, that's my cynical take. Do you have any higher cynical,
2661
02:52:02,841 --> 02:52:06,068
more cynicism that you could possibly contribute? No,
2662
02:52:06,374 --> 02:52:10,255
frankly, that's, that's the pro—
2663
02:52:10,384 --> 02:52:13,749
this is, this is all of
2664
02:52:13,781 --> 02:52:17,704
the problem, right? Like asking the government to do stuff like this is,
2665
02:52:18,060 --> 02:52:21,628
it's, it's like, you know, I am impressed that they can do it in the
2666
02:52:22,128 --> 02:52:26,824
same way I'm impressed when like a dog can walk and it's hanging feat.
2667
02:52:27,310 --> 02:52:30,276
You know, it is, it is remarkable that they can do it at all.
2668
02:52:31,135 --> 02:52:33,534
Like, and I, and I, and I don't expect it to be done elegantly.
2669
02:52:37,601 --> 02:52:40,832
So, you know, I mean,
2670
02:52:40,864 --> 02:52:44,080
I'm, I'm, I'm, I'm impressed. You know,
2671
02:52:44,256 --> 02:52:47,465
things like this don't happen more often. That, that's all that
2672
02:52:47,514 --> 02:52:51,574
my, my, you know, my impressed budget is spent there. I'm impressed
2673
02:52:52,074 --> 02:52:55,725
with that, no matter it's government or not. I've actually— though there have
2674
02:52:56,225 --> 02:52:59,925
been people who were recently graduating from college and were afraid of
2675
02:52:59,973 --> 02:53:03,820
the job market, and I advised them that once they actually get
2676
02:53:04,320 --> 02:53:08,147
into the job market, their perspective will change because the initial perception is
2677
02:53:08,227 --> 02:53:11,930
that everyone out there is so advanced
2678
02:53:12,154 --> 02:53:15,344
and good at their job that it's intimidating,
2679
02:53:15,392 --> 02:53:17,748
and since you've never had a job, you're going to be incompetent at it,
2680
02:53:17,844 --> 02:53:20,347
and everyone's going to know You're a fraud or whatever. And I said, no,
2681
02:53:20,363 --> 02:53:23,924
listen, it's actually surprising that things work when you,
2682
02:53:24,117 --> 02:53:27,550
when you get into a company and you see the amount of normal incompetence that
2683
02:53:28,050 --> 02:53:31,176
surrounds you, that somehow everything still functions in society.
2684
02:53:31,256 --> 02:53:34,224
It's going to be very impressive that planes just aren't falling out of the air
2685
02:53:34,724 --> 02:53:38,155
all the time. I have a challenge for you. Oh boy. All right.
2686
02:53:38,877 --> 02:53:42,262
All right. Okay. First of all, make a short list of like around
2687
02:53:42,762 --> 02:53:44,139
about 5 of your nearest coworkers.
2688
02:53:47,899 --> 02:53:51,266
Okay. All right. Just, I don't know, geographic, like in the office. I don't
2689
02:53:51,766 --> 02:53:54,283
know, whatever. Some, some people who are— I mean, I don't really have,
2690
02:53:54,965 --> 02:53:56,439
I really don't have those anymore, but okay.
2691
02:53:58,605 --> 02:54:02,065
Um, I don't know. Think of a previous job then. Uh, previous. Oh, that's easy.
2692
02:54:02,178 --> 02:54:05,562
Okay. How many of them could you replace with
2693
02:54:06,062 --> 02:54:07,587
AI were you tasked to do so?
2694
02:54:15,002 --> 02:54:18,498
Easily or with some care? Like, are we giving
2695
02:54:19,461 --> 02:54:22,415
me several months to develop a system that replaces the people or?
2696
02:54:23,490 --> 02:54:24,438
Give me both numbers.
2697
02:54:26,797 --> 02:54:30,682
Most people not trivially just by putting
2698
02:54:30,730 --> 02:54:34,872
in an LLM. Yeah, not a single prompt in a Grok model.
2699
02:54:34,920 --> 02:54:38,500
But you could probably reduce several headcounts
2700
02:54:39,000 --> 02:54:42,803
simply through redundancy, meaning you could have pre-filters easily done
2701
02:54:43,477 --> 02:54:48,390
that manage specific escalation workflows that make
2702
02:54:48,890 --> 02:54:52,838
sure a human's job did not require
several people
2703
02:54:52,870 --> 02:54:56,788
to take it. I see. So, you're saying, sorry, just because
2704
02:54:56,804 --> 02:55:00,256
redundancy is also like you've been made redundant. I think what
2705
02:55:00,756 --> 02:55:04,607
you're saying is you merge a job, like 3 or 4 jobs into one person's
2706
02:55:05,107 --> 02:55:08,920
job with AI. Yeah, yeah. Even, even just old, old server
2707
02:55:09,420 --> 02:55:12,510
maintenance people, right? You actually would have to go do regular checks and,
2708
02:55:12,686 --> 02:55:16,468
and look at logs. A human being reading logs is important, but if
2709
02:55:16,968 --> 02:55:19,978
a system can do pre-filtering, you can get a lot more done, you know,
2710
02:55:20,234 --> 02:55:22,926
if you're looking for anomalies and stuff like that. And,
2711
02:55:23,680 --> 02:55:26,756
oh yeah, and then writing the scripts to do the automation
2712
02:55:26,885 --> 02:55:30,106
and yada yada. Oh, sure. Okay. So got it.
2713
02:55:31,035 --> 02:55:31,211
But yeah.
2714
02:55:35,309 --> 02:55:38,942
Yeah. So moving forward here, Alex, we are now going to
2715
02:55:39,442 --> 02:55:43,414
move into the Jensen Huang scenario. The section of
2716
02:55:43,914 --> 02:55:47,830
our variety show. But first
2717
02:55:47,943 --> 02:55:51,811
we are going to play a little bit of Donald Trump being
2718
02:55:52,151 --> 02:55:55,476
interviewed by 60 Minutes last week because that was airing as
2719
02:55:55,591 --> 02:55:59,020
we were on where Do that
2720
02:55:59,485 --> 02:56:03,285
from now. Now, the 60 Minutes ensemble was giving
2721
02:56:03,397 --> 02:56:06,811
a rather long, hour-long interview with Trump, and Trump
2722
02:56:06,924 --> 02:56:10,418
is referring to questions involving AI.
2723
02:56:10,707 --> 02:56:14,731
And so we're going to get into that as we talk about Jensen. About that
2724
02:56:15,231 --> 02:56:19,524
economy, a big part of that economy has been the success of artificial
2725
02:56:20,024 --> 02:56:22,313
intelligence. I know we call it superintelligence in this house.
2726
02:56:23,500 --> 02:56:27,307
And you have said repeatedly, whoever wins superintelligence wins the
2727
02:56:27,807 --> 02:56:31,499
future. But again, if you go into polls, Americans, they don't love data
2728
02:56:31,999 --> 02:56:35,611
centers. They're pretty negative about AI superintelligence.
2729
02:56:36,301 --> 02:56:39,433
What does the average American get out of AI?
2730
02:56:39,514 --> 02:56:42,951
First, let me just tell you that when you say a big part,
2731
02:56:43,031 --> 02:56:47,111
yes, it's a part. But we're also building automobile factories
2732
02:56:47,384 --> 02:56:51,445
like we've never built before. Just announced four billion dollars.
2733
02:56:51,638 --> 02:56:54,734
No, but you said a big part automobiles.
2734
02:56:55,392 --> 02:56:59,708
How about pharmaceutical? Eli Lilly, Merck, Pfizer—they're building hundreds
2735
02:56:59,724 --> 02:57:02,291
of billions of dollars. There's a lot going on in the economy, but I want
2736
02:57:02,791 --> 02:57:06,414
to know about superintelligence. I know, but there has never been a time where
2737
02:57:06,799 --> 02:57:10,553
there is the activity. Trillions and trillions of dollars right
2738
02:57:11,053 --> 02:57:14,019
now as we sit is being built. In a year from now, because they started
2739
02:57:14,519 --> 02:57:16,892
a year ago, in a year from now, many of those plants are going to
2740
02:57:17,392 --> 02:57:20,810
be open, operating. But I want— I want— what do Americans
2741
02:57:20,922 --> 02:57:24,310
get? You have— I talked to people in Silicon Valley who
2742
02:57:24,810 --> 02:57:28,356
say that the decisions you make around AI are the most important things
2743
02:57:28,453 --> 02:57:32,017
that you'll be remembered for. That's what— that's what they— that's what they think.
2744
02:57:32,146 --> 02:57:35,036
I don't know if that's true. But at the same time, there is all of
2745
02:57:35,536 --> 02:57:38,729
this anxiety and concern. And I just want you to know, as someone— you are
2746
02:57:39,229 --> 02:57:42,580
an AI bull. I am. What do— what does the
2747
02:57:42,660 --> 02:57:45,948
average American get out of that technology? Excuse me,
2748
02:57:45,964 --> 02:57:49,604
before I answer that, the reason I'm an AI bull— yes,
2749
02:57:49,877 --> 02:57:53,806
one of the reasons is because President Xi left here 2
2750
02:57:54,306 --> 02:57:58,810
weeks ago. We had a great meeting. He's all in to
2751
02:57:59,211 --> 02:58:02,819
SI, as I call it, because it really is superintelligence. It's a— it's a much
2752
02:58:02,883 --> 02:58:08,224
better name, I think, because it's not artificial. It's— it's actually that at least
2753
02:58:08,724 --> 02:58:10,651
It's what they think they're racing towards, yes. So China— so we're in a race
2754
02:58:11,151 --> 02:58:14,135
with China. And if— and I've said whoever wins that race wins.
2755
02:58:14,761 --> 02:58:18,021
The level of power of winning that race is
2756
02:58:18,053 --> 02:58:21,232
very important. What is the power? And China has no problem with data centers.
2757
02:58:21,408 --> 02:58:24,941
Right. China has no problem with electricity. And by the way, we don't either,
2758
02:58:24,957 --> 02:58:28,811
because I'm letting— I'm demanding that they build their own electric
2759
02:58:29,311 --> 02:58:32,743
plant. Every one of those centers that goes up now under me
2760
02:58:33,080 --> 02:58:36,526
is building their own electric. You have— and they're giving the excess electric—
2761
02:58:37,312 --> 02:58:40,774
they're giving the excess electric right now. They're giving it into
2762
02:58:40,838 --> 02:58:44,654
the old grid that we had. We have an old broken grid, and that juice
2763
02:58:44,830 --> 02:58:47,956
is pouring into the grid. It's going to lower electric. You had all of the
2764
02:58:48,456 --> 02:58:51,707
AI tycoons in here. Yeah, I did.
And Elon Musk said then,
2765
02:58:51,980 --> 02:58:56,260
as he said often, the endgame with AI superintelligence
2766
02:58:56,372 --> 02:59:00,383
is that we're incredibly rich and nobody has to work and
2767
02:59:00,399 --> 02:59:03,867
everyone gets paid a universal high income. Yeah. You like
2768
02:59:03,915 --> 02:59:07,207
to work. Yeah. Does that sound like a good future to
2769
02:59:07,707 --> 02:59:11,510
you? No. What, what he meant, I think, is that when
2770
02:59:11,799 --> 02:59:15,155
a data center, as an example, goes to a community— look at West Virginia,
2771
02:59:15,171 --> 02:59:17,291
they're going wild, they love it, they say come on in.
2772
02:59:18,623 --> 02:59:22,075
And many, many countries, many places like it. By the way,
2773
02:59:22,220 --> 02:59:24,805
other countries like it. They're building them in Finland all of a sudden.
2774
02:59:25,672 --> 02:59:28,932
Because if we lose that, we're losing a cash cow. It's a cash cow.
2775
02:59:29,140 --> 02:59:32,432
But here's the thing. I think you're saving— so look, I think you
2776
02:59:32,932 --> 02:59:36,558
are taking his very weird statement and because you're a good politician,
2777
02:59:36,607 --> 02:59:40,733
making it sound normal. But the people in the AI
2778
02:59:40,846 --> 02:59:44,009
industry, I almost said the superintelligence industry, you're working on me,
2779
02:59:44,860 --> 02:59:47,991
are weird. They're strange. They do things like stand outside
2780
02:59:48,491 --> 02:59:51,828
their company and say, hey, We're creating this amazing company
2781
02:59:51,941 --> 02:59:55,730
and there's a 10% chance it kills humanity. Like you standing outside a hotel
2782
02:59:55,875 --> 02:59:58,444
and saying, here's my new hotel and it might collapse.
2783
02:59:58,749 --> 03:00:01,591
But you think that's fake.
2784
03:00:01,719 --> 03:00:05,396
Why are you not worried about the doom fears
2785
03:00:05,492 --> 03:00:09,169
that the people building it have? Because first of all, because China is
2786
03:00:09,202 --> 03:00:11,819
going to have it. Okay? So if we don't have it, China is going to
2787
03:00:12,319 --> 03:00:15,656
have it. That would put us at an unbelievable disadvantage. It's all between us
2788
03:00:15,704 --> 03:00:19,173
and China. There's nobody else. China is going to have it. China is now
2789
03:00:19,381 --> 03:00:23,315
one year behind us, which is amazing. And President Xi said, you are
2790
03:00:23,815 --> 03:00:26,799
really moved. President Xi got off his plane. He said, wow,
2791
03:00:26,944 --> 03:00:29,368
have you done a great job. And he means it. He's not a guy that,
2792
03:00:29,497 --> 03:00:32,611
you know— He doesn't trade in casual flattery?
2793
03:00:33,029 --> 03:00:36,898
I don't think so. I mean, I know him better than probably almost anybody.
2794
03:00:37,846 --> 03:00:41,314
But so we're racing China and that makes it worth the risk? Or you
2795
03:00:41,814 --> 03:00:44,975
just think the risk is not real? I think that the risk is not nearly
2796
03:00:45,087 --> 03:00:48,571
as real as it as they talk, and I
2797
03:00:48,604 --> 03:00:51,686
think the risk can be managed. And it's managed.
2798
03:00:51,751 --> 03:00:53,758
You know who by who? The DOJ.
2799
03:00:55,042 --> 03:00:58,639
We had a situation where somebody was reported
2800
03:00:59,139 --> 03:01:02,958
to us to be doing something bad with AI. We called them very strongly
2801
03:01:03,472 --> 03:01:06,380
and said, "Get ready to go to jail. Please give your wife a kiss." good
2802
03:01:06,880 --> 03:01:09,119
night. Who was this person? I can't tell you. Or this company? I can't tell
2803
03:01:09,619 --> 03:01:13,593
you. But we said, do you like the color orange?
2804
03:01:13,658 --> 03:01:17,520
Because get rid— go home right now. See your
2805
03:01:18,020 --> 03:01:20,686
wife and children. Give them a kiss good night.
2806
03:01:21,562 --> 03:01:25,080
And that's the story. Because tomorrow morning
2807
03:01:25,097 --> 03:01:28,179
you're going to jail. If the Democrats wanted to make a deal to put that
2808
03:01:28,260 --> 03:01:30,649
to— This person totally changed everything.
2809
03:01:32,260 --> 03:01:36,120
Totally. Yes, sir. Yes, sir. Was it a leader of a major AI company?
2810
03:01:36,500 --> 03:01:39,939
One of the big ones. Yeah. Okay. One of the big ones. And if the
2811
03:01:40,439 --> 03:01:43,780
Democrats said, we want to codify this into law, would you work
2812
03:01:44,280 --> 03:01:46,060
with them? I don't think it was done maliciously, but it could have been malicious.
2813
03:01:46,560 --> 03:01:51,320
Would you work with the Democrats? You're saying the DOJ can handle this on AI
2814
03:01:51,820 --> 03:01:55,260
safety. I don't mind guardrails, but I don't want to put us in a
2815
03:01:55,760 --> 03:01:58,846
deep second position when leading them will one. And China
2816
03:01:59,346 --> 03:02:02,341
has no guardrails, by the way. So if AI is dangerous— well, they have the
2817
03:02:02,841 --> 03:02:06,160
guardrails of being a totalitarian state, right? They have the guardrails that
2818
03:02:06,209 --> 03:02:09,814
whatever everyone says goes. Yeah. With me, it's pretty
2819
03:02:10,314 --> 03:02:13,764
strong, but it's not quite that strong. By the way, you know,
2820
03:02:13,844 --> 03:02:17,877
we have very brilliant people watching every one of these companies. Brilliant people.
2821
03:02:18,648 --> 03:02:21,540
People— I hope so. Well, they're the smartest.
2822
03:02:21,877 --> 03:02:26,993
They're of Hopefully equal. Equal intelligence,
2823
03:02:27,041 --> 03:02:30,214
though, to the people running the company? Which is not even— Are they
2824
03:02:30,714 --> 03:02:32,445
super intelligent? They're super intelligent. Okay, good.
2825
03:02:34,476 --> 03:02:37,781
All right, Alex, now he clearly doesn't understand what
2826
03:02:38,281 --> 03:02:41,723
Elon's future envisioning was.
2827
03:02:41,853 -->
03:02:45,690
It was much closer to WALL-E than Trump understands,
2828
03:02:46,775 --> 03:02:49,875
or at least will admit. Now,
2829
03:02:50,004 --> 03:02:51,723
interestingly enough, Alex,
2830
03:02:53,362 --> 03:02:57,121
if we narrow down who the potential big AI companies
2831
03:02:57,716 --> 03:03:01,198
people are, we have to rule out Elon Musk because he doesn't have a
2832
03:03:01,280 --> 03:03:03,968
wife to kiss before he goes into an orange jumpsuit.
2833
03:03:06,175 --> 03:03:10,814
So who does that leave, Alex? Because I don't
2834
03:03:11,041 --> 03:03:13,279
think Sam Altman has a wife either.
2835
03:03:14,890 --> 03:03:17,847
He has a husband and some kids by surrogate. Does that count?
2836
03:03:17,990 --> 03:03:21,297
Yes, yes. Well, he said kiss your wife goodbye, so I, you know,
2837
03:03:21,313 --> 03:03:24,669
we'll see what— I don't know how they call each other. I don't
2838
03:03:25,169 --> 03:03:27,495
like— This is true. This is true. We don't know that. We don't know the
2839
03:03:27,995 --> 03:03:30,321
pronouns. That's right.
2840
03:03:30,690 --> 03:03:33,998
All right. I was gonna say we finally have proof
2841
03:03:34,498 --> 03:03:37,739
of Russiagate. I don't know if you heard it when he said it,
2842
03:03:38,333 --> 03:03:40,565
but I've attached a tweet to the space.
2843
03:03:41,946 --> 03:03:45,184
There's a 2017 quote by Putin. Who says, uh,
2844
03:03:45,460 --> 03:03:49,183
whoever, uh, leads in super— in artificial intelligence
2845
03:03:49,443 --> 03:03:52,372
will be the ruler of the world. Now,
2846
03:03:52,597 --> 03:03:55,585
is Donnie quoting his handler? Oh,
2847
03:03:58,236 --> 03:04:01,337
now look, look, look, if anyone is gonna end the
2848
03:04:01,369 --> 03:04:05,569
world, it might as well be us. That's a, that's a nice—
2849
03:04:05,669 --> 03:04:08,997
It's our world and we'll end it whenever we pretty damn well
2850
03:04:09,110 --> 03:04:11,878
please, okay? Yes. Yes. All right.
2851
03:04:12,921 --> 03:04:16,789
Now, now, some people were wondering whether he was referring to—
2852
03:04:16,997 --> 03:04:20,239
they only because the news, I believe only because the
2853
03:04:20,739 --> 03:04:24,749
news was coinciding. They were wondering if he was referring to Jensen Huang. I don't
2854
03:04:24,781 --> 03:04:28,681
think he was. He was probably, he was probably referring to a
2855
03:04:29,034 --> 03:04:32,404
larger AI company whose LLMs in a
2856
03:04:32,904 --> 03:04:36,143
harness were touching a system they weren't supposed to be exploring.
2857
03:04:36,420 --> 03:04:39,499
That would be my guess. And they said,
2858
03:04:39,579 --> 03:04:43,192
knock it off or we actually are going to bring the DOJ against you for
2859
03:04:43,240 --> 03:04:46,789
this intrusion that we've been letting you slide on. That's my hypothesis.
2860
03:04:46,853 --> 03:04:50,531
Yeah. I don't know what company, but they've been going very lax on those
2861
03:04:50,579 --> 03:04:54,224
companies for the shenanigans that would typically
2862
03:04:54,256 --> 03:04:57,837
pass the threshold for cybercrimes,
2863
03:04:57,853 --> 03:05:01,065
as you were talking about, not with the thing you created,
2864
03:05:01,081 --> 03:05:06,019
but that you mentioned several times, which is the felony bench That's
2865
03:05:06,519 --> 03:05:10,165
right. Record. Well, I mean, let's just say, you know, Aaron Swartz was
2866
03:05:10,665 --> 03:05:13,150
driven to suicide for a lot less.
2867
03:05:13,909 --> 03:05:17,289
That, okay, that was the gentleman who committed suicide after he
2868
03:05:17,789 --> 03:05:21,358
was being prosecuted because he had access
2869
03:05:21,390 --> 03:05:25,458
to a bunch of journal
2870
03:05:26,133 --> 03:05:29,637
articles. Several hundred fifty. Yeah, so scientific,
2871
03:05:29,766 --> 03:05:33,039
mathematics, et cetera, et cetera, and released them for
2872
03:05:33,072 --> 03:05:36,732
free. Uh, and, and was being prosecuted
2873
03:05:36,748 --> 03:05:39,699
for that. Yeah, well, they, they,
2874
03:05:41,101 --> 03:05:45,281
they considered his accessing the articles
2875
03:05:45,378 --> 03:05:49,253
at a high rate as hacking and
2876
03:05:49,753 --> 03:05:53,075
basically threatened to throw the book at him, as prosecutors do,
2877
03:05:53,928 --> 03:05:56,486
in order to get a plea to a lesser charge,
2878
03:05:57,388 --> 03:05:58,755
uh, which he didn't take.
2879
03:06:01,011 --> 03:06:03,786
Well, he took a higher charge. He pled to a higher charge.
2880
03:06:03,867 --> 03:06:06,993
Remember, they threatened him with like 40 years in prison or whatever. Well, I mean,
2881
03:06:07,073 --> 03:06:09,612
committing suicide is a higher charge than that.
2882
03:06:10,802 --> 03:06:12,775
Well, I mean, I guess he,
2883
03:06:14,330 --> 03:06:18,492
whatever, you know, somebody with a— has a depressive streak could
2884
03:06:18,540 --> 03:06:21,848
be destabilized by the prospect
2885
03:06:22,348 --> 03:06:25,913
of decades in prison. Who knows? Who knows? Life's a
2886
03:06:25,930 --> 03:06:29,476
mystery. Yes. Anyway, long story short,
2887
03:06:30,341 --> 03:06:33,481
um, I'm saying what he did is a lot less than what
2888
03:06:33,981 --> 03:06:37,102
these people are doing routinely as part of development. And they don't seem
2889
03:06:37,151 --> 03:06:41,541
to be getting threatened with anything. So, um, yeah, it's fascinating.
2890
03:06:42,999 --> 03:06:46,908
All right. Uh, now the memes, Alex, the memes recently
2891
03:06:47,084 --> 03:06:50,273
have been prime. They've been premium memes.
2892
03:06:50,609 --> 03:06:53,429
Premium memes. Have you seen, uh, have you seen, uh,
2893
03:06:53,654 --> 03:06:56,820
Jordan Peterson, uh, Chastises Jensen Huang.
2894
03:06:58,445 --> 03:07:02,162
No one blames you. It's just, you know, the spy
2895
03:07:02,662 --> 03:07:06,370
thing and the harem thing. You can't fuck
2896
03:07:06,870 --> 03:07:08,920
too many
bitches, Jensen. You got to choose one.
2897
03:07:10,300 --> 03:07:13,541
Like Carl Jung or Solzhenitsyn, whoever said
2898
03:07:13,702 --> 03:07:17,338
in The Gulag Archipelago, it says the line between
2899
03:07:17,516 --> 03:07:20,355
ass and asset runs through every human heart.
2900
03:07:21,969 --> 03:07:25,751
This is unbelievable. You work in AI, make a few billion dollars,
2901
03:07:25,783 --> 03:07:28,887
and suddenly beautiful Chinese supermodel spies are throwing
2902
03:07:28,919 --> 03:07:31,328
themselves at you. Yeah, terrible problem to have.
2903
03:07:32,492 --> 03:07:35,273
Wait a minute, wait a minute, I've got an idea.
2904
03:07:36,227 --> 03:07:39,538
Oh no. At OpenAI, we're not
2905
03:07:40,038 --> 03:07:43,347
just building artificial intelligence, we're building the future. And let me tell
2906
03:07:43,379 --> 03:07:47,374
you, the future is looking very good. Oh, you've Very important man.
2907
03:07:47,637 --> 03:07:49,917
Here, my number. Maybe we go drink together?
2908
03:07:52,088 --> 03:07:52,735
Unbelievable.
2909
03:07:54,337 --> 03:07:57,622
So, George,
2910
03:07:57,897 --> 03:08:01,037
you work OpenAI, yes? You see all secret AI models? Oh,
2911
03:08:01,117 --> 03:08:04,772
absolutely. Very secret. Top secret models. Maybe you show
2912
03:08:05,272 --> 03:08:09,148
me? Just little look. I promise I no tell anybody. Well, I suppose
2913
03:08:09,247 --> 03:08:10,795
I could arrange a private viewing.
2914
03:08:14,292 --> 03:08:17,819
You know, I've never really, uh, dated anyone from China
2915
03:08:18,319 --> 03:08:21,219
before. China? No, no, George, I not Chinese. I Thai.
2916
03:08:22,076 --> 03:08:25,551
Thai? Yes, and I have little surprise for you.
2917
03:08:27,486 --> 03:08:28,362
Oh, come on!
2918
03:08:32,936 --> 03:08:35,320
Oh, Mario, I didn't expect to see you here.
2919
03:08:38,519 --> 03:08:42,046
Let's-a go! Dear Jensen, I have baked a cake for you.
2920
03:08:42,126 --> 03:08:46,041
Thank you for being my husband for 40 years. It sure would
2921
03:08:46,057 --> 03:08:49,859
be bad if you destroyed our marriage along with the US economy by cheating
2922
03:08:49,972 --> 03:08:52,972
on me. Oh, it's you again.
2923
03:08:53,662 --> 03:08:56,470
Listen, Mario, I'm kind of busy today.
2924
03:08:57,529 --> 03:09:00,866
Isn't it
2925
03:09:01,781 --> 03:09:05,038
great, Mario? We banned the data centers!
2926
03:09:05,688 --> 03:09:08,813
Now our Black community won't be plagued with millions of dollars
2927
03:09:08,845 --> 03:09:12,444
in revenue and countless jobs. Gee, it's too
2928
03:09:12,944 --> 03:09:16,652
bad that AI in America is facing so many headwinds of
2929
03:09:17,152 --> 03:09:20,882
a purely organic nature. I guess the future belongs
2930
03:09:21,382 --> 03:09:25,650
to the Chinese. Oh well. We proudly stand against scientific
2931
03:09:26,150 --> 03:09:29,572
breakthroughs and medical advances which could save millions of lives.
2932
03:09:29,830 --> 03:09:30,939
Oh, hi, Chinese Toad.
2933
03:09:33,903 --> 03:09:37,526
Listen here, you little shit. I don't know what you want
2934
03:09:38,026 --> 03:09:41,832
from me, but you're going to be speaking to my lawyers. But we
2935
03:09:42,332 --> 03:09:46,051
begin tonight with the breaking news: the Chinese spy scandal involving the CEO of
2936
03:09:46,551 --> 03:09:50,869
one of America's top chipmakers, Nvidia. CEO Jensen Huang now being investigated
2937
03:09:50,901 --> 03:09:54,664
by his own company, admitting to an affair with a businesswoman. The New York Times
2938
03:09:54,696 --> 03:09:57,878
tonight reporting that U.S. intelligence had identified her
2939
03:09:57,926 --> 03:10:01,816
as a suspected spy for China. Which China denies tonight.
2940
03:10:02,281 --> 03:10:05,959
Late today, President Trump weighing in on Nvidia's CEO, who had lobbied
2941
03:10:06,459 --> 03:10:09,572
the president to be able to sell those cutting-edge chips to China.
2942
03:10:10,086 --> 03:10:14,068
Tonight, Nvidia's stock falling more than 4% at the close. ABC's Jay O'Brien
2943
03:10:14,568 --> 03:10:17,826
leading us off tonight. Tonight, President Trump defending Jensen
2944
03:10:17,922 --> 03:10:21,310
Huang, the CEO of the world's most powerful chipmaker Nvidia,
2945
03:10:21,471 --> 03:10:25,405
who admitted to an affair with an alleged Chinese spy. He's an
2946
03:10:25,905 --> 03:10:29,175
amazing guy. I just hope it's I hope there's no problem.
2947
03:10:29,480 --> 03:10:32,045
If there is a problem, we'll let you know about it. But I hope there's
2948
03:10:32,545 --> 03:10:36,117
no problem. I believe he's suffering greatly. The New York Times
2949
03:10:36,133 --> 03:10:39,660
reporting U.S. intelligence officials determined this woman, Alice Wong,
2950
03:10:39,836 --> 03:10:43,187
was recruited by the Chinese government to get information from the
2951
03:10:43,687 --> 03:10:48,140
Nvidia CEO and gain access to the company's cutting-edge chips, something Beijing
2952
03:10:48,253 --> 03:10:51,764
denies. Trump saying he spoke today with the Nvidia CEO
2953
03:10:51,780 --> 03:10:54,877
at the center of the scandal. I asked him that question. Did you have any
2954
03:10:54,909 --> 03:10:57,831
idea that she was a spy? He said, "I had no idea that she was
2955
03:10:57,943 --> 03:11:01,058
a spy," and I really believe him. Alice Huang is a Chinese
2956
03:11:01,154 --> 03:11:04,269
tech executive identified here in a video from Taiwanese
2957
03:11:04,301 --> 03:11:08,379
media wearing a pink dress alongside the Nvidia CEO last
2958
03:11:08,879 --> 03:11:12,698
year. In an email to Nvidia employees, Huang writing, "I was unfaithful to
2959
03:11:12,731 --> 03:11:16,744
my wife. My actions hurt her and our children. There is no excuse
2960
03:11:17,244 --> 03:11:20,965
for what I did. I am deeply sorry." The company's co-founder Jensen
2961
03:11:21,062 --> 03:11:24,896
Huang has led Nvidia's meteoric rise, becoming one of the most
valuable
2962
03:11:24,928 --> 03:11:28,522
companies in the world. Its high-tech chips power almost every
2963
03:11:28,683 --> 03:11:32,068
AI company. Huang also serves as a close advisor
2964
03:11:32,568 --> 03:11:36,159
to President Trump, flying with him aboard Air Force One on a trip to China
2965
03:11:36,659 --> 03:11:40,010
last spring, attending a state dinner for Chinese President Xi Jinping in
2966
03:11:40,510 --> 03:11:44,133
September, and days later front and center at a White House summit for AI
2967
03:11:44,181 --> 03:11:48,272
leaders. And last year, after lobbying from the Nvidia CEO, President Trump
2968
03:11:48,352 --> 03:11:52,122
loosened government restrictions on sending high-powered chips to China.
2969
03:11:52,443 --> 03:11:56,534
The Times reports Trump was briefed roughly a year ago on Jensen Huang's interactions
2970
03:11:57,034 --> 03:12:00,352
with the alleged Chinese spy and chose not to act. And this spring,
2971
03:12:00,400 --> 03:12:03,817
Secretary of State Marco Rubio reportedly warned the two had developed
2972
03:12:04,317 --> 03:12:07,603
a close relationship. Today, Rubio pressed if national security was
2973
03:12:07,683 --> 03:12:11,132
compromised. Is it a national security concern? I'm just not going I'm not going to
2974
03:12:11,164 --> 03:12:13,457
comment on it. Can I ask you to? I think they'll be rude. And I
2975
03:12:13,521 --> 03:12:17,673
can't, and I'm not going to. Tonight, President Trump says the FBI is investigating,
2976
03:12:18,025 --> 03:12:21,231
and Nvidia's board has launched its own investigation, saying they have
2977
03:12:21,731 --> 03:12:25,768
no reason to believe that either national security or Nvidia's secrets were compromised.
2978
03:12:26,040 --> 03:12:29,535
But on Capitol Hill, lawmakers telling me they're concerned. Are you
2979
03:12:29,551 --> 03:12:32,260
concerned by what you've seen publicly in this Huang case?
2980
03:12:32,853 --> 03:12:36,235
There are real concerns that need to be addressed with respect to
2981
03:12:36,300 --> 03:12:39,541
the Nvidia situation. We're going to need a full
2982
03:12:40,055 --> 03:12:44,021
briefing specifically as it relates to the Nvidia
2983
03:12:44,230 --> 03:12:47,827
issue so we can all get to the bottom of what
2984
03:12:47,843 --> 03:12:51,793
has happened and what the implications are in
2985
03:12:52,293 --> 03:12:56,032
terms of America's national security. David, tonight Nvidia's board
2986
03:12:56,532 --> 03:13:00,431
saying it has confidence in Jensen Huang's leadership. He's been deeply ingrained in that company
2987
03:13:00,447 --> 03:13:04,459
for years. Even his children work there today. Nvidia stock tumbling more
2988
03:13:04,959 --> 03:13:08,017
than 4%. David. Jay O'Brien leading us off here on a Monday night.
2989
03:13:08,146 --> 03:13:11,656
Jay, thank you. First thing in the morning. Thanks, Ginger. There is news tonight involving
2990
03:13:11,688 --> 03:13:15,584
the CEO of American
chipmaker Nvidia and his affair with an alleged Chinese
2991
03:13:15,616 --> 03:13:19,143
spy. But the New York Times is now reporting tonight that President Trump
2992
03:13:19,643 --> 03:13:23,118
was warned— still CEO Jensen Huang boarding Air Force One.
2993
03:13:23,696 --> 03:13:27,286
Here's Jay O'Brien. Tonight we're learning new details
2994
03:13:27,318 --> 03:13:30,733
about the warnings President Trump received that Nvidia CEO
2995
03:13:30,749 --> 03:13:33,987
Jensen Huang was the target of an alleged Chinese spy.
2996
03:13:34,517 --> 03:13:38,470
The New York Times reporting Secretary of State Marco Rubio was traveling to China
2997
03:13:38,551 --> 03:13:41,604
on Air Force One in May when he shared intelligence with the
2998
03:13:42,104 --> 03:13:45,444
president showing the Nvidia chief was in a relationship with a woman believed
2999
03:13:45,944 --> 03:13:48,818
to be a Chinese spy. Still, a short time later,
3000
03:13:48,899 --> 03:13:52,498
Jensen Huang would board Air Force One in Alaska, a late addition
3001
03:13:52,530 --> 03:13:56,194
to the trip, arriving in Beijing steps behind
3002
03:13:56,210 --> 03:13:59,665
Rubio. It was the second time in seven months Trump had been reported.
3003
03:13:59,810 --> 03:14:03,939
Reportedly warned about the plot, but the president now says he had no idea.
3004
03:14:04,180 --> 03:14:07,682
When you went to China earlier this year in May, did you know about these
3005
03:14:08,035 --> 03:14:11,441
alleged ties with the Chinese spy then, or? No, I didn't hear
3006
03:14:11,554 --> 03:14:14,864
about it then, but a lot of people have heard about it,
3007
03:14:15,571 --> 03:14:19,185
and Jensen totally denies that. The woman at the center of the scandal,
3008
03:14:19,346 --> 03:14:23,555
Alice Wong, is a Chinese tech executive who grew close to the Nvidia CEO
3009
03:14:23,684 --> 03:14:27,715
over several years, posting pictures of their time together. Taiwanese media
3010
03:14:28,215 --> 03:14:31,873
capturing this kiss on the cheek at a dinner last year. But around the same
3011
03:14:31,969 --> 03:14:36,190
time, Alice Huang's company, MegaSpeed, was reportedly under investigation
3012
03:14:36,223 --> 03:14:40,155
by the U.S. government for allegedly smuggling cutting-edge Nvidia chips
3013
03:14:40,655 --> 03:14:44,216
to China, something the company denies. Tonight, House Republicans are demanding
3014
03:14:44,312 --> 03:14:47,860
Jensen Huang explain behind closed doors what he discussed
3015
03:14:48,360 --> 03:14:51,840
with the alleged spy, threatening to subpoena him if he doesn't respond.
3016
03:14:52,049 --> 03:14:54,168
Are you worried about the national security implications of this?
3017
03:14:54,906 --> 03:14:58,425
100%. David, House Republicans have given Jensen Huang
3018
03:14:58,457 --> 03:15:02,107
until tomorrow to respond to their demand for a briefing. David.
3019
03:15:02,237 --> 03:15:03,678
Jay O'Brien with us tonight live at Wall Street.
3020
03:15:06,155 --> 03:15:09,591
So Alex, we're going through this a little bit in chronological order since it's been
3021
03:15:09,705 --> 03:15:12,979
a wild week. We're not jumping to the end of the story. We're going to
3022
03:15:13,044 --> 03:15:16,566
get there and we're going to go through what the New York Times
3023
03:15:17,389 --> 03:15:21,280
sort of did as they were investigating this story
3024
03:15:21,393 --> 03:15:26,578
because we want the details. And I know that there's a lot
3025
03:15:27,078 --> 03:15:28,540
of nuance and intricacy in it. So we're not going to blaze past it.
3026
03:15:28,588 --> 03:15:31,607
We're not going to jump to the conclusions. We're going to actually go through it
3027
03:15:32,137 --> 03:15:35,813
and discuss it broadly. So starting at that specific portion, this is
3028
03:15:36,313 --> 03:15:39,875
before the announcement of additional women
3029
03:15:40,469 --> 03:15:44,306
and his resignation. This was the opening that everyone
3030
03:15:44,403 --> 03:15:48,834
was getting basically as the initial reports were coming. You were
3031
03:15:49,334 --> 03:15:52,831
pointing out last week as we were announcing the breaking news, so to
3032
03:15:53,331 --> 03:15:56,956
speak, that the sources were all anonymized.
3033
03:15:57,020 --> 03:16:00,918
There were no specific names as being involved
3034
03:16:01,833 --> 03:16:05,298
in commenting to the press about this supposed
3035
03:16:05,798 --> 03:16:09,325
report. No, Marco Rubio was named as the source.
3036
03:16:10,015 --> 03:16:13,561
Well, yes, right. Marco Rubio's name was mentioned, but not as the
3037
03:16:14,061 --> 03:16:17,395
source. It was just sources close to— could say. Now,
3038
03:16:18,663 --> 03:16:21,980
I think it is safe to say that yes, there were
3039
03:16:22,432 --> 03:16:26,107
probably people inside of the government
3040
03:16:26,123 --> 03:16:29,397
who had written that they were investigating to see if she
3041
03:16:29,429 --> 03:16:33,474
was a spy. I'm sure that that is something that we could take to
3042
03:16:33,974 --> 03:16:35,884
the bank, that someone's there. Now, the question is,
3043
03:16:37,950 --> 03:16:41,201
the US intelligence apparatus is lots of
3044
03:16:41,701 --> 03:16:45,918
people, and firm conclusions are something that is a process.
3045
03:16:46,030 --> 03:16:49,841
And there might have been a situation situation where an analyst here or there
3046
03:16:50,531 --> 03:16:54,682
had a strong opinion and it's them who went to the press and
3047
03:16:54,778 --> 03:16:58,097
gave their opinion, and that is the opinion that the press is running with
3048
03:16:58,562 --> 03:17:01,960
and asserting as the opinion of the full US government's
3049
03:17:04,028 --> 03:17:07,394
intelligence agencies, right? They're creating the impression that it's
3050
03:17:07,894 --> 03:17:11,482
a unanimous sort of conclusion that has the weight, but it's really
3051
03:17:11,982 --> 03:17:15,489
just an
analyst or two. Presumably, if it was
3052
03:17:15,538 --> 03:17:19,048
a unanimous conclusion that he was consorting with a spy and presumably doing
3053
03:17:19,385 --> 03:17:23,818
illicit illicit things, um, he would have been prosecuted,
3054
03:17:25,329 --> 03:17:28,574
right? I mean, this isn't like— they're not alleging that he,
3055
03:17:28,799 --> 03:17:32,287
like, played chess with her or something. They are alleging that he did something
3056
03:17:32,432 --> 03:17:35,814
illicit, or at least that was the implication, in which
3057
03:17:35,911 --> 03:17:37,866
case it would be criminal, right?
3058
03:17:39,467 --> 03:17:42,942
Um, if he didn't— if he knew that he was consorting
3059
03:17:42,974 --> 03:17:46,696
with a spy, perhaps, but if he didn't know that
3060
03:17:47,196 --> 03:17:49,474
he was consorting with a spy, then I don't believe there would be any crimes.
3061
03:17:49,908 --> 03:17:53,585
Then what's the wrongdoing though? The faithlessness to
3062
03:17:54,085 --> 03:17:57,438
his wife? Well, that's one issue.
3063
03:17:57,551 --> 03:18:01,404
Yeah, that's a problem. But then the
3064
03:18:01,904 --> 03:18:05,772
further scandal involves the fact that they're alleging that Trump was
3065
03:18:05,868 --> 03:18:10,219
warned about this inside of his intelligence and did not go and
3066
03:18:11,550 --> 03:18:16,539
deal with Hwang so that that base knew they were covering up for Hwang.
3067
03:18:16,668 --> 03:18:20,336
If I was Trump, I would definitely take intelligence warnings extremely
3068
03:18:26,731 --> 03:18:30,659
Now, let's go a little bit through the New York Times
3069
03:18:30,836 --> 03:18:34,587
telling of how they broke this story, because there's additional details that are
3070
03:18:35,087 --> 03:18:37,939
not necessarily in the articles themselves. That might, uh,
3071
03:18:38,293 --> 03:18:42,534
that might have some pertinence. You have to back up to last August
3072
03:18:42,647 --> 03:18:46,358
when Anna Swanson and I got a message from a tech
3073
03:18:46,858 --> 03:18:51,001
industry contact of ours who said, hey, you really need
3074
03:18:51,501 --> 03:18:54,488
to drop everything and look at this company called Megaspeed.
3075
03:18:55,644 --> 03:18:59,147
It's a Singaporean data center firm.
3076
03:18:59,420 --> 03:19:03,147
I mean, it's literally a company that buys chips from Nvidia,
3077
03:19:03,612 --> 03:19:07,461
and builds data centers. And there's nothing about that on the surface
3078
03:19:07,961 --> 03:19:11,069
that says this is something we should drop what we're doing and devote time and
3079
03:19:11,569 --> 03:19:15,335
resources to figuring out what this company is and why it matters.
3080
03:19:16,105 --> 03:19:20,451
But Nvidia was selling chips to this company at
3081
03:19:20,951 --> 03:19:24,412
a time of increasing controversy over
3082
03:19:24,540 --> 03:19:28,276
where AI chips go. The U.S. government has been concerned for
3083
03:19:28,453 --> 03:19:31,470
years about the arms race,
3084
03:19:31,518 --> 03:19:35,130
so to speak, between the US and China over who's
3085
03:19:35,630 --> 03:19:39,592
going to achieve supremacy in artificial intelligence. And when the AI revolution
3086
03:19:39,641 --> 03:19:43,156
started in late 2022, the US
3087
03:19:43,656 --> 03:19:46,832
had restricted China from buying Nvidia's
3088
03:19:46,848 --> 03:19:50,781
best chips and best technology because it wanted to stop China
3089
03:19:50,797 --> 03:19:55,405
from being able to build technology that was as sophisticated
3090
03:19:55,662 --> 03:19:59,023
and comprehensive as what US companies were building.
3091
03:20:00,263 --> 03:20:04,514
But there was a big loophole, and that loophole was that
3092
03:20:05,014 --> 03:20:08,307
Nvidia could sell chips to companies in Southeast Asia and
3093
03:20:08,388 --> 03:20:12,382
that those companies could then provide access to
3094
03:20:12,882 --> 03:20:17,357
that technology. What was
3095
03:20:17,857 --> 03:20:21,198
that? I need a safe word for whenever the word loophole comes
3096
03:20:21,698 --> 03:20:25,210
in because we need to clarify. That when they say loophole, they mean
3097
03:20:25,710 --> 03:20:29,150
a, a thing that is absolutely legal. Yes. Multiple people have tried
3098
03:20:29,650 --> 03:20:33,250
to make it illegal and proposed bills to do so. Yes. Those bills have failed,
3099
03:20:33,710 --> 03:20:38,070
as in being voted against, because presumably if,
3100
03:20:38,570 --> 03:20:41,790
you know, the Congress and whatever is the expression of the people of the United
3101
03:20:42,290 --> 03:20:45,850
States, they want it to be legal. Okay. If you, you could say democracy doesn't
3102
03:20:46,350 --> 03:20:49,630
represent people, but I mean, longstanding pet peeve of mine. I don't think
3103
03:20:50,130 --> 03:20:52,970
those reporters feel that way. Longstanding pet peeve of mine, Alex, that you've just mentioned.
3104
03:20:53,470 --> 03:20:56,619
Yeah, I know. I know. There's a loophole, there's a loophole,
3105
03:20:56,747 --> 03:20:59,700
which is that the law can be used to do things that are supposed to
3106
03:21:00,200 --> 03:21:03,070
be illegal. Yeah, yeah. And we don't like that, goddammit!
3107
03:21:03,552 --> 03:21:07,066
Loophole is code word for, in our idealized set of laws,
3108
03:21:07,243 --> 03:21:10,757
this would not be allowed, but it is allowed because we haven't figured out
3109
03:21:10,774 --> 03:21:14,657
how to criminalize it. And it's
3110
03:21:15,157 --> 03:21:18,059
not because it was missed, but it was because we have suggested that it should
3111
03:21:18,156 --> 03:21:21,349
not be legal multiple times, and people just don't want to agree with us.
3112
03:21:22,104 --> 03:21:25,673
The real version of what a loophole would be would be that everyone
3113
03:21:25,737 --> 03:21:29,755
agreed that a certain thing would be illegal, but through some Rube
3114
03:21:29,803 -->
03:21:32,958
Goldberg mechanics of law, you could, with a
3115
03:21:33,458 --> 03:21:36,452
special combination of things, you can actually achieve it.
3116
03:21:36,742 --> 03:21:40,348
That's what a loophole would be, right? But this is— A loophole is like,
3117
03:21:40,444 --> 03:21:45,154
for example, let's say people say each
3118
03:21:45,266 --> 03:21:49,181
person or company could get only this much funding from the government. government and
3119
03:21:49,213 --> 03:21:52,489
somebody forms a company, gets funding, bankrupts the
3120
03:21:52,989 --> 03:21:56,520
company, forms another, gets more funding. Ergo, they are the
3121
03:21:57,020 --> 03:21:59,764
beneficiary. But because of the indirection,
3122
03:22:00,214 --> 03:22:03,666
right, they managed somehow to— because people had not foreseen that
3123
03:22:04,166 --> 03:22:07,585
this could be done and it was not explicitly covered, and they can get away
3124
03:22:08,085 --> 03:22:11,551
with it. That's a loophole, right? This is like everybody knew it
3125
03:22:11,599 --> 03:22:14,908
could be done. It was being done. It is being done by US companies as
3126
03:22:14,972 --> 03:22:18,473
well. Yeah. Like Google has, uh, data centers in Southeast Asia. Of course,
3127
03:22:18,780 --> 03:22:22,288
like everything, it's not— there's nothing weird about this at all,
3128
03:22:22,821 --> 03:22:23,484
but we don't like it.
3129
03:22:26,080 --> 03:22:29,753
Go on. Remotely to
3130
03:22:30,253 --> 03:22:33,529
Chinese companies. And as we began to examine
3131
03:22:34,029 --> 03:22:37,477
Megaspeed, it was pretty clear they're kind of the poster child for
3132
03:22:37,606 --> 03:22:41,722
exploiting this loophole. How so? How so?
3133
03:22:41,949 --> 03:22:44,536
I mean, in the span of roughly 2 years,
3134
03:22:45,057 --> 03:22:48,882
they had purchased more than $4.8 billion in
3135
03:22:48,979 --> 03:22:52,512
chips from Nvidia. I mean, they went from being a company that nobody
3136
03:22:52,544 --> 03:22:55,722
knew of overnight to a company that had become
3137
03:22:55,770 --> 03:22:57,933
one of Nvidia's largest customers in the region.
3138
03:23:01,386 --> 03:23:03,911
And I mean, this was a company with clear ties to China.
3139
03:23:04,866 --> 03:23:08,088
So we wanted to learn more about those ties and dispatched
3140
03:23:08,588 --> 03:23:12,513
our colleague Paul Moser from Taipei to Singapore to visit
3141
03:23:12,690 --> 03:23:16,126
the Megaspeed office. And when he showed up, he found that there were only
3142
03:23:16,142 --> 03:23:19,707
2 employees at the office. And when he spoke to them,
3143
03:23:20,397 --> 03:23:24,363
they said that most all of their customers were Chinese and most
3144
03:23:24,863 --> 03:23:28,762
of the executives were Chinese and rarely came to Singapore. In fact,
3145
03:23:28,939 --> 03:23:32,761
most of them stayed mostly in China. This was significant because
3146
03:23:33,261 --> 03:23:36,678
at the time Nvidia was telling us that Megaspeed didn't really have ties
3147
03:23:36,791 --> 03:23:40,172
to China. Paul also went and
3148
03:23:40,220 --> 03:23:43,634
found that Megaspeed was owned by a shell company, and he
3149
03:23:43,746 --> 03:23:47,271
visited the shell company and found that it was based in an
3150
03:23:47,771 --> 03:23:51,293
accounting firm, almost like a shell company mill. All of
3151
03:23:51,793 --> 03:23:54,867
this was mystifying for us. Unusual,
3152
03:23:54,979 --> 03:23:58,120
just because why would such a big company be housed with an accounting firm?
3153
03:23:58,264 --> 03:24:01,533
Right. You're buying $4.8 billion of chips. You would think you
3154
03:24:02,033 --> 03:24:04,997
could spring for your own office space. Right, right.
3155
03:24:07,068 --> 03:24:10,790
So it sounds like you're finding there are a lot of questions to be asked
3156
03:24:10,920 --> 03:24:14,546
about this company's connections to China and how it became
3157
03:24:14,594 --> 03:24:18,145
such a major buyer of these chips basically out of the blue.
3158
03:24:18,615 --> 03:24:22,116
Yeah, absolutely. I mean, and as we dug into this company
3159
03:24:22,181 --> 03:24:25,958
and looked at it, we found that they had aroused the suspicions of
3160
03:24:26,296 --> 03:24:30,318
the Commerce Department and commerce officials who are supposed to track where chips
3161
03:24:30,367 --> 03:24:33,980
are going around the world. And so they dispatched investigators to
3162
03:24:34,061 --> 03:24:37,401
visit one of the data centers. And when they showed up in September of
3163
03:24:37,433 --> 03:24:40,886
2024, they found that Megaspeed still had a lot
3164
03:24:41,386 --> 03:24:44,837
of the Nvidia technology in boxes, which raised suspicion that
3165
03:24:44,886 --> 03:24:48,644
that technology was being shipped in and then onward
3166
03:24:48,676 --> 03:24:52,177
to potentially China. It really raised the prospects that this was
3167
03:24:52,225 --> 03:24:55,325
a smuggling operation, not just a data center operation.
3168
03:24:55,922 --> 03:24:59,477
Fascinating. And at the center of all of this was this
3169
03:24:59,574 --> 03:25:03,229
woman named Alice Wong, or Wong Le, a Chinese citizen
3170
03:25:03,758 --> 03:25:08,214
who claimed to be the company's chief executive officer on her LinkedIn,
3171
03:25:08,952 --> 03:25:12,671
and who had a biography that detailed that she
3172
03:25:12,719 --> 03:25:16,807
used to be a TV journalist. And this just
3173
03:25:17,307 --> 03:25:20,895
made us question, how did she get all this money? How did
3174
03:25:21,395 --> 03:25:23,684
she come to be in charge of this company? And then most of all,
3175
03:25:23,780 --> 03:25:27,724
and most mystifying, how did she get so close to Jensen Huang, the person
3176
03:25:28,224 --> 03:25:32,246
who is at the forefront of artificial intelligence. How close
3177
03:25:32,518 --> 03:25:36,318
a relationship are you aware that they have at
3178
03:25:36,818
--> 03:25:39,990
this point? Like, what do you know about that? We don't know a lot,
3179
03:25:40,134 --> 03:25:43,517
but she's been photographed with him on a
3180
03:25:44,017 --> 03:25:47,526
few different occasions and videoed with him at various industry
3181
03:25:47,590 --> 03:25:50,957
gatherings. And so, we do some reporting. We speak
3182
03:25:51,457 --> 03:25:54,965
to a man who took a photograph of them at an industry party.
3183
03:25:55,662 --> 03:25:59,180
at a bar in Taipei. And there's a photograph of Jensen
3184
03:25:59,244 --> 03:26:02,977
sitting beside Alice as they prepare to toast
3185
03:26:03,477 --> 03:26:07,143
whiskey. And the gentleman who took the photograph is a tech executive named Parker
3186
03:26:07,643 --> 03:26:10,717
Schmidt. And Parker got on the phone and he said, you know, the wild thing
3187
03:26:10,765 --> 03:26:14,530
about that photo is we were hanging out with Alice at this party and
3188
03:26:15,030 --> 03:26:18,424
she said something to the effect of, I bet I can get Jensen
3189
03:26:18,440 --> 03:26:21,693
here. And sure enough, a few minutes later,
3190
03:26:21,741 --> 03:26:24,776
Jensen shows An indication that, I mean,
3191
03:26:25,258 --> 03:26:28,567
she may have him on speed dial or something like that, at least. Right.
3192
03:26:28,711 --> 03:26:31,298
That was the way it was perceived by others at this party.
3193
03:26:31,924 --> 03:26:35,394
And you gotta remember, like, this is in Taipei. This is where
3194
03:26:35,894 --> 03:26:38,414
he is rock star among rock stars.
3195
03:26:39,073 --> 03:26:41,949
And this is the industry that he is building and leading.
3196
03:26:42,447 --> 03:26:44,808
So to be able to, in moments,
3197
03:26:45,210 --> 03:26:49,402
summon the most important figure in the industry really catches people's
3198
03:26:49,902 --> 03:26:53,830
attention. All right. So you drop this reporting. What happens?
3199
03:26:54,167 --> 03:26:57,616
Not much. It's quiet. Nvidia keeps doing business with
3200
03:26:58,116 --> 03:27:02,108
Megaspeed, and we don't hear a lot for a while. And then about
3201
03:27:02,608 --> 03:27:05,830
a month ago, we hear that our colleagues in Washington on the national
3202
03:27:06,330 --> 03:27:09,584
security beat all of a sudden are hearing things about Alice Wong,
3203
03:27:10,049 --> 03:27:12,375
and she's resurfaced in our universe.
3204
03:27:13,017 --> 03:27:18,294
And we start to do some reporting, and it turns out
3205
03:27:18,794 --> 03:27:22,095
It turns out that she has been under US surveillance for about a year,
3206
03:27:22,225 --> 03:27:22,890
if not longer.
3207
03:27:26,003 --> 03:27:29,313
And in the course of surveilling her and her
3208
03:27:29,813 --> 03:27:34,076
communication, US intelligence services have determined that she's
3209
03:27:34,576 --> 03:27:38,256
a Chinese asset funneling information to Beijing.
3210
03:27:38,758 --> 03:27:40,768
A spy. A spy, basically.
3211
03:27:42,998 --> 03:27:46,167
And lo and behold, who do they find out she's in touch with?
3212
03:27:46,617 --> 03:27:49,914
Jensen Wong. All right. Now I want to point out
3213
03:27:50,479 --> 03:27:53,490
that that's somewhat misleading because it
3214
03:27:53,990 --> 03:27:57,585
is not an announced universal conclusion by the US intelligence agencies
3215
03:27:57,683 --> 03:28:00,920
that she is a spy. So like,
3216
03:28:01,664 --> 03:28:07,005
there's this— also the sort
3217
03:28:07,505 --> 03:28:10,155
of the crime is being, you know, three-card monte.
3218
03:28:11,171 --> 03:28:14,639
It's supposed to be smuggling chips. All of a sudden, she's smuggling
3219
03:28:14,928 --> 03:28:18,155
information to you know like like I
3220
03:28:18,655 --> 03:28:20,724
mean I guess it could be both, but like which is it? You know like
3221
03:28:21,224 --> 03:28:24,529
it's these two things are quite distinct. There's there
3222
03:28:24,706 --> 03:28:28,302
is there is a definite major issue with the potential
3223
03:28:28,398 --> 03:28:31,770
to be blackmailed, right? If you are engaged in
3224
03:28:31,850 --> 03:28:35,334
an illicit affair that you're trying to keep secret, and this person
3225
03:28:35,834 --> 03:28:38,888
that's a completely different. Well, I'm just, I'm just saying this person, you know,
3226
03:28:39,049 --> 03:28:43,291
Jensen Huang is a very powerful person with possibly clearance to
3227
03:28:43,323 --> 03:28:47,440
things. And so, uh, you know, uh, monitoring his activity
3228
03:28:47,940 --> 03:28:51,676
to make sure that he's not, uh, someone who can be compromised through blackmail
3229
03:28:52,372 --> 03:28:56,093
is a— I find it fascinating though that this has not been reported
3230
03:28:56,593 --> 03:29:00,432
that he has clearance. Um, and I feel like it would have been.
3231
03:29:00,561 --> 03:29:03,928
I, I, I, I'm— I've been on the lookout. I am making an
3232
03:29:03,960 --> 03:29:07,189
assumption. I am making an assumption. just because a lot of
3233
03:29:07,689 --> 03:29:10,515
these people wind up having clearance for things. So I am just making an assumption,
3234
03:29:11,045 --> 03:29:13,825
but I'm following what you're— All I'm saying is that these are 3 different things,
3235
03:29:14,001 --> 03:29:17,311
right? Like somebody is getting close to you in order
3236
03:29:17,327 --> 03:29:20,942
to get you to sell them product so that they can, you know,
3237
03:29:21,007 --> 03:29:25,152
without perhaps without applying the full suite of checks in
3238
03:29:25,249 --> 03:29:27,723
order to smuggle it to China is one thing.
3239
03:29:28,944 --> 03:29:32,655
Somebody getting close to you in order to get information about what the
3240
03:29:33,155 --> 03:29:36,573
next generation, whatever, chips is another.
3241
03:29:36,686 --> 03:29:40,492
Somebody getting in touch, you know, honeypotting you basically so you can be blackmailed
3242
03:29:40,909 --> 03:29:43,655
about an affair is a third. Like,
3243
03:29:44,281 --> 03:29:46,979
and I understand, like, you know, in the world of espionage it could be all
3244
03:29:47,027 --> 03:29:50,319
3, who knows? But like, I'm noticing that, like, okay, what,
3245
03:29:50,560 --> 03:29:54,639
like, what is it though? At this point it's all like,
3246
03:29:54,880 --> 03:29:58,493
you know. We are given, we are given reasons for pause but
3247
03:29:58,509 --> 03:30:01,326
not given anything specifically to hook on.
3248
03:30:01,861 --> 03:30:05,422
Right. Now, I will— Gator, when you
3249
03:30:05,922 --> 03:30:09,452
know that she was under surveillance for over a
3250
03:30:09,468 --> 03:30:11,897
year with the— They didn't have anything specific.
3251
03:30:13,917 --> 03:30:17,008
I mean, again, that's a dog that didn't bark. All right. So you've done that.
3252
03:30:17,072 --> 03:30:20,244
And now we know that the US intelligence services are
3253
03:30:20,744 --> 03:30:23,836
not sure what her connection is and whatever. I guess we'll get to that.
3254
03:30:24,403 --> 03:30:27,514
But it's like, okay, so I mean, if you're not sure by now,
3255
03:30:28,049 --> 03:30:31,228
what are you You need an X-ray. Like, what's it gonna take?
3256
03:30:31,374 --> 03:30:34,332
Let me, let me give you, uh, let me give you something here, Alex.
3257
03:30:34,752 --> 03:30:37,389
Uh, you know who didn't blackmail Jensen Huang?
3258
03:30:39,283 --> 03:30:43,280
Who? Claude. None of the large language models
3259
03:30:43,912 --> 03:30:47,704
blackmailed Jensen Huang. Can you believe it?
3260
03:30:48,817 --> 03:30:52,010
I was told that there'd be rampant LLM blackmail by now.
3261
03:30:55,376 --> 03:30:58,985
Didn't happen. Yep. Didn't happen. So you learn that
3262
03:30:59,485 --> 03:31:03,271
the US intelligence community has for the past year or so been surveilling
3263
03:31:03,287 --> 03:31:06,065
this woman that they suspect is a Chinese spy.
3264
03:31:06,514 --> 03:31:10,367
They know this woman is talking quite a bit with Nvidia's
3265
03:31:10,867 --> 03:31:14,718
CEO, Jensen Huang. What do they come to know about
3266
03:31:15,007 --> 03:31:17,479
the nature of the relationship between these two?
3267
03:31:18,459 --> 03:31:22,148
They know that they have a very close relationship. that they've
3268
03:31:22,648 --> 03:31:26,107
obviously been doing business together, and that some people in the
3269
03:31:26,171 --> 03:31:29,297
intel community suspect that the relationship is intimate. Hmm.
3270
03:31:29,842 --> 03:31:33,929
And this obviously sets off alarm bells within the
3271
03:31:34,429 --> 03:31:38,048
US government because— Right, you'd think. Jensen Huang, over the course
3272
03:31:38,112 --> 03:31:41,590
of the past 2 years, has gone from somebody who did not
3273
03:31:41,783 --> 03:31:45,165
even know President Trump to one of President
3274
03:31:45,665 --> 03:31:50,037
Trump's closest allies, who has a huge voice and influence over over US
3275
03:31:50,537 --> 03:31:53,803
AI
policy and who has regular phone calls with
3276
03:31:54,303 --> 03:31:57,873
the president. Right. Um, they're in touch all the time. I mean, we saw recently
3277
03:31:58,946 --> 03:32:02,215
Jensen Huang was on stage at an event and he got a call from the
3278
03:32:02,715 --> 03:32:05,965
president and, and took it live. So they're clearly very close.
3279
03:32:06,542 --> 03:32:09,746
And I assume that the worry or the
3280
03:32:10,246 --> 03:32:13,656
question on the minds of these intelligence officers is, what is
3281
03:32:13,993 --> 03:32:17,092
Alice Huang trying to get Out of Jensen,
3282
03:32:17,768 --> 03:32:20,683
and what has she managed to glean from him?
3283
03:32:21,230 --> 03:32:25,160
Right, and the U.S. intelligence community determines that
3284
03:32:25,660 --> 03:32:28,456
she's working on behalf of China's Ministry of State Security,
3285
03:32:29,820 --> 03:32:31,498
which is essentially their CIA. Wow!
3286
03:32:33,629 --> 03:32:37,164
And they determine that she is trying
3287
03:32:37,261 --> 03:32:40,982
to learn more about American AI chips. In other words, you know,
3288
03:32:41,144 --> 03:32:45,116
Nvidia's magic. And potentially trying to
3289
03:32:45,196 --> 03:32:48,994
influence U.S. policy on AI
3290
03:32:49,042 --> 03:32:53,048
chips for China. In other words, encouraging Jensen to
3291
03:32:53,548 --> 03:32:56,766
try to advocate for China to get more AI chips.
3292
03:32:57,839 --> 03:33:01,428
And the concern and the worry here is she's incredibly close to him, which means
3293
03:33:01,605 --> 03:33:05,354
she could have access to his devices. She could be communicating
3294
03:33:05,386 --> 03:33:08,639
with him at all times. She could be weighing in on issues.
3295
03:33:09,048 --> 03:33:12,495
in real time? We don't know. We haven't seen the surveillance. But you
3296
03:33:12,511 --> 03:33:15,757
can imagine the type of questions that
3297
03:33:15,821 --> 03:33:19,549
the intel community was agonizing over as it was reviewing
3298
03:33:19,597 --> 03:33:23,035
the surveillance. And so what do they do when they find
3299
03:33:23,535 --> 03:33:26,908
this out? Like, how do they respond to this information? They take
3300
03:33:27,408 --> 03:33:30,924
it to the top. They briefed the president on this about a year ago,
3301
03:33:31,471 --> 03:33:34,748
and they let him know that Jensen Huang
3302
03:33:34,861 --> 03:33:38,159
has Has a relationship, a close relationship with this
3303
03:33:38,659 --> 03:33:40,859
woman who they've determined to be a Chinese spy.
3304
03:33:44,070 --> 03:33:47,113
Now, Alex, we're going to stop that
3305
03:33:47,483 --> 03:33:50,929
replay from, or that retread from the New York
3306
03:33:51,429 --> 03:33:54,103
Times before we continue because we got a lot to do and I'm trying to
3307
03:33:54,603 --> 03:33:58,207
be very thorough with this here. Okay? Yeah. Just so it's clear, this guy just
3308
03:33:58,707 --> 03:34:01,649
lied there, right? Because— That's where I'm heading.
3309
03:34:01,806 --> 03:34:06,164
Or so they are, they are, they are pushing that
3310
03:34:06,678 --> 03:34:10,579
they are blurring the lines, let's say, between an analyst
3311
03:34:10,691 --> 03:34:13,902
who is giving them information about things that may have actually happened,
3312
03:34:13,982 --> 03:34:17,434
like this person was, was given particular information, you know, some information or
3313
03:34:17,934 --> 03:34:21,447
a briefing or something that contains something. And then they are also then saying it
3314
03:34:21,947 --> 03:34:24,273
is the intelligence community's conclusion,
3315
03:34:25,525 --> 03:34:30,614
as I was saying earlier, the unanimous sort of thing that it is their official
3316
03:34:31,114 --> 03:34:34,442
intelligence. They've determined her to be, you know, this is a determination that has been
3317
03:34:34,942 --> 03:34:38,909
taken. CNN. U.S. intel isn't sure if businesswoman
3318
03:34:39,409 --> 03:34:42,947
spied on Nvidia for China doesn't mean she wasn't a risk, sources say. So this
3319
03:34:43,447 --> 03:34:47,166
is already contradictory. I love doesn't means.
3320
03:34:48,977 --> 03:34:52,745
U.S. intelligence community is still working to understand the relationship between a Chinese
3321
03:34:52,778 --> 03:34:55,899
businesswoman who had an affair with Nvidia CEO Jensen Huang and the Chinese
3322
03:34:56,160 --> 03:34:59,815
government, with multiple sources telling CNN that the intelligence collected
3323
03:34:59,863 --> 03:35:02,909
on her to date does not prove she was acting as a spy for Beijing.
3324
03:35:03,410 --> 03:35:06,879
So that contradicts what the New York Times was just saying, which is basically,
3325
03:35:07,073 --> 03:35:10,833
or at least significantly implying,
3326
03:35:12,750 --> 03:35:16,766
is that it is a US government intelligence assessment
3327
03:35:16,863 --> 03:35:20,157
that she is for sure a spy. So this is the contradiction here. Look,
3328
03:35:20,237 --> 03:35:23,724
she's either been determined to be a spy or she hasn't. And by the way,
3329
03:35:24,047 --> 03:35:27,851
there is a distinct— like, a spy is a
3330
03:35:28,175 --> 03:35:30,343
direct member of the, you know,
3331
03:35:30,505 --> 03:35:33,854
organizational hierarchy of the intelligence services, right? We're not.
3332
03:35:34,242 --> 03:35:38,024
An asset is a different thing, right? It's somebody who's been recruited. Sure.
3333
03:35:38,105 --> 03:35:41,806
Um, knowingly or unknowingly, you know, whatever. Um,
3334
03:35:42,130 --> 03:35:45,052
you know, it's, it's a far looser term, but they're using— like,
3335
03:35:45,149 --> 03:35:48,526
they're throwing whatever, you know. It's, it's an— it's a spy,
3336
03:35:48,802 --> 03:35:52,459
it's an It's the asset, it's the smuggling, it's information,
3337
03:35:52,941 --> 03:35:56,106
whatever, you know, it's— she's Chinese, you know, she must have something going.
3338
03:35:56,155 -->
03:35:59,368
There is disagreement among U.S. officials about the strength of
3339
03:35:59,868 --> 03:36:03,047
evidence that Huang was recruited by the Chinese Ministry of State Security,
3340
03:36:03,481 --> 03:36:07,369
the Chinese equivalent of the CIA, to spy on Jensen Huang and by extension
3341
03:36:07,803 --> 03:36:11,659
on U.S. policymakers. The FBI has been investigating Huang's affair for
3342
03:36:11,675 --> 03:36:15,226
at least a year, but Attorney General Todd Blanche said on Thursday there are
3343
03:36:15,726 --> 03:36:19,106
still no conclusions reached by the Department of Justice or by me or by the
3344
03:36:19,122 --> 03:36:22,393
FBI Director Patel around who she is or is
3345
03:36:22,505 --> 03:36:26,096
not. Some officials at the Office of the Director of National Intelligence,
3346
03:36:26,112 --> 03:36:29,703
including those close to Jay Clayton, have framed the characterization of
3347
03:36:30,203 --> 03:36:33,967
Alice Huang as a Chinese spy as not entirely accurate or overtorqued,
3348
03:36:34,063 --> 03:36:37,718
sources said. And officials at the CIA are also not yet convinced
3349
03:36:38,218 --> 03:36:41,389
that the intelligence collection on Huang to date, including intercepted communications,
3350
03:36:41,570 --> 03:36:45,163
communications proves she knowingly funneled information back to Beijing,
3351
03:36:45,339 --> 03:36:48,550
a source told CNN. Still, some Trump
3352
03:36:48,567 --> 03:36:51,874
officials told CNN it strains credulity to believe she
3353
03:36:52,374 --> 03:36:56,595
was not acting as an agent of the Chinese government. Now, strains credulity is a
3354
03:36:57,095 --> 03:36:59,694
statement, but that only really is,
3355
03:37:00,432 --> 03:37:03,772
you know, benefited by naming the person who is claiming that
3356
03:37:03,804 --> 03:37:07,689
it strains credulity, you know, like Some
3357
03:37:07,753 --> 03:37:11,491
people, if they say if it strains credulity, you know, have more credence.
3358
03:37:11,539 --> 03:37:14,426
I mean, do you have evidence or don't you? I mean,
3359
03:37:14,554 --> 03:37:17,586
if you do, yes, it does strain credulity. If you don't, I mean,
3360
03:37:17,698 --> 03:37:20,842
I don't know, like, are we doing our jobs here? Are we collecting intelligence?
3361
03:37:21,082 --> 03:37:23,633
Are we just saying, well, you know, she's Chinese, look at her?
3362
03:37:24,371 --> 03:37:28,332
Simply by the virtue of the fact that she's a successful Chinese businesswoman
3363
03:37:28,685 --> 03:37:31,059
working with one of America's most powerful companies.
3364
03:37:32,038 --> 03:37:35,968
Under Chinese national security law, Chinese citizens can be compelled to to help gather
3365
03:37:35,984 --> 03:37:39,708
intelligence on behalf of the Chinese Communist Party, even if she had not
3366
03:37:39,756 --> 03:37:42,886
done so yet. Giving her access to Huang and by extension to
3367
03:37:43,386 --> 03:37:46,722
President Donald Trump, some officials believe she posed a clear counterintelligence
3368
03:37:47,222 --> 03:37:50,622
risk. Now, look, that claim can certainly be
3369
03:37:50,814 --> 03:37:54,137
some validity, particularly if she has family back
3370
03:37:54,362 --> 03:37:57,459
in China. There have been plenty of stories where China has
3371
03:37:57,636 --> 03:38:01,504
used people's family to threaten people
3372
03:38:02,004 --> 03:38:05,341
in order to use them as assets and gain information. And cooperation and send
3373
03:38:05,841 --> 03:38:10,465
them on tasks, etc., etc. Some officials believe—
3374
03:38:10,577 --> 03:38:13,613
and if that, if that's the standard we're playing with, then, you know,
3375
03:38:13,693 --> 03:38:17,853
U.S. executives should be banned from having any, you know, a close connection
3376
03:38:17,901 --> 03:38:20,824
to any, any, like, like literally 1/6 of humanity.
3377
03:38:21,419 --> 03:38:24,776
Yeah. But to others, the fixation on Huang's relationship to the
3378
03:38:25,276 --> 03:38:28,502
Chinese government has obscured broader questions around the U.S. government's failure to crack down on
3379
03:38:29,002 --> 03:38:32,369
the diversion to China sophisticated technology vital to the AI
3380
03:38:32,417 --> 03:38:36,169
race. Alice Huang's Singapore-based Megaspeed is a
3381
03:38:36,669 --> 03:38:40,194
major purchaser of Nvidia chips. The Department of Justice and the Commerce Department
3382
03:38:40,694 --> 03:38:44,396
began an investigation into the firm in 2024, part of a broader probe into
3383
03:38:44,444 --> 03:38:48,661
an ecosystem of foreign shell companies suspected of diverting restricted
3384
03:38:48,677 --> 03:38:52,029
Nvidia chips to China, a person briefed on the investigation said.
3385
03:38:52,702 --> 03:38:55,989
There are now questions brewing inside the White House over why the Commerce
3386
03:38:56,489 --> 03:38:59,617
Department did not put Megaspeed on the so-called entity list, which,
3387
03:39:00,067 --> 03:39:03,953
which restricts trade with companies that may pose a risk
3388
03:39:04,453 --> 03:39:07,775
to US national security or foreign policy interests, at least
3389
03:39:08,275 --> 03:39:10,794
while investigation was ongoing, according to a person familiar.
3390
03:39:11,886 --> 03:39:15,740
Quote, an intelligence operation wasn't necessary to achieve that result, a former
3391
03:39:16,240 --> 03:39:19,642
Commerce Department official told CNN, noting that Jensen Huang has publicly
3392
03:39:19,658 --> 03:39:22,805
advocated for the US to sell more chips to China. The fact that chips
3393
03:39:23,305 --> 03:39:27,024
might be diverted to China through Megaspeed was theoretically a place where China's interests already
3394
03:39:27,040 --> 03:39:30,711
coincided with Nvidia's and regulatory— a regulatory
3395
03:39:30,807 --> 03:39:33,484
loophole already allowed it, the former Commerce official added.
3396
03:39:34,575 --> 03:39:37,732
Jensen Huang resigned from Nvidia
on Wednesday following
3397
03:39:37,797 --> 03:39:40,922
days of reports that he had an affair with a suspected Chinese intelligence
3398
03:39:41,067 --> 03:39:44,818
asset. Reports also emerged on Wednesday night that he had separately had
3399
03:39:44,834 --> 03:39:47,094
an affair with a subordinate at Nvidia.
3400
03:39:48,697 --> 03:39:52,416
Democratic Representative Jim Himes, the ranking member of the House
3401
03:39:52,528 --> 03:39:56,263
Intelligence Committee, told Politico Thursday that he left a classified briefing
3402
03:39:56,763 --> 03:40:00,529
with senior intelligence officials earlier this week, unsure of the nature of Huang's
3403
03:40:01,029 --> 03:40:04,828
relationship with the Chinese government. That would be Alice
3404
03:40:05,328 --> 03:40:09,246
Huang, not Jonathan Huang. So this is important because this
3405
03:40:09,375 --> 03:40:13,375
is the incoming Democrat person
3406
03:40:13,875 --> 03:40:17,645
in that committee, right? So now you can't
3407
03:40:17,677 --> 03:40:20,964
say, oh, this is Trump officials closing ranks
3408
03:40:21,464 --> 03:40:25,122
around the president trying to, you know, protect his close,
3409
03:40:25,395 --> 03:40:29,022
uh, confidant or whatever. Like, this is somebody who is motivated to
3410
03:40:29,522 --> 03:40:32,696
do the exact opposite and is coming out and saying, look, you know,
3411
03:40:32,728 --> 03:40:35,905
I've been briefed, so ergo, you know, I know, you know,
3412
03:40:36,628 --> 03:40:40,334
the key bits of what they know. And I,
3413
03:40:40,799 --> 03:40:44,233
you know, I haven't heard something to tell me that she's
3414
03:40:44,282 --> 03:40:47,988
a spy, even after, I, I add, a year
3415
03:40:48,181 --> 03:40:51,612
of surveillance of this person. Yeah. Guys.
3416
03:40:52,189 --> 03:40:55,186
What's the relationship with Alice Wong and the Chinese, and what do we know about
3417
03:40:55,410 --> 03:40:58,823
what influence or information was traded? Because today, at least, I don't know,
3418
03:40:59,080 --> 03:41:02,589
Himes said. Last year, when US intelligence officials
3419
03:41:02,621 --> 03:41:04,977
first learned of the relationship between Huang and Alice Wong,
3420
03:41:06,147 --> 03:41:09,865
officials across the government immediately wanted to know more, and some began sounding
3421
03:41:09,913 --> 03:41:13,743
the alarm. Secretary of State and National Security Advisor Marco
3422
03:41:13,759 --> 03:41:17,508
Rubio warned in a memo in the spring about
3423
03:41:18,008 --> 03:41:21,060
Jensen's relationship with Huang, one source told CNN. But it's unclear how detailed the memo
3424
03:41:21,560 --> 03:41:25,201
got on how much information Huang was or was not sharing with the Chinese
3425
03:41:25,234 --> 03:41:28,797
government. Rubio briefed Trump on Jensen Huang and his relationship
3426
03:41:28,830 --> 03:41:32,795
with Alice Huang aboard Air Force One en route to China in May.
3427
03:41:33,501 --> 03:41:36,792
The briefing occurred before Jensen Huang joined Trump in Alaska for the
3428
03:41:37,292 --> 03:41:40,854
final leg of a trip, a source briefed on the matter told CNN. Different agencies
3429
03:41:40,886 --> 03:41:44,113
have continued conducting their own reviews of the intelligence,
3430
03:41:44,161 --> 03:41:47,483
sources said. One particularly alarming Charming development occurred when U.S.
3431
03:41:47,547 --> 03:41:51,447
intelligence obtained communications that appeared to indicate someone with close access to Jensen
3432
03:41:51,463 --> 03:41:55,507
Huang and Nvidia was the source of information that ended up in Chinese intelligence
3433
03:41:55,635 --> 03:41:59,070
streams. A person familiar with the investigation told CNN.
3434
03:42:00,065 --> 03:42:04,044
The FBI, which is conducting the primary counterintelligence investigation,
3435
03:42:04,381 --> 03:42:08,137
told lawmakers last year that it was particularly suspicious of Alice Huang's seemingly
3436
03:42:08,313 --> 03:42:11,924
overnight career pivot from Chinese state television journalist
3437
03:42:12,424 --> 03:42:16,078
to businesswoman. Woman with an unusually large amount of purchasing power in the Chinese
3438
03:42:16,578 --> 03:42:19,976
economy, one source said. That, paired with her close relationship
3439
03:42:20,476 --> 03:42:24,568
to the then-Nvidia CEO, has continued to raise red flags inside the bureau,
3440
03:42:24,907 --> 03:42:28,206
where the counterintelligence probe remains ongoing, according to multiple sources
3441
03:42:28,706 --> 03:42:32,581
familiar with the matter. Definitively linking suspicious behavior
3442
03:42:33,081 --> 03:42:36,671
by Chinese nationals interacting with the United States to a state-directed spying campaign
3443
03:42:36,769 --> 03:42:41,199
remains a perennial problem for counterintelligence
3444
03:42:41,699 --> 03:42:44,782
Officials trying to defend against Chinese espionage is notoriously difficult to collect intelligence on China,
3445
03:42:44,798 --> 03:42:47,939
and China routinely compels its private citizens to do its bidding overseas.
3446
03:42:48,547 --> 03:42:51,591
When it comes to Chinese spying, the threat is in the eye of the beholder.
3447
03:42:51,912 --> 03:42:55,228
Some officials think the fears are overblown; other officials see a Chinese
3448
03:42:55,728 --> 03:42:59,346
spy in every encounter. A lot of people are guilty of just being Chinese,"
3449
03:42:59,458 --> 03:43:02,293
said a former senior counterintelligence official.
3450
03:43:03,046 --> 03:43:06,956
But Joshua Hodges, a commissioner on the U.S.-China Economic and Security
3451
03:43:06,972 --> 03:43:10,753
Review Commission, which investigates how U.S.-China trade. and Economic Ties
3452
03:43:11,253 --> 03:43:15,032
Affect American National Security, said that, quote, we've seen repeatedly that Chinese
3453
03:43:15,081 --> 03:43:18,174
are incredibly adept at using people both wittingly and unwittingly.
3454
03:43:18,751 -->
03:43:22,742
Quote, ultimately what we typically see— Is this commission a US government body
3455
03:43:22,838 --> 03:43:26,669
or just an anti-China think tank that is trying
3456
03:43:26,781 --> 03:43:29,730
to sound like one? That's a good question. I don't know that answer, sir.
3457
03:43:29,810 --> 03:43:33,208
Maybe while I'm reading the rest, you can perform a search.
3458
03:43:33,288 --> 03:43:36,975
Tell me their name again. US-China Economic
3459
03:43:37,103 --> 03:43:38,353
and Security Review Commission.
3460
03:43:41,500 --> 03:43:45,037
Okay. Ultimately, what we typically see is those people at a certain
3461
03:43:45,537 --> 03:43:48,809
point in time recognize that they know who they're working for
3462
03:43:49,309 --> 03:43:51,676
and they know the ends that they're working towards, he told CNN.
3463
03:43:52,752 --> 03:43:55,966
Hodges, who served on the National Security Council in Trump's first
3464
03:43:56,466 --> 03:44:00,080
term, noted Hwang's past rhetoric denouncing the phrase China hawks as
3465
03:44:00,096 --> 03:44:03,466
a badge of shame. That clearly is directed to make people stop
3466
03:44:03,966 --> 03:44:07,143
talking about the topic, Hodges said. You've got the biggest, one of the biggest CEOs
3467
03:44:07,176 --> 03:44:10,865
in the world attacking people for speaking out about something they care about that clearly
3468
03:44:10,962 --> 03:44:14,310
only benefits one country. So before I continue,
3469
03:44:14,439 --> 03:44:17,694
Alex, or did you find
3470
03:44:18,194 --> 03:44:20,176
if that was a think tank? Now, it said that he served in Trump's first
3471
03:44:20,676 --> 03:44:23,140
term, so I'll assume he's not currently working in government, which makes me think it's
3472
03:44:23,640 --> 03:44:27,174
a think tank. Yeah, I'm still looking. I'll tell you when I know.
3473
03:44:27,644 --> 03:44:31,193
All right. But I mean, the way he's talking is very specific,
3474
03:44:31,209 --> 03:44:34,069
which Oh, it is part of the U.S.
3475
03:44:34,118 --> 03:44:37,639
federal government. Oh, okay. One of
3476
03:44:37,671 --> 03:44:38,612
12 commissions.
3477
03:44:40,797 --> 03:44:45,169
Its 12 commissioners are selected through House
3478
03:44:45,476 --> 03:44:48,952
and Senate leadership, and it reports to Congress on the national security implications,
3479
03:44:48,968 --> 03:44:51,744
blah, blah, blah. Congress funds it through appropriations.
3480
03:44:53,437 --> 03:44:57,126
Okay, okay, so it's subservient
3481
03:44:57,158 --> 03:45:00,340
to Congress, but he's not a congressman. Or anything like that?
3482
03:45:00,550 --> 03:45:03,779
Is that the— No. Okay.
3483
03:45:04,796 --> 03:45:08,170
Well, Alex, this is gonna sound like a tangent
3484
03:45:08,302 --> 03:45:10,431
or an aside, but it does serve a purpose.
3485
03:45:12,815 --> 03:45:16,716
So you know how there's a question about whether or not
3486
03:45:16,912 --> 03:45:20,261
God exists and people are regularly looking
for
3487
03:45:20,326 --> 03:45:23,734
evidence for or against the
3488
03:45:24,234 --> 03:45:27,178
existence of God? Sorry, say it again?
3489
03:45:28,661 --> 03:45:33,918
You know how the question about whether or not there is a God is ubiquitous
3490
03:45:33,983 --> 03:45:37,318
amongst humans, and that we're constantly looking for evidence to see
3491
03:45:37,726 --> 03:45:41,185
if we can prove or disprove the existence of God? Yes,
3492
03:45:41,378 --> 03:45:44,945
as I'm aware. All right, well, one bit of evidence,
3493
03:45:45,009 --> 03:45:49,652
Alex, that I think might be slightly dispositive,
3494
03:45:49,668 --> 03:45:52,336
something that we can say is hinting towards dispositivity,
3495
03:45:52,746 --> 03:45:56,636
against the existence of God is that Hillary Rodham Clinton and
3496
03:45:57,136 --> 03:46:02,278
Kara Swisher now have a podcast. And oh God.
3497
03:46:05,931 --> 03:46:09,642
And maybe God wants us to bring us closer to him
3498
03:46:09,692 --> 03:46:11,254
and his name.
3499
03:46:13,563 --> 03:46:16,598
High stakes with Hillary Clinton and Kara Swisher.
3500
03:46:17,680 --> 03:46:21,102
And good. They, they, uh, brought up this issue this
3501
03:46:21,602 --> 03:46:25,309
week. But let's get into the news. There's a huge story broke during the
3502
03:46:25,809 --> 03:46:28,199
time, and it's something I took very seriously right away.
3503
03:46:28,552 --> 03:46:32,133
Nvidia's board is investigating CEO Jensen Huang after his admission
3504
03:46:32,583 --> 03:46:36,083
that he had an affair with a woman who U.S. intelligence believes was a Chinese
3505
03:46:36,115 --> 03:46:39,889
spy. She was reportedly recruited by China's top spy agency
3506
03:46:40,389 --> 03:46:43,726
to get Huang to push Washington to let Nvidia sell chips to China, and that's
3507
03:46:44,226 --> 03:46:47,802
exactly what he's been pushing for. Still, Nvidia NVIDIA's board says it has
3508
03:46:47,979 --> 03:46:51,682
full confidence in him. You've dealt with Chinese intelligence operations as Secretary
3509
03:46:52,182 --> 03:46:55,978
of State. NVIDIA is worth trillions of dollars, and Trump wants those
3510
03:46:56,074 --> 03:46:59,937
chip sales. But does the spy scandal change US-China
3511
03:47:00,437 --> 03:47:03,704
policy? I'd love just your take on the whole thing, and I'll talk about the
3512
03:47:04,204 --> 03:47:07,584
tech part. Yeah, well, this is a huge
3513
03:47:07,616 --> 03:47:11,655
story, and it says a lot
3514
03:47:11,864 --> 03:47:16,415
about how the Trump administration deals
3515
03:47:16,671 --> 03:47:19,139
with national security threats.
3516
03:47:20,277 --> 03:47:23,914
It doesn't appear from everything that I've been able to read up until
3517
03:47:24,414 --> 03:47:28,112
now that they did what is called a defensive briefing,
3518
03:47:28,689 --> 03:47:32,455
where they actually went to Jensen
3519
03:47:32,955 --> 03:47:36,140
Huang and said, we believe based on intelligence—
3520
03:47:36,640 --> 03:47:40,708
and remember, this is not people sitting around speculating, this is hard What are
3521
03:47:41,208 --> 03:47:44,460
called what's called signal intelligence intercepted calls, emails,
3522
03:47:45,118 --> 03:47:48,822
and they should have gone to him and
3523
03:47:49,322 --> 03:47:53,200
said, "We believe that this woman Alice Wong is
3524
03:47:53,537 --> 03:47:57,209
a Chinese spy, that she is part of the Ministry
3525
03:47:57,482 --> 03:48:00,737
of State Security in China, and she is
3526
03:48:01,330 --> 03:48:05,243
extracting information from you." This is not
3527
03:48:05,743 --> 03:48:09,669
a you know a moral issue about his personal relationships;
3528
03:48:09,749 --> 03:48:13,264
it's a national security. Right, right. But as far as I know,
3529
03:48:13,280 --> 03:48:16,540
they never went to him. No. Yeah, that's that's what Trump said.
3530
03:48:17,198 --> 03:48:21,485
But when I was running for president, Kara, I got a defensive briefing about
3531
03:48:21,550 --> 03:48:26,752
Russian interference. Not that I could talk about it, which sadly kind
3532
03:48:27,252 --> 03:48:30,365
of you know kept me from explaining a lot of what was going on in
3533
03:48:30,381 --> 03:48:33,993
that campaign. But you know the intelligence community came to me.
3534
03:48:34,058 --> 03:48:36,579
I also know they gave a defensive briefing to. Trump.
3535
03:48:37,350 --> 03:48:41,140
And not that he cared, because he wanted the help from Russia.
3536
03:48:41,542 --> 03:48:45,188
So if your government knows that an American citizen
3537
03:48:45,862 --> 03:48:50,247
is a target of an espionage operation,
3538
03:48:51,162 --> 03:48:54,519
and you have reason to believe he does not know that,
3539
03:48:54,551 --> 03:48:59,097
that he has fallen into what's called, you know, a honeypot operation
3540
03:48:59,193 --> 03:49:03,415
where a beautiful woman, you know, seduces the then
3541
03:49:03,784 --> 03:49:06,750
you owe it to him to tell him. They didn't tell him. And now we
3542
03:49:07,250 --> 03:49:12,008
know not only did they not tell him, they actively encouraged
3543
03:49:12,184 --> 03:49:15,903
the continuing sale of the Nvidia chips.
3544
03:49:16,111 --> 03:49:18,580
As you say, it's not a huge amount of their business.
3545
03:49:19,494 --> 03:49:23,052
It's in the single-digit percentage, but it is enough
3546
03:49:23,181 --> 03:49:27,236
to give a real advantage to the Chinese who had
3547
03:49:27,268 --> 03:49:31,051
been told in the Biden administration they couldn't have access to these
3548
03:49:31,099 --> 03:49:34,370
chips. Let me ask a specific question. So these are these defensive
3549
03:49:34,870 --> 03:49:37,693
hearings that you have. Trump was saying yesterday,
3550
03:49:37,838 --> 03:49:40,375
off the top of his head, he said it didn't matter. That was one of
3551
03:49:40,875 --> 03:49:44,019
the things he seemed to say. The second thing is that we didn't
3552
03:49:44,519
--> 03:49:46,990
tell him. Tell me the ins and outs of this, because something— it looks like
3553
03:49:47,311 --> 03:49:51,148
Marco Rubio leaked all this stuff because there was a memo Rubio wrote
3554
03:49:51,582 --> 03:49:55,580
before Kwang Jensen Wong went to China,
3555
03:49:55,660 --> 03:49:59,530
and he wasn't on the plane, and then he was. Yeah, but that decision that
3556
03:49:59,835 --> 03:50:04,139
Rubio Rubio sent a memo talking about the seriousness
3557
03:50:04,267 --> 03:50:08,089
of this, contrary to either the White House or the Commerce Department.
3558
03:50:09,005 --> 03:50:12,827
Maybe because of Rubio's memo and whatever follow-up he did,
3559
03:50:13,646 --> 03:50:17,580
they initially left him off the plane on Trump's trip
3560
03:50:18,080 --> 03:50:21,900
to China. But as you know, Trump then intervened
3561
03:50:22,013 --> 03:50:24,646
and said, no, no, I want him to come, so he should fly and meet
3562
03:50:25,146 --> 03:50:26,991
me in Alaska. So I have no idea.
3563
03:50:28,039 --> 03:50:31,039
doubt that Trump was briefed on this.
3564
03:50:31,633 --> 03:50:34,685
He was told, we have intelligence. But remember,
3565
03:50:34,766 --> 03:50:38,203
Trump doesn't believe the American intelligence community if it in any
3566
03:50:38,316 --> 03:50:41,255
way contradicts what he wants to believe.
3567
03:50:41,753 --> 03:50:45,737
I will predict to you what will happen when he sees Xi Jinping
3568
03:50:46,010 --> 03:50:49,448
in, you know, a few weeks. He will say to Xi Jinping,
3569
03:50:49,560 --> 03:50:52,162
like he said to Putin, is this true?
3570
03:50:52,612 --> 03:50:56,000
And Xi Jinping will say, of course it's not, Donald. You know,
3571
03:50:56,208 --> 03:50:59,592
that we wouldn't do that to you. And then Trump will come out and he
3572
03:51:00,092 --> 03:51:03,505
will say, I asked President Xi if it were true and he assured me it
3573
03:51:03,601 --> 03:51:07,915
wasn't. That's right. You can't make it up, right? This is a serious
3574
03:51:07,931 --> 03:51:11,219
breach of national security, which is made even worse by what
3575
03:51:11,719 --> 03:51:15,228
we have found out in the last 48 hours, which is that the Nvidia
3576
03:51:15,292 --> 03:51:19,125
general counsel intervened with the Commerce Department, which was actually trying
3577
03:51:19,238 --> 03:51:22,620
to investigate that these were Phony companies that had
3578
03:51:23,120 --> 03:51:26,568
been set up as pass-throughs to the Chinese government.
3579
03:51:26,908 --> 03:51:30,319
So, so a couple— All right, Alex, so there,
3580
03:51:30,528 --> 03:51:33,998
there's, uh, some of what they were discussing. And I, I thought it important
3581
03:51:34,498 --> 03:51:38,143
simply because this is, uh, somewhat the propagation of,
3582
03:51:38,336 --> 03:51:42,016
of more certainty than I think we actually have being broadcast
3583
03:51:42,129 --> 03:51:45,852
to large, large audiences that now are part
3584
03:51:45,917
--> 03:51:49,881
of the coalition question of both House and Senate going
3585
03:51:50,381 --> 03:51:54,526
to be pushing more of an anti-AI message writ
3586
03:51:54,623 --> 03:51:58,300
large. Well, so, no, no, with the— I think you missed
3587
03:51:58,800 --> 03:52:02,467
the smoking gun, like the big revelation
3588
03:52:02,967 --> 03:52:06,022
here. I was about to make this argument. I didn't know that Hillary stepped in
3589
03:52:06,055 --> 03:52:09,091
to defend Jensen. All right.
3590
03:52:09,367 --> 03:52:13,489
But for example, if a US congressperson and perhaps
3591
03:52:13,989 --> 03:52:17,621
member of the Intelligence committees,
3592
03:52:17,767 --> 03:52:22,288
Eric Swalwell is having
3593
03:52:22,482 --> 03:52:26,111
contact with a potential Chinese asset,
3594
03:52:26,208 --> 03:52:28,843
spy, call her what you will, Fang Fang.
3595
03:52:31,770 --> 03:52:35,149
What happens is exactly what— what happened is exactly
3596
03:52:35,165 --> 03:52:38,557
what she described, which is they take you aside, they give you a defensive briefing.
3597
03:52:38,995 --> 03:52:42,385
Hey dude, do you know who you're Do
3598
03:52:42,885 --> 03:52:45,949
you know that she's this and that? And they give you an opportunity to,
3599
03:52:46,977 --> 03:52:49,819
uh, you know, make it right,
3600
03:52:51,152 --> 03:52:54,379
as in to, you know, stop the
3601
03:52:54,879 --> 03:52:58,393
bleeding. And then you can continue to have your position in
3602
03:52:58,893 --> 03:53:02,552
your committees, uh, and, and so on and so forth. That's how it works,
3603
03:53:03,050 --> 03:53:05,908
right? The fact that they didn't do that, um,
3604
03:53:07,627 --> 03:53:11,181
means that it could mean 2 things, you know. If— well,
3605
03:53:11,246 --> 03:53:14,546
it could mean 3 things. A, there was never such a conclusion to begin with,
3606
03:53:15,351 --> 03:53:18,136
right? B, if there was a conclusion,
3607
03:53:18,909 --> 03:53:23,756
it could either mean that— B1, I suppose— they
3608
03:53:25,559 --> 03:53:26,865
didn't— did not want to tell him,
3609
03:53:27,495 --> 03:53:30,952
right? Or wait, what was—
3610
03:53:31,065 --> 03:53:34,634
I've lost track of my numbers. Oh, no worries. Short, either they were—
3611
03:53:34,939 --> 03:53:38,873
they weren't sure she was a spy, or they, um, wanted to,
3612
03:53:39,114 --> 03:53:42,277
uh, trap him, I guess, right? I mean, what else?
3613
03:53:42,325 --> 03:53:45,793
Okay, what else could it be for them to not do the default
3614
03:53:46,293 --> 03:53:50,289
thing that they do, especially when we're talking about— She's pinning it on Trump's specific
3615
03:53:50,353 --> 03:53:53,870
responsibility, that it is Trump's. You think
3616
03:53:54,370 --> 03:53:57,755
Trump is running around micromanaging who gets defensive briefings? Are you fucking kidding me?
3617
03:54:01,689 --> 03:54:06,955
Well, that's what— I understand what she wants to do, but procedurally
3618
03:54:07,455 --> 03:54:11,181
speaking, I mean, you think he has any idea who gets defensive briefings?
3619
03:54:11,246 --> 03:54:14,610
Are you kidding me? So that's
3620
03:54:15,110 --> 03:54:19,138
sort of where we stand, Alex. I want to— I have more
3621
03:54:19,750 --> 03:54:23,362
on this story, but I think we have enough for this that
3622
03:54:23,862 --> 03:54:27,298
we've covered this week on it that we could possibly get into the
3623
03:54:27,798 --> 03:54:29,900
rest of next week. Well, I think there's another interesting element I want to add,
3624
03:54:29,948 --> 03:54:33,288
which is that— so Alice Huang was
3625
03:54:33,561 --> 03:54:36,854
called to Singapore to be questioned about
3626
03:54:37,354 --> 03:54:40,901
this case, both by Commerce Department, the US Commerce
3627
03:54:41,401 --> 03:54:44,627
Department, and the Singaporean authorities. They then—
3628
03:54:44,964 --> 03:54:47,501
she went voluntarily. They questioned her,
3629
03:54:48,015 --> 03:54:52,641
and then they kept her passport, which means that she was not able
3630
03:54:53,141 --> 03:54:56,908
to leave Singapore for 8 months. Um, the Singaporean authorities kept asking for cooperation from,
3631
03:54:57,181 --> 03:54:59,865
uh, the Commerce Department. The Commerce Department did not respond,
3632
03:55:00,653 --> 03:55:03,866
and therefore they gave her her passport back and she left. Okay,
3633
03:55:03,948 --> 03:55:08,548
okay. So, all right, what is— what
3634
03:55:09,048 --> 03:55:09,312
does that sound like?
3635
03:55:13,749 --> 03:55:17,295
Uh, I mean Well,
3636
03:55:17,440 --> 03:55:19,817
she was literally— she's like, yeah, sure, what do you want to know? Right?
3637
03:55:19,898 --> 03:55:23,095
They're like, we got your passport, you can't leave. Fine,
3638
03:55:23,111 --> 03:55:26,469
I'll wait. And then she waits and nothing happens.
3639
03:55:29,310 --> 03:55:32,782
Well, presumably the Chinese can get a
3640
03:55:33,282 --> 03:55:36,351
person out of Singapore if they want to, if it's their spy, by the way.
3641
03:55:36,981 --> 03:55:39,725
Like, it's not a— There's one,
3642
03:55:39,757 --> 03:55:44,370
there's one additional sort of, um, hypothesis that
3643
03:55:44,578 --> 03:55:48,055
was sort of floating around that gained popularity this week.
3644
03:55:49,545 --> 03:55:53,550
And it was in a long Twitter thread by the
3645
03:55:54,050 --> 03:55:56,386
account @rnaudbertrand. Yes.
3646
03:55:57,315 --> 03:56:00,552
And it said, I'll read the thread here. This is the biggest
3647
03:56:00,600 --> 03:56:04,140
irony in the whole Alice Wong Nvidia saga. The Chinese government
3648
03:56:04,204 --> 03:56:07,409
actually doesn't want Nvidia chips and is actively keeping them out of the
3649
03:56:07,909 --> 03:56:11,751
country. Last year, it even ordered Alibaba and ByteDance to cancel their
3650
03:56:12,251 --> 03:56:15,346
Nvidia orders. Hang on, long link.
3651
03:56:16,069 --> 03:56:19,280
And sent customs teams to ports to
check every advanced
3652
03:56:19,780 --> 03:56:23,968
chip coming in. And right now, 600,000 Nvidia H200s
3653
03:56:24,113 --> 03:56:27,164
are stuck in Hong Kong because Beijing won't let them into the country.
3654
03:56:27,726 --> 03:56:31,226
Nvidia is also explicitly— Nvidia also explicitly frames
3655
03:56:31,322 --> 03:56:34,791
this as a threat to its business in its latest annual report to the SEC.
3656
03:56:35,272 --> 03:56:39,493
The Chinese government has encouraged customers to purchase from our China-based competitors
3657
03:56:39,846 --> 03:56:43,390
and discourage customers from purchasing, importing, or using our
3658
03:56:43,679 --> 03:56:47,464
data center products. Why does China do that? Because every Nvidia
3659
03:56:47,964 --> 03:56:52,163
chip a Chinese company buys is one that Huawei and
3660
03:56:52,275 --> 03:56:55,466
our Cabricorn doesn't sell, and domestic
3661
03:56:55,966 --> 03:56:59,219
chips can only catch up if people actually buy them, which means that the notion
3662
03:56:59,251 --> 03:57:02,571
that Chinese intelligence was running honey traps to get more Nvidia chips into
3663
03:57:02,619 --> 03:57:05,830
China Gets the whole thing exactly backwards. In fact,
3664
03:57:05,847 --> 03:57:08,947
it looks like MegaSpeed, Alice Huang's company, was a counter
3665
03:57:09,447 --> 03:57:13,044
to reaction to official Chinese policy, a private sector workaround to these efforts
3666
03:57:13,542 --> 03:57:17,141
by the Chinese government to curtail Nvidia. As was widely reported,
3667
03:57:17,157 --> 03:57:20,691
including by Western media, MegaSpeed built data centers in Southeast Asia
3668
03:57:21,191 --> 03:57:25,061
and rented their computing power to Alibaba, which the New York Times called a backdoor
3669
03:57:25,174 --> 03:57:28,579
way of getting access to Nvidia chips. That is a back
3670
03:57:28,885 --> 03:57:32,962
backdoor only, not allegedly around. Washington's export controls, but around China's
3671
03:57:33,123 --> 03:57:37,230
own efforts to wean its companies off Nvidia. I can't believe the Chinese government
3672
03:57:37,730 --> 03:57:40,021
was too thrilled about it. It would make just about zero sense for it to
3673
03:57:40,101 --> 03:57:43,904
run countless domestic measures to prevent Alibaba from using Nvidia chips,
3674
03:57:44,273 --> 03:57:47,674
yet at the same time have its top spy agency help the same Alibaba get
3675
03:57:48,174 --> 03:57:51,540
Nvidia chips through a backdoor in Southeast Asia. At the end of the day,
3676
03:57:51,974 --> 03:57:55,455
as with all such stories, you always need a qui bono—who
3677
03:57:55,955 --> 03:57:59,419
benefits—who benefits is pretty clear. Nvidia itself, it's in the business of selling
3678
03:57:59,435 --> 03:58:03,255
chips, so a customer willing to spend billions of dollars on chips in Southeast Asia
3679
03:58:03,287 --> 03:58:07,011
is a godsend. Again, as they themselves state in their SEC filings,
3680
03:58:07,316
--> 03:58:10,542
not being able to tap into Chinese demand would have a material and adverse
3681
03:58:10,750 --> 03:58:13,976
impact on our business. Chinese companies like Alibaba—
3682
03:58:14,281 --> 03:58:17,122
facts are: 1, Chinese chips haven't caught up yet; 2,
3683
03:58:17,251 --> 03:58:20,573
Chinese chip companies' production capabilities are still limited; and 3,
3684
03:58:21,022 --> 03:58:24,265
Chinese tech firms historically Tech stacks run on Nvidia's CUDA
3685
03:58:24,765 --> 03:58:28,039
system, and rewriting years of code for Hawaii chips is painful.
3686
03:58:28,729 --> 03:58:32,181
Renting Nvidia chips in Southeast Asia is the easiest way to stay
3687
03:58:32,681 --> 03:58:36,340
competitive and serve customer demand in the meantime. The U.S., according to
3688
03:58:36,356 --> 03:58:39,134
core members of Trump administration back in July last year,
3689
03:58:39,326 --> 03:58:43,003
U.S. Commerce Secretary Howard Lutnick explained that the point of letting
3690
03:58:43,100 --> 03:58:46,504
Nvidia sell more chips to China was to get Chinese developers addicted to
3691
03:58:47,004 --> 03:58:50,871
the American technology stack. Who does not benefit, according to Chinese policy,
3692
03:58:51,224 --> 03:58:55,640
is the Chinese state, which has made itself reliant on one self-reliance
3693
03:58:56,140 --> 03:58:59,494
one of its foremost priorities. Which means that to believe
3694
03:58:59,994 --> 03:59:02,962
the spy story, you have to believe that a person, Alice Huang, was effectively a
3695
03:59:03,462 --> 03:59:07,522
spy paid by her own government while simultaneously sabotaging its tech self-reliance
3696
03:59:08,022 --> 03:59:11,456
drive, which has spent years and trillions of dollars on pretty senseless.
3697
03:59:11,777 --> 03:59:15,358
Stranger things have happened, I guess. But it doesn't pass the smell test. There's another
3698
03:59:15,455 --> 03:59:19,165
absolutely major contradiction in the media story. According to a Financial Times
3699
03:59:19,229 --> 03:59:22,795
article published yesterday, quote, a person who knows Alice Huang said
3700
03:59:23,295 --> 03:59:26,361
they had been asked by MSS for information about her earlier this
3701
03:59:26,473 --> 03:59:30,296
year after news emerged that Singapore was investigating the company in relation
3702
03:59:30,328 --> 03:59:34,199
to chip smuggling allegations. The agency wanted to find out more information
3703
03:59:34,699 --> 03:59:38,295
about Huang, including the nature of her business and connections, which raises a
3704
03:59:38,359 --> 03:59:41,630
damn obvious If she were indeed an MSS asset,
3705
03:59:41,646 --> 03:59:44,916
why on earth would MSS need to task people around who she was and
3706
03:59:45,416 --> 03:59:48,267
the nature of her business? It's a complete contradiction with the spy story,
3707
03:59:48,379 --> 03:59:52,515
but very much not a contradiction with what I described above—a private businesswoman
3708
03:59:52,531 --> 03:59:55,640
running a workaround to Beijing's own policy,
3709
03:59:55,993 --> 03:59:58,990
which the Chinese security apparatus would naturally want to look into.
3710
03:59:59,455 --> 04:00:03,206
Especially since around the same time Shanghai police detained her business partner,
3711
04:00:03,511 --> 04:00:06,313
reportedly her husband, in a stock manipulation case. Case,
3712
04:00:06,781 --> 04:00:10,781
also coherent with Chinese authorities looking into her network, and much harder
3713
04:00:10,813 --> 04:00:13,120
to square with her being a prized MSS asset.
3714
04:00:15,670 --> 04:00:19,156
Reportedly, her husband Alex— this gets thicker— was
3715
04:00:19,189 --> 04:00:22,486
she cheating on her husband? Were two
3716
04:00:22,583 --> 04:00:23,647
Huangs make a Wright?
3717
04:00:27,276 --> 04:00:29,937
You know, Chris Wright.
3718
04:00:32,182 --> 04:00:35,137
And to top it all, there is to this day not a shred of evidence
3719
04:00:35,185 --> 04:00:39,011
that's been published that She is in fact a Chinese spy, just anonymous intelligence officials
3720
04:00:39,511 --> 04:00:43,156
repeated by journalists who never verified it. So the claim went to print without
3721
04:00:43,656 --> 04:00:46,627
anyone seeing the evidence, without anyone being able to check it, without anyone knowing whether
3722
04:00:47,127 --> 04:00:51,030
it even exists. And given everything above, from Beijing blocking the very chips
3723
04:00:51,530 --> 04:00:55,191
she was supposedly recruiting to help get into China, to the MSS asking
3724
04:00:55,239 --> 04:00:58,501
around about her, to her partner being detained in Shanghai,
3725
04:00:58,517 --> 04:01:01,763
my bet is the evidence doesn't exist at all. If anything, everything we can.
3726
04:01:02,072 --> 04:01:05,483
checks and points the other way. So Alex, I think we've done
3727
04:01:05,676 --> 04:01:09,204
a decent job at painting the
3728
04:01:09,704 --> 04:01:13,224
picture spectrum here, even if we haven't resolved anything. You know, at least going
3729
04:01:13,256 --> 04:01:17,059
through this to describe the scenarios. Well, we're trying to find
3730
04:01:17,123 --> 04:01:20,293
out if there's anything to resolve, and we're having a fucking hard time of it.
3731
04:01:21,002 --> 04:01:26,060
True. I'll tell you that. I mean, but I need to
3732
04:01:26,076 --> 04:01:29,864
add another little tangent to the story, which I've
3733
04:01:30,364 --> 04:01:34,315
sent to you, but I can read the tweet at least from Bloomberg. Ousted Nvidia
3734
04:01:34,815 --> 04:01:38,666
chief executive Jensen Huang is discussing a possible comeback
3735
04:01:39,166 --> 04:01:42,454
to the company just days after the board of directors pushed him out, the Financial
3736
04:01:42,954 --> 04:01:46,701
Times reported. Yes, I was wondering if he was going to be doing a Sam
3737
04:01:47,201 --> 04:01:50,584
Altman. Um, now Nvidia is nothing without
3738
04:01:51,084 --> 04:01:54,712
its
people. It's a little
3739
04:01:54,745 --> 04:02:00,625
difficult for me to envision that,
3740
04:02:00,866 --> 04:02:05,498
simply because if he did have more extramarital
3741
04:02:05,547 --> 04:02:09,487
affairs under his belt, he's going to have to get divorced.
3742
04:02:09,551 --> 04:02:12,797
I can't imagine him being able to mount a return,
3743
04:02:13,311 --> 04:02:17,006
manage the company, and keep a marriage. That's just a whole lot to— Last
3744
04:02:17,506 --> 04:02:20,154
I heard, his wife wants to work on the marriage.
3745
04:02:22,816 --> 04:02:25,652
So, I mean, look, for As far as I can tell,
3746
04:02:26,798 --> 04:02:30,028
I don't know. You know, maybe they were—he was staying together with his wife
3747
04:02:30,528 --> 04:02:34,432
for the kids or whatever. This guy seemed to be slinging
3748
04:02:34,480 --> 04:02:37,831
it around. It wasn't—there wasn't any particular,
3749
04:02:39,660 --> 04:02:43,291
you know, friction. I mean, his his daughter apparently took
3750
04:02:43,791 --> 04:02:47,448
photos of him and Alice Huang and whatever.
3751
04:02:47,773 --> 04:02:51,376
You know, he's going around being photographed. and videoed with, you know, various women
3752
04:02:51,825 --> 04:02:55,238
in various states. Like, you know, it's not a situation of like,
3753
04:02:55,254 --> 04:02:58,732
oh, it's my wife. I mean, he's in, he's in video with Al
3754
04:02:58,796 --> 04:03:02,081
Swang. Yes, exactly. He's in video.
3755
04:03:02,417 --> 04:03:05,927
Um, um, you see what I
3756
04:03:06,427 --> 04:03:09,116
mean? Like, you don't, I mean, I don't know what, how their marriage works or
3757
04:03:09,616 --> 04:03:12,818
whatever, but I've, I've, you know, it was funny that Hillary was like, I'm not
3758
04:03:13,318 --> 04:03:16,367
gonna get into the morality of cheating. No, I bet Of course not.
3759
04:03:19,710 --> 04:03:22,317
Our marriage is a convenience. People are not kidding. I mean, who would have,
3760
04:03:22,757 --> 04:03:24,402
who could, who could do such a thing?
3761
04:03:27,150 --> 04:03:30,796
Anyway. Um, yeah. So, you know, that doesn't
3762
04:03:31,296 --> 04:03:35,816
seem to be a problem apparently. Okay. But, you know,
3763
04:03:35,880 --> 04:03:39,580
but the, what has happened is a trillion dollars off of NVIDIA's,
3764
04:03:39,660 --> 04:03:39,790
um,
3765
04:03:42,881 --> 04:03:46,271
Market share, market cap. And believe
3766
04:03:46,771 --> 04:03:49,597
you me, these things are, you know, if let me put it this way.
3767
04:03:49,936 --> 04:03:53,582
It takes about 15% to call a
3768
04:03:53,727 --> 04:03:57,985
shareholder meeting to vote in or out the
3769
04:03:58,485 --> 04:04:03,453
board. And, you know, it doesn't take a ton of coordination
3770
04:04:03,550 --> 04:04:09,091
from Huang. Already one set of investors in Nvidia have
3771
04:04:09,591 --> 04:04:12,636
said, you know, like, okay, you've got an interim CEO in. Now we advise you
3772
04:04:12,828 --> 04:04:16,741
to take things slow and consider all options
3773
04:04:16,757 --> 04:04:19,532
now that you have a safe pair of hands in the interim,
3774
04:04:20,077 --> 04:04:23,028
including the option of Jensen coming back, wink wink.
3775
04:04:24,471 --> 04:04:27,566
You know, these letters don't go out at random. Like, this is somebody
3776
04:04:27,598 --> 04:04:30,950
saying, we are the shareholders. If you fuck this up, we're gonna fuck
3777
04:04:31,450 --> 04:04:34,814
you up. This isn't, you know, this is
3778
04:04:34,974 --> 04:04:39,349
billions and trillions of dollars at play. And by the way, people who
3779
04:04:39,381 --> 04:04:42,554
are hearing us probably don't know much of this, but Jensen Huang
3780
04:04:42,586 --> 04:04:46,255
was running NVIDIA as a fiefdom, right? Like he was giving
3781
04:04:46,755 --> 04:04:49,845
his word on orders of billions of dollars of chips. And he was,
3782
04:04:50,502 --> 04:04:54,108
he had 60 direct reports. Like, I don't know if people
3783
04:04:54,608 --> 04:04:57,777
understand what this means, but basically it means he's the company. He's the company.
3784
04:04:58,354 --> 04:05:01,611
And yeah, yeah. the company. Like, that's— I was gonna say,
3785
04:05:03,007 --> 04:05:06,345
we're running a little bit— We're running a little bit long, so I
3786
04:05:06,377 --> 04:05:09,008
want to make— I, I don't, you know, I want to make sure everybody has,
3787
04:05:09,121 --> 04:05:12,763
especially you, Alex, have, have your evening. Um, uh,
3788
04:05:13,068 --> 04:05:16,422
so I— we could possibly, if he doesn't return this
3789
04:05:16,922 --> 04:05:20,290
week in some sort of, you know, whiplash scenario,
3790
04:05:20,434 --> 04:05:24,253
we can discuss, um, potential futures in
3791
04:05:24,350 --> 04:05:27,609
his absence because it will rock the whole
3792
04:05:27,737 --> 04:05:31,031
environment. I mean, things change every— like the, the, uh,
3793
04:05:31,433 --> 04:05:35,193
the pause the AI might have kind of happened by accident if
3794
04:05:35,693 --> 04:05:38,663
he leaves, right? Like there's, there's a lot of consequences and potential moving parts.
3795
04:05:38,888 --> 04:05:41,909
There's a lot of consequences. And I'm sure we'll learn more about this, but I
3796
04:05:42,037 --> 04:05:45,524
am glad we did a thorough, like, walking through. There's more
3797
04:05:45,878 --> 04:05:49,557
elements of this that we haven't gone through, but yes, people need to understand.
3798
04:05:49,605 --> 04:05:52,722
I think what you brought up there at the end is extremely important. China,
3799
04:05:53,589 --> 04:05:58,085
as far as we know, does not want the chips. Effectively, what happened
3800
04:05:58,585 --> 04:06:01,907
with the export bans of the Biden era was a
3801
04:06:02,517 --> 04:06:06,338
gift to the Chinese industry that previously
3802
04:06:06,531 --> 04:06:09,806
they kind of knew that they needed to develop their own, but the Nvidia chips
3803
04:06:10,306 --> 04:06:13,837
were coming in and the distance from the Huawei alternatives was so
3804
04:06:14,337 --> 04:06:18,060
big that it was very hard to generate the collective
3805
04:06:18,880 --> 04:06:22,422
of like, we need to drop everything and work on closing the
3806
04:06:22,922 --> 04:06:26,914
gap, right? And what happened with the export bans is people
3807
04:06:26,930 --> 04:06:30,728
were like, you know, they had their oh shit moment where multiple layers
3808
04:06:31,228 --> 04:06:35,110
of the stack needed to be rebuilt. From not
3809
04:06:35,610 --> 04:06:38,055
just the silicon level where, you know, the Chinese are behind, they're catching up,
3810
04:06:38,071 --> 04:06:41,529
whatever, that's a process that's ongoing, but also on, you know,
3811
04:06:41,546 --> 04:06:45,061
the GPU technology and the software stack and the cooling stack alternatives.
3812
04:06:45,157 --> 04:06:48,218
And, you know, there's like a lot of elements that are not just built by
3813
04:06:48,718 --> 04:06:53,105
Huawei, but all the companies that are building on top of Huawei
3814
04:06:53,249 --> 04:06:56,967
that would have needed to plug all sorts of gaps
3815
04:06:57,015 --> 04:07:00,380
that occur when you're training a model without NVIDIA chips.
3816
04:07:01,021 --> 04:07:04,803
And as far as we know, much of that
3817
04:07:05,303 --> 04:07:07,736
gap has been plugged already.
3818
04:07:09,130 --> 04:07:12,283
So that's sort of, you know, coordination signal that's
3819
04:07:12,381 --> 04:07:14,558
been gifted to the Chinese, the Chinese are not wasting.
3820
04:07:16,433 --> 04:07:20,694
They are, you know, investing it quite strongly
3821
04:07:21,375 --> 04:07:24,850
in catching up. So, and apparently
3822
04:07:25,301 --> 04:07:28,507
there was this viral post going around showing that a lot of Chinese
3823
04:07:29,007 --> 04:07:32,586
models have been trained on NVIDIA chips. But what that person,
3824
04:07:32,942 --> 04:07:36,231
very, very dirty move, I think, did not put in there is
3825
04:07:36,329 --> 04:07:40,079
that he did not find a signature. So, so this person
3826
04:07:40,579 --> 04:07:44,077
identified a particular signature in the initialization of the neural networks,
3827
04:07:44,703 --> 04:07:47,866
uh, that indicated that they had been using the latest NVIDIA chips,
3828
04:07:47,914 --> 04:07:51,447
which again could have been gotten through, uh, could have been
3829
04:07:51,947 --> 04:07:55,092
gotten through, uh, um, an API, right?
3830
04:07:55,654 --> 04:07:58,223
Right. So, so that, that doesn't necessarily mean that they had,
3831
04:07:58,833 --> 04:08:03,295
you know, the whole run on, on American chips, but What
3832
04:08:03,795 --> 04:08:06,438
that person did not put in their report is that he did not find that
3833
04:08:06,938 --> 04:08:10,241
signature in the latest DeepSeek models, right? Which, if we
3834
04:08:10,741 --> 04:08:14,742
accept that finding it means that they use Blackwells, not finding
3835
04:08:15,242 --> 04:08:18,068
it might mean that they use the sends, like Huawei chips, right? Which is a
3836
04:08:18,197 --> 04:08:21,330
big deal because that model is in the top 10 right now. Right. It's not
3837
04:08:21,830 --> 04:08:25,363
like some long tail type thing.
3838
04:08:25,702 --> 04:08:26,896
It's a legitimate contender.
3839
04:08:28,818 --> 04:08:31,195
So, you know, the gap is closing,
3840
04:08:32,111 --> 04:08:35,579
and you know, the last thing the U.S. needs—let me put
3841
04:08:36,079 --> 04:08:39,980
it this way—the last thing the U.S. needs is to literally put its the brain
3842
04:08:40,480 --> 04:08:43,722
of its chip industry on the bench. Like that is so
3843
04:08:43,770 --> 04:08:46,789
fucking stupid. I can't even believe we're talking about it.
3844
04:08:48,267 --> 04:08:51,912
It's been a wild week, man. It's going to get even more complicated if if
3845
04:08:52,041 --> 04:08:55,188
Hwang tries to mount a. Yeah,
3846
04:08:55,797 --> 04:08:59,596
yeah. Uh, if he, if he tries to mount a recovery and come back,
3847
04:08:59,740 --> 04:09:02,978
or, or we, we wait, like, the— either way,
3848
04:09:03,811 --> 04:09:07,257
um, it's gonna be a scramble this week, uh, as people try to figure
3849
04:09:07,757 --> 04:09:11,264
out what to do. Um, and, uh, we'll, we'll try to play catch-up,
3850
04:09:11,584 --> 04:09:14,966
right? Some of it, like, Elon had tweeted out that, you know,
3851
04:09:15,047 --> 04:09:18,605
it will be an honor for, you know, because he's having his own semiconductor move
3852
04:09:18,653 --> 04:09:22,355
that he's making. Like, sure, if he goes to like a competitor
3853
04:09:22,419 --> 04:09:25,821
or whatever, now you don't You aren't just missing Nvidia's brain.
3854
04:09:26,255 --> 04:09:29,707
He's actively working against them. Well, I mean, you mentioned
3855
04:09:30,207 --> 04:09:34,347
several months ago that Huang was developing
3856
04:09:34,460 --> 04:09:38,233
this sort of lease thing with
3857
04:09:38,698 --> 04:09:42,086
the GPUs to try to make almost like a
3858
04:09:42,472 --> 04:09:46,101
credit system with the GPUs and become somewhat
3859
04:09:46,601 --> 04:09:49,555
of a pseudo bank when it comes to GPUs to try to— Right, right,
3860
04:09:49,636 --> 04:09:53,643
right. Well, all of these things, a lot of these were deals
3861
04:09:53,659 --> 04:09:56,907
that run on Huang's word, right? People that he knows and
3862
04:09:57,407 --> 04:09:59,951
trusted. You eject all of that. I mean,
3863
04:10:00,774 --> 04:10:04,775
I don't look, I, in general, when a founder
3864
04:10:04,807 --> 04:10:08,761
leaves a company like Steve Jobs with Apple the second time around,
3865
04:10:11,048 --> 04:10:15,107
and the next generation that comes in are usually caretakers, but counterintuitively,
3866
04:10:15,246 --> 04:10:18,778
Effectively what they do is they slaughter the
3867
04:10:19,278 --> 04:10:23,258
goose with the golden
eggs. So, for a while, you actually have an improvement
3868
04:10:23,290 --> 04:10:26,389
in financial performance because a lot of the intangibles are monetized.
3869
04:10:26,983 --> 04:10:30,178
So, they start giving out dividends, they start doing all the things that the
3870
04:10:30,678 --> 04:10:34,209
market likes. And what happens is you have a countdown
3871
04:10:34,709 --> 04:10:38,929
of 5 to 10 years before the company has really lost its momentum
3872
04:10:39,429 --> 04:10:43,164
because the founder usually does— Sell. long-term investments
3873
04:10:44,616 --> 04:10:47,729
that a caretaker doesn't necessarily understand,
3874
04:10:47,777 --> 04:10:52,073
appreciate, or care about, or want to continue doing because they
3875
04:10:52,573 --> 04:10:56,581
feel a much more intense pressure for short-term results.
3876
04:10:57,386 --> 04:11:01,081
But whatever it is, that is a common pattern. So I'm not saying Nvidia is
3877
04:11:01,113 --> 04:11:04,070
going to die tomorrow, but what I'm saying is a lot of— No, but it
3878
04:11:04,570 --> 04:11:08,554
changes the future landscape hugely. Yeah. Yeah. I mean, all of
3879
04:11:08,570 --> 04:11:12,117
those, all of the— like Anthropic has commitments and
3880
04:11:12,133 --> 04:11:15,953
things like that on purchasing. Are those out the window? That's all
3881
04:11:16,082 --> 04:11:20,059
in their recent perspective
3882
04:11:20,751 --> 04:11:24,406
for their IPO. And it's like, well, do those commitments evaporate?
3883
04:11:24,921 --> 04:11:28,528
All sorts of things go into question, including— I don't know if
3884
04:11:28,641 --> 04:11:32,285
you— I noticed that the CEO
3885
04:11:32,301 --> 04:11:36,537
of AMD looks suspiciously like Jensen
3886
04:11:36,553 --> 04:11:40,192
Huang if he were performing a Mrs. Doubtfire. And I'm
3887
04:11:40,641 --> 04:11:44,600
like, what's going on here? So,
3888
04:11:44,824 --> 04:11:48,896
you know, Nvidia has the biggest
3889
04:11:49,396 --> 04:11:52,823
opportunity here to not drop the ball if he's actually out, right? It's like,
3890
04:11:52,984 --> 04:11:56,045
come on, you got the great opportunity to play catch up here.
3891
04:11:56,606 --> 04:11:59,764
But Alex, I think we did a good job
3892
04:12:00,264 --> 04:12:01,914
this week. I think we did a good job this week. A lot of stuff
3893
04:12:01,962 --> 04:12:05,058
to cover. I even had a half hour
3894
04:12:05,170 --> 04:12:09,003
before the show that I had finished all of the prep. It's— that's sometimes
3895
04:12:09,019 --> 04:12:12,580
down to the wire. Wow. I had a half hour with all this. Okay,
3896
04:12:13,751 --> 04:12:17,521
so, so what's your bet, Gator? Next time— this week— next time— next
3897
04:12:18,021 --> 04:12:21,563
time— next week, this same time, is Jensen
3898
04:12:21,611 --> 04:12:24,915
Huang back? He is not back as of next Sunday.
3899
04:12:25,220 --> 04:12:28,332
He could be back the following one, but as of now one, he's not back.
3900
04:12:28,926 --> 04:12:32,264
That's my bet. All right, let's see what happens. Do you take the opposite bet,
3901
04:12:32,280 --> 04:12:36,853
or do you agree? I don't know.
3902
04:12:37,255 --> 04:12:41,283
I was unfair because I gave you a bet I wouldn't know how to
3903
04:12:41,783 --> 04:12:45,503
respond myself. I'm going
3904
04:12:45,616 --> 04:12:49,018
to say no, simply because the moving parts
3905
04:12:49,066 --> 04:12:52,644
to reinstate him are probably slower than
3906
04:12:52,933 --> 04:12:55,777
it takes to write a resignation letter. Like,
3907
04:12:56,084 --> 04:12:59,685
yeah, we'll, we'll see. We'll see. I, I found it fascinating that
3908
04:12:59,718 --> 04:13:03,335
the Financial Times reported that he is discussing coming back because
3909
04:13:03,691 --> 04:13:07,069
they said, you know, he's discussing with his close associates. Well, why would he be
3910
04:13:07,569 --> 04:13:10,708
discussing it if there wasn't an offer on the table, right? Um,
3911
04:13:11,307 --> 04:13:15,220
so clearly there must be feelers
3912
04:13:15,253 --> 04:13:18,358
if this is true. There must be feelers out. Um,
3913
04:13:19,603 --> 04:13:22,868
or You know, it depends, right? It could be that
3914
04:13:23,368 --> 04:13:26,660
there are a group of shareholders that have said, like, you know, we're gonna
3915
04:13:27,160 --> 04:13:30,174
fuck up the board, which could take time. Or it could be the board themselves,
3916
04:13:30,680 --> 04:13:33,799
or, you know, there could be all sorts of things. We'll find out. We'll find
3917
04:13:34,299 --> 04:13:37,445
out. All right, sir. Thank you for joining me this week.
3918
04:13:38,462 --> 04:13:41,272
Thank you, sir. Everyone in the audience enjoyed,
3919
04:13:41,369 --> 04:13:44,179
and we will see you all next Sunday.
3920
04:13:46,135 --> 04:13:47,084
For the next dumb week.