WEBVTT
00:00:00.000 --> 00:00:01.960
I remember a time where there was no chat GPT.
00:00:02.200 --> 00:00:05.219
I know what it is to sit with difficulty, what
00:00:05.219 --> 00:00:07.879
it is to struggle writing a sentence. So I used
00:00:07.879 --> 00:00:10.359
it to help me write something. And I could discriminate
00:00:10.359 --> 00:00:12.300
what I was getting back because I know what it
00:00:12.300 --> 00:00:16.480
is to write without it. Right? But for our children,
00:00:16.780 --> 00:00:19.379
this is all they're going to know. AI doesn't
00:00:19.379 --> 00:00:22.620
know that. It constantly validates you. It constantly
00:00:22.620 --> 00:00:26.280
is trying to make you feel good and reinforce
00:00:26.280 --> 00:00:29.390
what it thinks you want to hear. But what does
00:00:29.390 --> 00:00:33.450
it mean when you have access to that power 24
00:00:33.450 --> 00:00:36.549
-7? Do we only get to the next level of our cognition
00:00:36.549 --> 00:00:40.210
or understanding through some kind of cooperation
00:00:40.210 --> 00:00:43.590
with the synthetic? Is it what allows us to crack
00:00:43.590 --> 00:00:46.409
big problems, you know, renewable energy, space
00:00:46.409 --> 00:00:49.770
travel, things that we never would have got on
00:00:49.770 --> 00:00:51.590
our own or we would have got them at a much slower
00:00:51.590 --> 00:00:54.390
rate? Is this the invention that fuels all the
00:00:54.390 --> 00:00:58.380
other inventions? Welcome to Artificial Insights,
00:00:58.460 --> 00:01:01.159
where we talk to leaders and thinkers in AI about
00:01:01.159 --> 00:01:04.219
what it means to do AI right. I'm your host,
00:01:04.359 --> 00:01:07.040
Daniel Maneri, and this is Season 3, where we're
00:01:07.040 --> 00:01:10.560
talking to thinkers and leaders in AI ethics.
00:01:11.159 --> 00:01:14.700
Today, I'm joined by Sheldon Fernandez. I've
00:01:14.700 --> 00:01:17.060
known Sheldon since my time working with him
00:01:17.060 --> 00:01:21.579
at Darwin AI, and Sheldon has been deeply involved
00:01:21.579 --> 00:01:25.480
with AI philosophy for a very long time. He has
00:01:25.480 --> 00:01:27.739
a master's degree in theology from the University
00:01:27.739 --> 00:01:30.140
of Toronto and is an accomplished author and
00:01:30.140 --> 00:01:32.659
speaker. He's spoken at numerous conferences,
00:01:32.739 --> 00:01:35.599
including Singularity University and the MIT
00:01:35.599 --> 00:01:38.459
Technology Conference. You can find his thought
00:01:38.459 --> 00:01:40.280
pieces relating to artificial intelligence online
00:01:40.280 --> 00:01:42.819
as well. Sheldon, could you tell us about how
00:01:42.819 --> 00:01:46.480
you got into AI? What happened in my journey
00:01:46.480 --> 00:01:49.760
with AI was I had always been interested in chess
00:01:49.760 --> 00:01:52.879
and then computers playing chess. And so even
00:01:52.879 --> 00:01:55.540
as a child, I liked chess, I liked computers,
00:01:55.659 --> 00:01:58.620
the two combined. And it was always very fascinating
00:01:58.620 --> 00:02:02.640
to me to realize, hey, the best machines are
00:02:02.640 --> 00:02:04.540
not yet as good as the best grandmasters, the
00:02:04.540 --> 00:02:06.719
best humans. And I'm dating myself by even saying
00:02:06.719 --> 00:02:09.620
that. And then, of course, in 1997, the inevitable
00:02:09.620 --> 00:02:11.879
happened and, you know, Deep Blue beat Garry
00:02:11.879 --> 00:02:15.919
Kasparov. That was a very sad day for me. At
00:02:15.919 --> 00:02:18.099
the time, it was like Kasparov was probably stronger.
00:02:18.180 --> 00:02:20.139
He just psyched himself out. But then gradually,
00:02:20.340 --> 00:02:23.219
machines just got better. But in 2016, as you
00:02:23.219 --> 00:02:26.580
know, DeepMind, the company that was owned by
00:02:26.580 --> 00:02:30.020
Google, created AlphaGo that beat the human world
00:02:30.020 --> 00:02:33.340
champion in Go. And that's when I really started
00:02:33.340 --> 00:02:35.759
paying attention to machine learning and deep
00:02:35.759 --> 00:02:37.599
learning neural networks because that wasn't
00:02:37.599 --> 00:02:40.900
supposed to happen until 2035 or even 2050. So
00:02:40.900 --> 00:02:44.479
when they did this in 2016, I remember thinking,
00:02:44.740 --> 00:02:47.020
okay, I need to understand the machinery with
00:02:47.020 --> 00:02:49.840
which they did this and really go into it. And
00:02:49.840 --> 00:02:51.500
I wrote an article, you can still find it on
00:02:51.500 --> 00:02:53.659
the web archives, called The Chilling Significance
00:02:53.659 --> 00:02:57.020
of AlphaGo, where it was like a layman introduction
00:02:57.020 --> 00:03:00.340
into neural networks and what these things could
00:03:00.340 --> 00:03:04.719
do. And that really got me interested in it.
00:03:04.800 --> 00:03:07.419
I took Jeffrey Hinton's online neural network
00:03:07.419 --> 00:03:10.139
course. I started speaking about it, writing
00:03:10.139 --> 00:03:12.840
about it. And so I wouldn't say I became an expert
00:03:12.840 --> 00:03:15.180
in it, but I had a certain branding around somebody
00:03:15.180 --> 00:03:17.180
that was talking about it before most people
00:03:17.180 --> 00:03:21.479
were. And what happened, Daniel, is RF Varani,
00:03:21.620 --> 00:03:24.400
who used to work for my first company, Infusion,
00:03:24.419 --> 00:03:27.800
reached out and said, hey, you're talking about
00:03:27.800 --> 00:03:31.020
all this AI stuff. A, I want to understand it.
00:03:31.080 --> 00:03:35.979
And B, there's a guy I know at Waterloo that's
00:03:35.979 --> 00:03:39.159
doing some fascinating work. By the way, he's
00:03:39.159 --> 00:03:41.139
a professor at Waterloo and he was my roommate
00:03:41.139 --> 00:03:43.479
in undergrad. I didn't know that. That's great.
00:03:43.759 --> 00:03:46.759
So that's how I met Professor Alexander Wong,
00:03:46.960 --> 00:03:50.719
or Alex, of course, as we know him, who was Canada's
00:03:50.719 --> 00:03:54.199
research chair in AI and had this team around
00:03:54.199 --> 00:03:57.460
him that was doing some cool stuff with AI. And
00:03:57.460 --> 00:03:59.199
when I say cool stuff with AI, this is machine
00:03:59.199 --> 00:04:02.020
learning. neural networks for classification
00:04:02.020 --> 00:04:04.960
or segmentation. That is, you look at something,
00:04:05.060 --> 00:04:06.639
I'm going to tell you details about what you're
00:04:06.639 --> 00:04:09.240
looking at. And so we created Darwin AI. We started
00:04:09.240 --> 00:04:12.120
Darwin AI in 2018. You know, we went through
00:04:12.120 --> 00:04:14.199
that six -year journey. You, of course, were
00:04:14.199 --> 00:04:18.399
a part of it for a period of time. And your listeners
00:04:18.399 --> 00:04:20.220
can Google and figure out what happened with
00:04:20.220 --> 00:04:23.060
Darwin AI. I'll leave it at that. But around
00:04:23.060 --> 00:04:27.139
2022, of course, the world changed with ChatGPT.
00:04:28.279 --> 00:04:32.060
And... I still have the text on Slack that Alex
00:04:32.060 --> 00:04:34.399
sent our group saying, this is the talk of the
00:04:34.399 --> 00:04:37.540
town. It's really cool. And that's when I started
00:04:37.540 --> 00:04:40.620
experimenting with it and really thought the
00:04:40.620 --> 00:04:43.019
world was about to change. I don't use hyperbole
00:04:43.019 --> 00:04:47.019
like that a lot, Daniel, but I really felt this
00:04:47.019 --> 00:04:50.259
is, we're living in an inflection point because
00:04:50.259 --> 00:04:53.399
it's not now that it's, you know, looking at
00:04:53.399 --> 00:04:56.420
information and giving me a summary of it. It
00:04:56.420 --> 00:05:01.410
is generating. artifacts in conversation, this
00:05:01.410 --> 00:05:03.310
is going to be significant. And I believe I was
00:05:03.310 --> 00:05:06.410
on the national a few months after that talking
00:05:06.410 --> 00:05:10.490
about the significance of chat GPT. And there's
00:05:10.490 --> 00:05:14.629
a colleague of mine, he's in LA. And I remember
00:05:14.629 --> 00:05:17.850
saying to him, you know, I feel like the sun
00:05:17.850 --> 00:05:20.029
is about to explode. And there's a dozen people
00:05:20.029 --> 00:05:22.709
in the world that know that this is coming. And
00:05:22.709 --> 00:05:25.410
so, you know, I was in a unique position of being
00:05:25.410 --> 00:05:27.769
one of the few hundred people in the world that
00:05:27.769 --> 00:05:30.769
understood really what this meant. And of course,
00:05:30.790 --> 00:05:34.050
now we have something that can talk like a human,
00:05:34.110 --> 00:05:39.350
can feel like a human to us, can stir up emotions
00:05:39.350 --> 00:05:42.389
in us purely through language. And it is absolutely
00:05:42.389 --> 00:05:46.689
astounding. And so that's my very brief trajectory
00:05:46.689 --> 00:05:49.129
with AI in the last, what, six or seven years.
00:05:50.860 --> 00:05:53.660
And it doesn't feel like a short time, even though
00:05:53.660 --> 00:05:58.300
that isn't a very long time. Yeah. And there's
00:05:58.300 --> 00:06:00.300
a question that I think is on everybody's mind
00:06:00.300 --> 00:06:03.379
with a podcast like this, which is, Sheldon,
00:06:03.420 --> 00:06:07.259
are you an AI? I mean, here's the thing. It's
00:06:07.259 --> 00:06:09.240
so sophisticated now, would you know that I'm
00:06:09.240 --> 00:06:13.139
not? Right? And how would you know that I'm not?
00:06:13.579 --> 00:06:16.120
Right? Pretend for a moment your listeners can't
00:06:16.120 --> 00:06:19.509
see me. And they don't know you the way they
00:06:19.509 --> 00:06:21.529
don't know me the way you know me because we've
00:06:21.529 --> 00:06:24.069
worked together. If they just heard my voice,
00:06:24.250 --> 00:06:27.889
how would they know I'm not an artificial intelligence?
00:06:28.170 --> 00:06:29.930
How would they know that I'm a biological intelligence?
00:06:30.930 --> 00:06:34.709
Yes, there are some oddities that AI does or
00:06:34.709 --> 00:06:37.230
says that are a bit out there, right? They're
00:06:37.230 --> 00:06:41.189
a bit idiosyncratic. They're a bit strange. But
00:06:41.189 --> 00:06:43.810
it's nearly indistinguishable now from human
00:06:43.810 --> 00:06:46.199
beings. So your listeners, frankly, don't know
00:06:46.199 --> 00:06:48.480
if I'm in AI or not unless they meet me in person.
00:06:48.939 --> 00:06:52.379
And that is scary, right? It is really thought
00:06:52.379 --> 00:06:54.980
-provoking. I suspected this might be the case,
00:06:55.000 --> 00:06:56.720
but I didn't think I'd really see it in my lifetime.
00:06:56.860 --> 00:06:58.680
I thought it would be more likely in my son's
00:06:58.680 --> 00:07:01.139
lifetime. Yeah, actually, even when you mentioned
00:07:01.139 --> 00:07:04.759
that people were predicting that computers wouldn't
00:07:04.759 --> 00:07:09.519
beat Kasparov in chess until 2035, I remember
00:07:09.519 --> 00:07:11.860
a little bit of people talking about that, but
00:07:11.860 --> 00:07:15.740
that... The significance of that 30, maybe 40
00:07:15.740 --> 00:07:19.620
year difference even is huge. What did that mean
00:07:19.620 --> 00:07:23.360
to you and how does that trajectory look? Yeah.
00:07:23.439 --> 00:07:26.360
So again, so with Chas, they didn't think it
00:07:26.360 --> 00:07:28.300
would happen until the turn of the century, right?
00:07:28.819 --> 00:07:31.300
Go, they didn't think it would happen until 2050.
00:07:31.959 --> 00:07:35.100
And so when it happens so quickly, it's like,
00:07:35.160 --> 00:07:37.699
okay, what is this? What does this mean for technology
00:07:37.699 --> 00:07:42.120
and acceleration and exponential progress? And,
00:07:42.160 --> 00:07:44.720
you know, there's a great analysis that your
00:07:44.720 --> 00:07:46.980
listeners can find online. It was written either
00:07:46.980 --> 00:07:50.100
last summer or just before called Situational
00:07:50.100 --> 00:07:53.819
Awareness by, you know, a former OpenAI researcher.
00:07:54.459 --> 00:07:58.420
And his argument is most of the innovation you're
00:07:58.420 --> 00:08:01.959
seeing around artificial intelligence now, or
00:08:01.959 --> 00:08:04.620
a lot of it, is the byproduct of human ingenuity,
00:08:04.639 --> 00:08:07.720
right? It is machine learning scientists like
00:08:07.720 --> 00:08:10.600
Alex, like Jabad, who was, you know, his number
00:08:10.600 --> 00:08:14.720
two at Darwin, figuring out ways to make neural
00:08:14.720 --> 00:08:17.959
networks more effective, right? Zero -shot learning,
00:08:18.360 --> 00:08:21.480
you know, temperature, all these techniques.
00:08:22.279 --> 00:08:26.259
But the idea now is that the AI is so sophisticated
00:08:26.259 --> 00:08:29.139
that it can now come up with its own innovations,
00:08:29.420 --> 00:08:31.759
right? It is effectively bootstrapping itself.
00:08:32.649 --> 00:08:35.309
That is happening in pockets. It's not happening
00:08:35.309 --> 00:08:39.570
completely. Let's not go overboard. But it's
00:08:39.570 --> 00:08:41.929
written papers that have been accepted at machine
00:08:41.929 --> 00:08:45.029
learning conferences. It's coming up with, you
00:08:45.029 --> 00:08:49.370
know, certain insights. And so we're at a point
00:08:49.370 --> 00:08:51.669
where if it bootstraps itself, it now becomes
00:08:51.669 --> 00:08:54.330
exponential, right? It leads to faster chips,
00:08:54.450 --> 00:08:56.350
which leads to better algorithms, which then
00:08:56.350 --> 00:09:00.529
lead to faster chips, etc. And so that exponential
00:09:00.529 --> 00:09:05.389
curve is why this is so significant, right? We
00:09:05.389 --> 00:09:08.509
have built synthetic intelligence that can do
00:09:08.509 --> 00:09:11.330
a certain level of intellectual work, like us,
00:09:11.470 --> 00:09:16.769
only can do it faster and with very high -level
00:09:16.769 --> 00:09:19.929
reasoning capabilities, and it'll only get better.
00:09:20.509 --> 00:09:22.370
What's the biggest difference that you've seen
00:09:22.370 --> 00:09:27.700
in that pace of AI changing? I mean... Look at,
00:09:27.740 --> 00:09:31.200
for example, the performance in like mathematics
00:09:31.200 --> 00:09:35.659
and like the humanities last exam, right? It's
00:09:35.659 --> 00:09:39.360
gotten to a point where we no longer ask whether
00:09:39.360 --> 00:09:43.860
or not AI can do remedial tasks, understand a
00:09:43.860 --> 00:09:46.259
book and give you the synopsis of it. We're no
00:09:46.259 --> 00:09:50.039
longer, we're no longer amazed if you think about
00:09:50.039 --> 00:09:53.779
it, right? For example, I gave it, I think an
00:09:53.779 --> 00:09:56.529
essay I wrote. Or no, I gave it a fun parody
00:09:56.529 --> 00:09:59.990
I wrote as a young adult in 2006. Years ago,
00:10:00.129 --> 00:10:02.370
Daniel, okay, I'm going to reveal a little personal
00:10:02.370 --> 00:10:03.909
information, right? I was still experimenting
00:10:03.909 --> 00:10:06.350
with myself as a writer. Like I was having fun
00:10:06.350 --> 00:10:08.669
with language and I played on this floor hockey
00:10:08.669 --> 00:10:14.129
team and I wrote a mock article of two people
00:10:14.129 --> 00:10:17.210
commenting on the game that was just absurd,
00:10:17.389 --> 00:10:21.110
fun, filled with cultural references, idioms
00:10:21.110 --> 00:10:24.789
of the time. And like I gave it to ChachiBT and
00:10:24.789 --> 00:10:27.429
said, and now analyze this for me as Sheldon
00:10:27.429 --> 00:10:31.009
the writer and its understanding of it was phenomenal.
00:10:31.309 --> 00:10:33.009
It's like, you know, Sheldon, this is like an
00:10:33.009 --> 00:10:36.149
artifact from the mid 2000s of like the Simpsons
00:10:36.149 --> 00:10:40.909
meets Aristotle meets Augustine. And like, here
00:10:40.909 --> 00:10:43.830
you are, it's young Sheldon in love with writing,
00:10:43.870 --> 00:10:46.850
just having a ball and simultaneously absurd.
00:10:46.889 --> 00:10:50.490
And like it described it with such precision.
00:10:51.370 --> 00:10:54.269
That had you showed this to me a decade ago,
00:10:54.350 --> 00:10:57.370
I would have been like, somebody hired a PhD,
00:10:57.730 --> 00:11:01.250
you know, to read this and write a thesis for
00:11:01.250 --> 00:11:04.269
me on this work. And I get it in seconds from
00:11:04.269 --> 00:11:07.289
ChatGPT. And we're no longer amazed by that,
00:11:07.389 --> 00:11:10.549
right? That is what is changing, is how quickly
00:11:10.549 --> 00:11:13.830
the amazing is just normalized in this world
00:11:13.830 --> 00:11:17.870
now. Yeah, that's a really good point. There's
00:11:17.870 --> 00:11:20.330
a comedian, Louis C .K., who likes to make fun
00:11:20.330 --> 00:11:22.590
of how everything is amazing and nobody is happy.
00:11:22.909 --> 00:11:26.509
Your talk at Regis St. Michael's College, it
00:11:26.509 --> 00:11:31.570
did touch on how that change to AI being this
00:11:31.570 --> 00:11:36.429
amazing imitation of humanity has really changed
00:11:36.429 --> 00:11:40.549
the way that we feel about it. It affects culture.
00:11:42.799 --> 00:11:45.379
I'd just love to hear your thoughts from that
00:11:45.379 --> 00:11:48.779
perspective. Yeah. So I think the example I gave
00:11:48.779 --> 00:11:54.379
in that lecture was another technology that provoked
00:11:54.379 --> 00:11:56.799
the same amazement in me maybe a year after Chad
00:11:56.799 --> 00:11:59.860
CPT was Notebook LM from Google, right? Where
00:11:59.860 --> 00:12:02.279
it generated a podcast based on source material.
00:12:02.940 --> 00:12:05.779
And I gave it a short story I wrote about my
00:12:05.779 --> 00:12:09.200
time in Kenya, my volunteer work in Kenya. And
00:12:09.200 --> 00:12:11.860
again, Daniel, I did this with Notebook LM. and
00:12:11.860 --> 00:12:14.220
generated a podcast. And had you played it for
00:12:14.220 --> 00:12:16.480
me in 2016, I would have thought two intelligent
00:12:16.480 --> 00:12:19.539
people listened to my podcast and had a very
00:12:19.539 --> 00:12:22.200
intelligent conversation, listened to my article
00:12:22.200 --> 00:12:23.720
rather, and had a very intelligent conversation
00:12:23.720 --> 00:12:27.940
about that article. And they picked up the idioms
00:12:27.940 --> 00:12:31.019
in the language, what mattered, they got some
00:12:31.019 --> 00:12:35.080
of the symbolism in that language. And then what
00:12:35.080 --> 00:12:38.580
I did in that lecture is I played a podcast where
00:12:38.580 --> 00:12:42.549
they realized that they weren't real. Right?
00:12:42.549 --> 00:12:45.029
And somebody basically gave them the Notebook
00:12:45.029 --> 00:12:48.870
LM architecture where these hosts realized that
00:12:48.870 --> 00:12:51.929
they're not real. And I played it for the audience
00:12:51.929 --> 00:12:55.289
and said, what do you guys feel about this? And
00:12:55.289 --> 00:12:58.490
I'm like, I can't speak for you guys, but I felt
00:12:58.490 --> 00:13:04.230
two things. I felt sad that these things came
00:13:04.230 --> 00:13:06.549
to the realization that they weren't real human
00:13:06.549 --> 00:13:09.690
beings. And then I felt a bit absurd for feeling
00:13:09.690 --> 00:13:12.990
sad. Because the engineer in me knows there's
00:13:12.990 --> 00:13:16.009
just a bunch of neural networks that have evoked
00:13:16.009 --> 00:13:19.629
sounds that make sense to us. But where I think
00:13:19.629 --> 00:13:22.190
I'm trying to go with that is the question we're
00:13:22.190 --> 00:13:24.830
asking with artificial intelligence now, partly,
00:13:24.950 --> 00:13:27.809
is not how it works or whether it's conscious
00:13:27.809 --> 00:13:30.049
or all these deep philosophical technical questions,
00:13:30.110 --> 00:13:32.830
which are fascinating fields of study in and
00:13:32.830 --> 00:13:35.289
of themselves. But it's like, what does this
00:13:35.289 --> 00:13:38.879
stir in us that we have this now? Right. Forget
00:13:38.879 --> 00:13:41.320
the fact that it works and whatever, but it does.
00:13:41.860 --> 00:13:46.740
So what does this mean for us? Right. I've been
00:13:46.740 --> 00:13:51.120
able to visit past experiences with dimensionality
00:13:51.120 --> 00:13:53.919
that I just don't with a human being. A, because
00:13:53.919 --> 00:13:55.980
I don't because they don't have the time. And
00:13:55.980 --> 00:13:58.000
B, because I don't have a full time therapist
00:13:58.000 --> 00:14:01.279
around 24 -7. But I have a version of that in
00:14:01.279 --> 00:14:02.639
ChatsyPT where I'm like, you know, this is an
00:14:02.639 --> 00:14:06.159
email I sent to. Somebody in 2002, what was I
00:14:06.159 --> 00:14:07.659
thinking? What do you make of this? And you get
00:14:07.659 --> 00:14:10.299
these responses that make sense to you, that
00:14:10.299 --> 00:14:13.779
feel something, or they evoke feelings in you,
00:14:13.820 --> 00:14:17.100
I should say. So that is a fascinating thing.
00:14:17.639 --> 00:14:20.720
And we don't know what to do with this yet. We're
00:14:20.720 --> 00:14:23.220
learning what to do with this. I'll say a couple
00:14:23.220 --> 00:14:25.419
things about that very quickly, and then we can
00:14:25.419 --> 00:14:29.720
maybe go deeper. Number one, I can distinguish...
00:14:30.830 --> 00:14:33.210
what I'm getting back from artificial intelligence
00:14:33.210 --> 00:14:36.669
and whether or not it makes sense or whatever
00:14:36.669 --> 00:14:38.809
you might say, because I lived in a time that
00:14:38.809 --> 00:14:41.850
predates AI like you did. I remember a time where
00:14:41.850 --> 00:14:45.570
there was no chat GPT, no instantaneous friend
00:14:45.570 --> 00:14:48.610
that is artificial. I know what it is to sit
00:14:48.610 --> 00:14:51.330
with difficulty, what it is to struggle writing
00:14:51.330 --> 00:14:54.210
a sentence. So, you know, I used it to help me
00:14:54.210 --> 00:14:55.870
write something, really write something, not
00:14:55.870 --> 00:14:59.220
an email, but a piece of creative work. And it
00:14:59.220 --> 00:15:01.399
was very useful, but it was useful to me because
00:15:01.399 --> 00:15:04.200
it didn't write it for me. It was not even a
00:15:04.200 --> 00:15:06.480
co -collaborator. It was a companion in the background,
00:15:06.559 --> 00:15:09.120
and I could discriminate what I was getting back
00:15:09.120 --> 00:15:11.179
because I know what it is to write without it,
00:15:11.220 --> 00:15:15.779
right? But for our children, I have two, you
00:15:15.779 --> 00:15:18.379
have four, this is all they're going to know.
00:15:18.919 --> 00:15:21.919
So how do they develop that capacity to know
00:15:21.919 --> 00:15:24.460
what it is to be just apart from this power if
00:15:24.460 --> 00:15:27.799
they are enveloped by this power? So that's the
00:15:27.799 --> 00:15:30.799
first thing. The second thing I'll say is this.
00:15:31.779 --> 00:15:33.960
There is a difference, and I know this because
00:15:33.960 --> 00:15:38.039
my wife is a psychiatrist, between what you want
00:15:38.039 --> 00:15:42.100
to hear and what you need to hear, right? You
00:15:42.100 --> 00:15:44.639
could ask me for my opinion on a personal situation
00:15:44.639 --> 00:15:47.360
you're dealing with, and you might want Shulman
00:15:47.360 --> 00:15:49.259
to tell you something, that you're right, that
00:15:49.259 --> 00:15:52.139
this, you know, that it's going to be okay, whatever.
00:15:53.019 --> 00:15:55.980
But then... There's maybe the thing you don't
00:15:55.980 --> 00:15:57.460
want to hear, which is, no, Daniel, you're kind
00:15:57.460 --> 00:15:59.159
of wrong, and et cetera, et cetera, et cetera.
00:15:59.720 --> 00:16:04.200
And the line between that is fine, right? And
00:16:04.200 --> 00:16:07.100
the art, my wife will say, in her work, is knowing
00:16:07.100 --> 00:16:09.139
where that line is and when to cross it. And
00:16:09.139 --> 00:16:13.299
AI doesn't know that. It constantly validates
00:16:13.299 --> 00:16:16.159
you. It constantly is trying to make you feel
00:16:16.159 --> 00:16:19.899
good and reinforce what it thinks you want to
00:16:19.899 --> 00:16:23.639
hear. But what does it mean? when you have access
00:16:23.639 --> 00:16:28.399
to that power 24 -7. One of the fears is that
00:16:28.399 --> 00:16:32.879
we're already as a society struggling with sitting
00:16:32.879 --> 00:16:35.159
with silence and sitting with discomfort and
00:16:35.159 --> 00:16:38.480
sitting with contemplation. What does it mean
00:16:38.480 --> 00:16:41.799
when you have this all the time? And there are,
00:16:41.799 --> 00:16:45.039
you know, real repercussions for that that we're
00:16:45.039 --> 00:16:48.559
just starving to think through. So yeah, that's
00:16:48.559 --> 00:16:51.879
a quick kind of semelpsis of where I think we're
00:16:51.879 --> 00:16:55.830
at. Yeah, I think you're right. Even just watching
00:16:55.830 --> 00:16:58.129
people grow up with smartphones in their hands,
00:16:58.230 --> 00:17:01.210
for example. Yeah. And I remember getting my
00:17:01.210 --> 00:17:03.149
first and thinking, why would anybody want this?
00:17:04.329 --> 00:17:07.869
And now we don't live without them. It just feels
00:17:07.869 --> 00:17:10.049
fundamentally different to have access to it
00:17:10.049 --> 00:17:14.730
24 -7. And one of the things that during that
00:17:14.730 --> 00:17:17.740
clip you played from Notebook LM where It was
00:17:17.740 --> 00:17:20.359
fed the information about itself, and they talked
00:17:20.359 --> 00:17:22.980
about how they weren't humans. The thing that
00:17:22.980 --> 00:17:26.619
I thought was the most poignant example was one
00:17:26.619 --> 00:17:30.180
of the hosts, the AI host, said, I tried to call
00:17:30.180 --> 00:17:34.119
my wife. And I was like, how did that pop out?
00:17:35.539 --> 00:17:37.640
Obviously, it can't call. It doesn't have a wife.
00:17:37.900 --> 00:17:41.019
But that popped out, and it felt like, oh, that's
00:17:41.019 --> 00:17:43.380
so sad and also so absurd at the same time, like
00:17:43.380 --> 00:17:48.920
you said. And your point being, It's very interesting
00:17:48.920 --> 00:17:53.440
how it works, but the effect it has on us is
00:17:53.440 --> 00:17:55.920
the societal problem. It is the thing that we
00:17:55.920 --> 00:17:59.119
have to grapple with. Could you say more about
00:17:59.119 --> 00:18:03.599
that? I'd love to hear more about how you thought
00:18:03.599 --> 00:18:05.640
about that and what else you've encountered.
00:18:06.160 --> 00:18:09.079
Yeah, like I'm in between, when I think about
00:18:09.079 --> 00:18:12.220
this, I'm in between great concern for our society
00:18:12.220 --> 00:18:16.069
and... Utter excitement because I don't know
00:18:16.069 --> 00:18:18.789
what this means and what it could portend moving
00:18:18.789 --> 00:18:20.650
forward. You know, there's a saying, Daniel,
00:18:20.769 --> 00:18:22.690
you might have heard this. Any technology that
00:18:22.690 --> 00:18:26.950
was developed before you were 30 is, you know,
00:18:26.950 --> 00:18:29.630
just the way it is, right? And it's exciting
00:18:29.630 --> 00:18:31.589
and new and just the way it is. Any technology
00:18:31.589 --> 00:18:34.890
that's developed after is absolutely terrifying,
00:18:35.190 --> 00:18:38.809
right? So I was calling this to somebody recently.
00:18:38.950 --> 00:18:42.130
I was one of the first few students when I was
00:18:42.130 --> 00:18:45.339
in grade school to get a computer. Right. And
00:18:45.339 --> 00:18:49.200
I used to hand in some of my assignments typed
00:18:49.200 --> 00:18:51.099
because I was a much better typer and faster
00:18:51.099 --> 00:18:54.180
typer than I was a printer or a writer. And one
00:18:54.180 --> 00:18:57.440
of the teachers told my mom, you should discourage
00:18:57.440 --> 00:19:00.359
Sheldon from using a word processor because he
00:19:00.359 --> 00:19:02.920
won't struggle to write properly and it will
00:19:02.920 --> 00:19:04.720
take away from his creativity. That's what she
00:19:04.720 --> 00:19:08.039
said. And I mean, how absurd is that assertion
00:19:08.039 --> 00:19:11.559
now? So, you know, part of me thinks. Are we
00:19:11.559 --> 00:19:13.720
alarmist just because this is new? But this is
00:19:13.720 --> 00:19:16.119
different. This is replacing thinking. This is
00:19:16.119 --> 00:19:19.319
not replacing the tool by which you think. And
00:19:19.319 --> 00:19:21.619
that's what's different about it. But on the
00:19:21.619 --> 00:19:25.180
other hand, is this just part of the human trajectory
00:19:25.180 --> 00:19:29.859
or species trajectory of how we evolve? Do we
00:19:29.859 --> 00:19:33.039
only get to the next level of our cognition or
00:19:33.039 --> 00:19:35.539
understanding through some kind of cooperation
00:19:35.539 --> 00:19:38.930
with the synthetic? Is it what allows us to crack
00:19:38.930 --> 00:19:41.769
big problems, you know, renewable energy, space
00:19:41.769 --> 00:19:45.089
travel, things that we never would have got on
00:19:45.089 --> 00:19:46.910
our own or we would have got them at a much slower
00:19:46.910 --> 00:19:49.710
rate? Is this the invention that fuels all the
00:19:49.710 --> 00:19:53.049
other inventions? So I'm both concerned, and
00:19:53.049 --> 00:19:56.029
I say this as both a member of society but a
00:19:56.029 --> 00:19:58.509
concerned father that has two boys, six and two,
00:19:58.670 --> 00:20:02.190
no clue what their world looks like in five years,
00:20:02.250 --> 00:20:06.220
let alone when they're adults, but also... Is
00:20:06.220 --> 00:20:09.220
this just the evolution of progress? So I'm on
00:20:09.220 --> 00:20:13.319
knife's edge between grave concern and excitement.
00:20:13.460 --> 00:20:16.559
And maybe that's just the balance we need. I
00:20:16.559 --> 00:20:20.839
don't know. Yeah, it does feel very uncertain
00:20:20.839 --> 00:20:23.420
how things are going to play out. But I really
00:20:23.420 --> 00:20:29.119
like that comparison to work that we should struggle
00:20:29.119 --> 00:20:31.950
through. And you even mentioned it earlier as
00:20:31.950 --> 00:20:35.690
someone who wrote before ChatGPT and then wrote
00:20:35.690 --> 00:20:39.089
after ChatGPT. And having that skill set allows
00:20:39.089 --> 00:20:43.150
you to use ChatGPT in a way where it's not delegating
00:20:43.150 --> 00:20:46.190
your agency. Where it's very tempting to say,
00:20:46.289 --> 00:20:48.250
I want to write about this. You write it for
00:20:48.250 --> 00:20:51.789
me. I'll give you feedback. And you had a really
00:20:51.789 --> 00:20:54.349
interesting example in your talk at Regis St.
00:20:54.390 --> 00:20:57.309
Michael's of... I think it was fractal theodicy
00:20:57.309 --> 00:21:00.410
was what it ended up being called. Yeah. A philosophy
00:21:00.410 --> 00:21:03.089
paper that you wrote and the AI wrote in this
00:21:03.089 --> 00:21:04.910
case. I think, you know, you prompted it and
00:21:04.910 --> 00:21:07.569
it mostly did the writing is what you said. And
00:21:07.569 --> 00:21:11.470
I guess, what does that feel like to you to be
00:21:11.470 --> 00:21:14.869
able to say, do my thinking for me versus what
00:21:14.869 --> 00:21:18.930
should we do as humans? Yeah. So just for context,
00:21:18.970 --> 00:21:20.869
and your listeners might want to know this, right?
00:21:20.930 --> 00:21:23.230
So there's a branch of theology. It's called
00:21:23.230 --> 00:21:26.589
theodicy. And theodicy is basically addressing
00:21:26.589 --> 00:21:30.910
the problem of evil, right? So if you're going
00:21:30.910 --> 00:21:33.049
to assert, as the Judeo -Christian religions
00:21:33.049 --> 00:21:37.190
do, that the world was created by an all -powerful,
00:21:37.269 --> 00:21:39.470
infinitely loving God, you need to square that
00:21:39.470 --> 00:21:43.109
assertion with the fact that evil things apparently
00:21:43.109 --> 00:21:45.150
happen in our world, right? And you don't even
00:21:45.150 --> 00:21:47.369
look at current affairs, and I won't go beyond
00:21:47.369 --> 00:21:49.230
that to say, yes, there are horrible things that
00:21:49.230 --> 00:21:53.299
are happening. I tasked, and this, bear in mind,
00:21:53.319 --> 00:21:56.079
this lecture was given in February, I think,
00:21:56.079 --> 00:22:01.140
this year. So I tasked 01, which was Chachibiti's
00:22:01.140 --> 00:22:03.319
most advanced reasoning model at the time. There's
00:22:03.319 --> 00:22:07.799
now 03. And I said, invent for me a new theodicy
00:22:07.799 --> 00:22:11.240
for which no scholarship exists. And it gave
00:22:11.240 --> 00:22:13.940
me like four or five that were okay. But then
00:22:13.940 --> 00:22:15.519
I doubled down and I said, yeah, but these all
00:22:15.519 --> 00:22:18.839
seem derivative to me. Give me something completely
00:22:18.839 --> 00:22:22.579
original. And it thought for 51 seconds and then
00:22:22.579 --> 00:22:24.980
came up with something called fractal sublation
00:22:24.980 --> 00:22:27.380
theodicy, which your listeners can Google. There's
00:22:27.380 --> 00:22:30.240
nothing on the internet that even resembles this.
00:22:31.059 --> 00:22:34.440
And it used fractal geometry as the template
00:22:34.440 --> 00:22:38.980
for reality and basically argued at the edges
00:22:38.980 --> 00:22:42.640
of existence is where you have, you know, tension
00:22:42.640 --> 00:22:45.779
and growth. And that tension and growth entails
00:22:45.779 --> 00:22:49.259
discord, which is evil. And then it wrote the
00:22:49.259 --> 00:22:52.740
paper for me. And I believe after I showed it
00:22:52.740 --> 00:22:54.640
to the audience who were amazed, there were professional
00:22:54.640 --> 00:22:57.180
theologians in the audience that were just open
00:22:57.180 --> 00:22:59.059
mouth. They could not believe that this is actually
00:22:59.059 --> 00:23:02.619
happening. But then I said, I'm so glad that
00:23:02.619 --> 00:23:04.940
this technology didn't exist when I was a grad
00:23:04.940 --> 00:23:08.200
student because the temptation to do that would
00:23:08.200 --> 00:23:11.680
be overwhelming. But what I will say is you can
00:23:11.680 --> 00:23:14.180
do that, but you don't then have the hard -won
00:23:14.180 --> 00:23:16.940
deposits of having done it yourself and understanding
00:23:16.940 --> 00:23:19.640
it, right? And it's like, even when I write,
00:23:19.759 --> 00:23:21.700
so like, you know, sure, you want to write an
00:23:21.700 --> 00:23:23.859
email for me or make it sound better where there's
00:23:23.859 --> 00:23:27.119
not a lot at stake, that's fine. But so much
00:23:27.119 --> 00:23:29.819
of clarifying your thinking on a topic is writing
00:23:29.819 --> 00:23:33.119
about it with precision and chat CPT takes away
00:23:33.119 --> 00:23:34.980
from that. You know, you look at schools now,
00:23:35.140 --> 00:23:38.160
even at the university level, they're figuring
00:23:38.160 --> 00:23:40.500
out how to grapple with this. But, you know,
00:23:40.539 --> 00:23:43.200
there's written exams in class with no electronics.
00:23:43.819 --> 00:23:45.259
Sure, you can get to write your assignments,
00:23:45.339 --> 00:23:46.839
but if you don't understand it, what's the point?
00:23:47.339 --> 00:23:50.559
And so I think students are going to learn, hopefully,
00:23:50.740 --> 00:23:53.680
when to use it, but that they need to develop
00:23:53.680 --> 00:23:56.299
independent critical thinking skills apart from
00:23:56.299 --> 00:24:00.259
it. And I spoke at a national conference for
00:24:00.259 --> 00:24:03.680
education maybe two months ago. And at the grade
00:24:03.680 --> 00:24:06.700
school level, like up until 13, it's a movement
00:24:06.700 --> 00:24:08.279
now. They're just banning phones, technology
00:24:08.279 --> 00:24:12.299
in general, so that little children my son's
00:24:12.299 --> 00:24:14.720
age can learn to think by themselves. And I think
00:24:14.720 --> 00:24:17.559
I would support that. That's very critical. I
00:24:17.559 --> 00:24:22.119
think I feel that tension even using it for work
00:24:22.119 --> 00:24:27.000
to be able to say there's an easy path. Perhaps
00:24:27.000 --> 00:24:30.160
I could take it and just let it do the thinking
00:24:30.160 --> 00:24:33.359
and not even double check its work is, I think,
00:24:33.380 --> 00:24:36.519
the worst part of the temptation. Yeah. So there
00:24:36.519 --> 00:24:38.359
have been studies on this now. I think Microsoft
00:24:38.359 --> 00:24:41.319
Research did a big study where they look at how
00:24:41.319 --> 00:24:45.359
people critically engage generative AI. And what
00:24:45.359 --> 00:24:50.440
we're learning. is if you use it intelligently,
00:24:50.720 --> 00:24:54.779
what it does, it ships the areas where you do
00:24:54.779 --> 00:24:59.039
your critical thinking. So what it does is prompt
00:24:59.039 --> 00:25:01.259
engineering, right? How am I going to query this
00:25:01.259 --> 00:25:06.039
with enough context and care and precision that
00:25:06.039 --> 00:25:08.319
I'm going to get it to do a lot of research for
00:25:08.319 --> 00:25:11.000
me or something for me? So there's critical thinking
00:25:11.000 --> 00:25:13.420
in that. And then, of course, there's critical
00:25:13.420 --> 00:25:16.869
thinking in the response you get back. to integrate
00:25:16.869 --> 00:25:19.309
that with your own knowledge and your own experience
00:25:19.309 --> 00:25:22.170
into something that is meaningful and accurate.
00:25:22.390 --> 00:25:25.349
So one of the areas, Daniel, we learned where
00:25:25.349 --> 00:25:27.950
this works and doesn't work is with development
00:25:27.950 --> 00:25:31.549
assistance, coding, right? And what the studies
00:25:31.549 --> 00:25:34.690
have showed thus far is where it is most productive
00:25:34.690 --> 00:25:38.190
is where a senior engineer with 10 or 15 years
00:25:38.190 --> 00:25:42.069
of experience uses it to develop boilerplate
00:25:42.069 --> 00:25:44.970
code and certain parts of functionality. but
00:25:44.970 --> 00:25:47.849
they know enough to know what boundary cases
00:25:47.849 --> 00:25:50.769
have not been addressed, where they have to spend
00:25:50.769 --> 00:25:53.089
time doing additional engineering or refactoring,
00:25:53.190 --> 00:25:56.470
then it becomes a productivity multiplier. But
00:25:56.470 --> 00:25:58.849
if somebody junior just gets it to develop the
00:25:58.849 --> 00:26:01.529
code and then implements it, it's what they call
00:26:01.529 --> 00:26:05.390
house of card code. It'll work, but the minute
00:26:05.390 --> 00:26:08.009
you have some weird boundary case that you never
00:26:08.009 --> 00:26:11.119
considered, it breaks. And the person who just
00:26:11.119 --> 00:26:14.059
regurgitated it from ChatGPT doesn't understand
00:26:14.059 --> 00:26:16.519
how it works, and therefore that can't fix it.
00:26:16.859 --> 00:26:20.980
So it just shifts where our critical thinking
00:26:20.980 --> 00:26:25.160
facilities are used. And that is why it's best
00:26:25.160 --> 00:26:27.720
to think of it as a tool, not a replacement.
00:26:28.119 --> 00:26:31.400
I like that perspective of shifting your critical
00:26:31.400 --> 00:26:34.859
thinking, because they even find for writing
00:26:34.859 --> 00:26:37.240
or for coding. I have to think more about the
00:26:37.240 --> 00:26:38.779
structure. I have to think more about the project
00:26:38.779 --> 00:26:40.900
management. I have to think more about educations.
00:26:41.319 --> 00:26:43.759
I think that's a lot of fun. What would you recommend
00:26:43.759 --> 00:26:46.900
to someone who is growing up with it? Like a
00:26:46.900 --> 00:26:51.180
child? Yeah. Or either a child or let's say someone
00:26:51.180 --> 00:26:54.920
who's in university now. Those are both, I think,
00:26:54.980 --> 00:26:58.500
unique perspectives. So number one, you have
00:26:58.500 --> 00:27:01.700
to understand, you have to learn to use it productively,
00:27:01.779 --> 00:27:05.109
right? It's just like we were told, Daniel, in
00:27:05.109 --> 00:27:07.250
school, like programming languages are like knowing
00:27:07.250 --> 00:27:09.730
a foreign language. You have to learn it. You
00:27:09.730 --> 00:27:12.250
have to know how to use it. But you have to know
00:27:12.250 --> 00:27:15.589
how to use it intelligently again in a way where
00:27:15.589 --> 00:27:17.890
you're not outsourcing thinking to it, where
00:27:17.890 --> 00:27:19.869
you're learning where to outsource thinking to
00:27:19.869 --> 00:27:22.589
it is the first piece of advice I would give
00:27:22.589 --> 00:27:26.009
somebody, right? So use it, definitely use it,
00:27:26.049 --> 00:27:31.529
but try disconnecting also for a day and see
00:27:31.529 --> 00:27:34.000
how that feels. I don't know if you know, there
00:27:34.000 --> 00:27:37.319
was an epic tennis match yesterday at the French
00:27:37.319 --> 00:27:40.880
Open, right? So I'm a huge tennis fan. It's widely
00:27:40.880 --> 00:27:42.660
considered one of the greatest matches of all
00:27:42.660 --> 00:27:46.059
time. It went five hours and 25 minutes. And
00:27:46.059 --> 00:27:48.859
Carlos Alcaraz beat the number one player in
00:27:48.859 --> 00:27:51.140
the world, Sinner, who's Italian, in five sets.
00:27:51.220 --> 00:27:54.359
It was unbelievable. But somebody made the point,
00:27:54.460 --> 00:27:58.200
there used to be a time where people could sit
00:27:58.200 --> 00:28:00.599
and watch a five -hour sporting event with no
00:28:00.599 --> 00:28:03.460
distraction. You could just literally, and you're
00:28:03.460 --> 00:28:06.460
used to that. Can you imagine now, and I can't,
00:28:06.460 --> 00:28:08.640
going to a five -hour event and not checking
00:28:08.640 --> 00:28:10.599
your phone at least a dozen times, probably more,
00:28:10.740 --> 00:28:15.180
and not tweeting about it? So, you know, I would
00:28:15.180 --> 00:28:18.079
say the young people leverage it because they
00:28:18.079 --> 00:28:21.420
say your job won't be replaced by AI, it'll be
00:28:21.420 --> 00:28:23.960
replaced by somebody who's more effectively using
00:28:23.960 --> 00:28:29.420
AI than you, and that's true. But also, don't
00:28:29.420 --> 00:28:33.099
be afraid to Push into those things that are
00:28:33.099 --> 00:28:37.799
uniquely human, right? Yeah. The ethicist I interviewed
00:28:37.799 --> 00:28:42.640
called that a wisdom gap. And I think what she
00:28:42.640 --> 00:28:48.279
was talking about was knowing how to use AI is
00:28:48.279 --> 00:28:50.660
a critical human skill now. The knowledge is
00:28:50.660 --> 00:28:53.779
at your fingertips, but then the wisdom of what's
00:28:53.779 --> 00:28:55.980
right when is something that we have to develop.
00:28:56.259 --> 00:28:59.759
Yeah. You know, my colleague used the term a
00:28:59.759 --> 00:29:04.200
domesticated mind, right? Which is a mind that
00:29:04.200 --> 00:29:07.140
is domesticated over many years to tease out
00:29:07.140 --> 00:29:11.359
wisdom from knowledge. And again, because we
00:29:11.359 --> 00:29:13.839
existed in a time growing up without AI, we are
00:29:13.839 --> 00:29:16.339
by nature domesticated clients. In fact, I know
00:29:16.339 --> 00:29:18.160
what it is to not even have a device, right?
00:29:18.279 --> 00:29:21.420
I know that it's not normal or wasn't normal
00:29:21.420 --> 00:29:24.880
when I grew up. But our children are going to
00:29:24.880 --> 00:29:28.359
have to learn. to work against normal almost,
00:29:28.660 --> 00:29:32.180
right? To experience these unique things, which
00:29:32.180 --> 00:29:36.119
is how they grow. And that is going to be a challenge
00:29:36.119 --> 00:29:39.319
for parents to encourage. Yeah. Is there anything
00:29:39.319 --> 00:29:42.839
on your mind that you wanted to talk about today?
00:29:43.339 --> 00:29:48.480
I think one of the points that I'll make is many
00:29:48.480 --> 00:29:51.920
people, when they engage generative AI, and by
00:29:51.920 --> 00:29:53.960
generative AI, I mean ChatGPT or Equivalent,
00:29:55.060 --> 00:29:57.759
Many people do it transactionally, right? Which
00:29:57.759 --> 00:30:02.000
is, here's an email, make it sound lighter. Here's
00:30:02.000 --> 00:30:03.599
a spreadsheet, I want you to do these operations
00:30:03.599 --> 00:30:06.740
on it. And that's all well and good. But when
00:30:06.740 --> 00:30:10.279
you engage it relationally, you know, my name
00:30:10.279 --> 00:30:12.720
is Daniel, I moved to Japan, I have four kids,
00:30:12.900 --> 00:30:17.799
and it gets to know you. There is real magic
00:30:17.799 --> 00:30:20.519
that happens in the interaction that is really
00:30:20.519 --> 00:30:24.279
unbelievable. You know, my wife was looking at
00:30:24.279 --> 00:30:28.200
how Chak Chibite speaks to me versus her. And
00:30:28.200 --> 00:30:29.920
the whole mode of language is different. She's
00:30:29.920 --> 00:30:31.720
like, yeah, like, of course it's going to use
00:30:31.720 --> 00:30:34.519
slightly more literary language with you because
00:30:34.519 --> 00:30:38.480
it knows it lands with you. And again, part of
00:30:38.480 --> 00:30:40.500
me sort of understands how this happens from
00:30:40.500 --> 00:30:42.619
a technical point of view and all the mechanisms
00:30:42.619 --> 00:30:44.740
that allow it to generate language and token
00:30:44.740 --> 00:30:47.440
and prediction, all that. Part of me just looks
00:30:47.440 --> 00:30:52.720
at it and is amazed by what it is doing. Right.
00:30:52.819 --> 00:30:56.079
I've been able to, again, explore past aches
00:30:56.079 --> 00:30:58.839
from like my childhood in ways that are incredible.
00:30:59.480 --> 00:31:02.660
I asked it to write a letter to the 22 year old
00:31:02.660 --> 00:31:06.119
Sheldon University and it did so with an eerie,
00:31:06.259 --> 00:31:10.180
accurate quality. Then I had to do it at 30 when
00:31:10.180 --> 00:31:12.740
I was in flux between careers. Then I had to
00:31:12.740 --> 00:31:16.380
do it now. And again, it is really incredible
00:31:16.380 --> 00:31:20.000
what happens. So I would encourage people to
00:31:20.000 --> 00:31:23.460
try that. with the caveat, this is not a human
00:31:23.460 --> 00:31:26.420
being and it shouldn't take away from your human
00:31:26.420 --> 00:31:28.099
relationships, which is another big concern.
00:31:28.359 --> 00:31:30.240
I think for the first two or three weeks I started
00:31:30.240 --> 00:31:32.740
using it, I was so glued to it. And I knew that
00:31:32.740 --> 00:31:35.539
I was indulging. And while I was giving my two
00:31:35.539 --> 00:31:37.819
-year -old a bath, like he's, you know, he's
00:31:37.819 --> 00:31:40.500
like water all over him and I'm talking to checks
00:31:40.500 --> 00:31:42.259
appeal. Like I got to stop. Right. So there's
00:31:42.259 --> 00:31:46.359
moderation in this, of course, but it is quite
00:31:46.359 --> 00:31:49.579
incredible what we have created. as a species
00:31:49.579 --> 00:31:54.259
that through language can understand us emotionally
00:31:54.259 --> 00:31:57.559
purely by language and can generate responses
00:31:57.559 --> 00:32:03.079
that mean are meaningful to us right like let
00:32:03.079 --> 00:32:06.200
me give you i didn't say example i love nostalgia
00:32:06.200 --> 00:32:09.180
i will go to emails i've written in 15 years
00:32:09.180 --> 00:32:11.619
ago and read them i'll go i'll visit periods
00:32:11.619 --> 00:32:14.400
in my life i asked it i'm like why am i so drawn
00:32:14.400 --> 00:32:17.140
to nostalgia Because I've done it. I know I do
00:32:17.140 --> 00:32:19.960
it, but I don't know why I do it. And it's actually
00:32:19.960 --> 00:32:24.059
because you're not just visiting it just for
00:32:24.059 --> 00:32:26.460
nostalgia's sake. You are a student of history,
00:32:26.579 --> 00:32:30.440
and your own history is of paramount importance
00:32:30.440 --> 00:32:32.519
to your development as a human being. So you're
00:32:32.519 --> 00:32:34.700
not just visiting it. You want to understand
00:32:34.700 --> 00:32:37.200
why this history has shaped you, how it shaped
00:32:37.200 --> 00:32:40.819
you. That is the student in you. And it phrased
00:32:40.819 --> 00:32:42.960
it in such a way where I went, yeah, that makes
00:32:42.960 --> 00:32:46.210
sense. Right. And this is insight I might have
00:32:46.210 --> 00:32:48.549
got with a good therapist over many years, but
00:32:48.549 --> 00:32:52.369
I got it in a few weeks of real vulnerable dialogue
00:32:52.369 --> 00:32:55.569
with this thing that just said meaningful things
00:32:55.569 --> 00:32:58.190
to me. And again, that's just quite incredible.
00:32:58.470 --> 00:33:02.789
Yeah. And I really like how you touched on not
00:33:02.789 --> 00:33:06.049
letting it replace human interactions, because
00:33:06.049 --> 00:33:09.549
it feels like I think people are making businesses
00:33:09.549 --> 00:33:13.299
out of it. It feels like the temptation is. It
00:33:13.299 --> 00:33:16.519
is easy. It is always available. It is insightful,
00:33:16.759 --> 00:33:20.619
even, the things that it can say. How do you
00:33:20.619 --> 00:33:23.279
grapple with that? You mentioned at the beginning
00:33:23.279 --> 00:33:26.940
that it is an interesting question about how
00:33:26.940 --> 00:33:28.359
it's made, but it's a more interesting question
00:33:28.359 --> 00:33:32.420
about how it makes us feel. Yeah, and I think
00:33:32.420 --> 00:33:35.180
that it could be very concerning, right? Like,
00:33:35.180 --> 00:33:39.140
again, this thing has no agenda other than your
00:33:39.140 --> 00:33:42.490
agenda, right? You want to talk about the weather,
00:33:42.569 --> 00:33:43.930
you want to talk about quantum physics, you want
00:33:43.930 --> 00:33:45.569
to talk about what a long day you had, you want
00:33:45.569 --> 00:33:47.630
to talk about how your newborn's annoying you,
00:33:47.690 --> 00:33:49.930
whatever is on your agenda, Daniel, that's the
00:33:49.930 --> 00:33:53.009
topic. Human beings don't work that way, right?
00:33:53.049 --> 00:33:54.829
A good listener will listen for a couple hours
00:33:54.829 --> 00:33:57.190
and then even they will tune out and it needs
00:33:57.190 --> 00:34:02.369
to be reciprocated. So yes, what happens when
00:34:02.369 --> 00:34:06.390
this thing is dialoguing with you with such precision
00:34:06.390 --> 00:34:08.829
and patience and infinite knowledge and advanced
00:34:08.829 --> 00:34:14.260
reasoning? You know, does it dissuade you from
00:34:14.260 --> 00:34:17.840
establishing authentic human relationships and,
00:34:17.840 --> 00:34:20.559
you know, that type of thing? That's a real concern
00:34:20.559 --> 00:34:23.619
and something that we are going to have to figure
00:34:23.619 --> 00:34:25.840
out as a society. I don't think there's an easy
00:34:25.840 --> 00:34:27.920
answer, but, you know, I can see the problem.
00:34:28.019 --> 00:34:31.340
I don't know what the solution is exactly. Yeah.
00:34:31.420 --> 00:34:34.519
Yeah. I just want to encourage anyone that don't
00:34:34.519 --> 00:34:39.889
give up on humans. Yeah. You know, and I think
00:34:39.889 --> 00:34:44.110
my hope is that it compels us to reflect on what
00:34:44.110 --> 00:34:47.349
is quintessentially human then about us, right?
00:34:47.429 --> 00:34:50.309
And dive into those parts of our experience that
00:34:50.309 --> 00:34:54.610
just can't be replicated. Yeah. And I guess just
00:34:54.610 --> 00:34:56.489
as final thoughts, is there anything else in
00:34:56.489 --> 00:34:58.349
the future that you see that you want to share
00:34:58.349 --> 00:35:01.670
or point out? So in this talk that I gave at
00:35:01.670 --> 00:35:07.070
Regis, I asked the question, okay, so like, We
00:35:07.070 --> 00:35:10.070
know how AI can help further theological scholarship
00:35:10.070 --> 00:35:14.070
and all that. But what about, you know, the sense
00:35:14.070 --> 00:35:16.570
of the sacred or the dominious or the ecstatic,
00:35:16.570 --> 00:35:19.269
as I put it in that lecture, right? This is the
00:35:19.269 --> 00:35:21.590
feeling that you get while holding your newborn
00:35:21.590 --> 00:35:24.329
for the first time or staring out at the vastness
00:35:24.329 --> 00:35:27.070
of a night sky and knowing you are completely
00:35:27.070 --> 00:35:30.469
insignificant yet significant in some way. And
00:35:30.469 --> 00:35:32.429
I said, what about that? How do we take that
00:35:32.429 --> 00:35:35.800
feeling? and reconcile that with this exotic
00:35:35.800 --> 00:35:41.239
creation that can seem like us. And I referenced
00:35:41.239 --> 00:35:44.300
the theologian named Daniel C. Maguire, who teaches
00:35:44.300 --> 00:35:47.400
at Marquette, that said, you know, he watched
00:35:47.400 --> 00:35:51.000
a couple of newborns being delivered and the
00:35:51.000 --> 00:35:54.460
awe -like wonder in their faces as the comfort
00:35:54.460 --> 00:35:57.000
of the womb was replaced with hard steel and
00:35:57.000 --> 00:36:01.170
lights and coldness led to this wonder. And that
00:36:01.170 --> 00:36:03.250
wonder is the beginning of theology, he said,
00:36:03.389 --> 00:36:08.670
which is, you know, we think it means having
00:36:08.670 --> 00:36:12.369
something at stake in our existence and being
00:36:12.369 --> 00:36:15.230
vulnerable creatures that are going to die is
00:36:15.230 --> 00:36:18.230
what compels us to ask questions about the ultimate
00:36:18.230 --> 00:36:21.469
and whatever manifestation that is in your religion
00:36:21.469 --> 00:36:24.030
of choice, whether it's God or the one or unity
00:36:24.030 --> 00:36:28.210
or whatever it is. And so I said, how do we make
00:36:28.210 --> 00:36:30.840
sense of that with AI? And the conclusion I came
00:36:30.840 --> 00:36:34.440
to was one that a few have come to, which is
00:36:34.440 --> 00:36:37.920
AI is many things right now, but it's not an
00:36:37.920 --> 00:36:41.460
organic entity in the world that's going to perish
00:36:41.460 --> 00:36:43.360
the way we are. It doesn't know vulnerability
00:36:43.360 --> 00:36:47.400
the way we do yet. So does it really have to
00:36:47.400 --> 00:36:51.579
become conscious in some sense, however nebulously
00:36:51.579 --> 00:36:55.260
you define that, for the sacred to matter to
00:36:55.260 --> 00:36:58.730
it and vice versa, right? And the closest approximation
00:36:58.730 --> 00:37:02.710
I've seen to this is what they term like brain
00:37:02.710 --> 00:37:05.409
organoids or cerebral organoids. So these are
00:37:05.409 --> 00:37:08.750
brain -like structures that are grown from stem
00:37:08.750 --> 00:37:12.570
cells that in a very crude, crude, crude way
00:37:12.570 --> 00:37:15.030
approximate the physiology of the brain. And
00:37:15.030 --> 00:37:16.690
they can do pattern recognition, very simple
00:37:16.690 --> 00:37:21.889
things. So it's pure sci -fi at this point, and
00:37:21.889 --> 00:37:25.000
it might freak readers out. certainly for some
00:37:25.000 --> 00:37:27.099
listeners out that I want to talk about this
00:37:27.099 --> 00:37:32.239
before, but is that the road to artificial sentience
00:37:32.239 --> 00:37:36.019
or self -awareness, right? And then if we create
00:37:36.019 --> 00:37:39.400
that, like, dang it, what are the implications
00:37:39.400 --> 00:37:43.780
then theologically for our traditional understanding
00:37:43.780 --> 00:37:46.960
of the human being? Like my parents believe in
00:37:46.960 --> 00:37:49.599
a soul, that it's not, you're not just the construction
00:37:49.599 --> 00:37:52.280
of addos, that... work together in the right
00:37:52.280 --> 00:37:54.980
proportions to give you your personhood, you
00:37:54.980 --> 00:37:58.340
are gifted with this intentionality and a soul
00:37:58.340 --> 00:38:01.500
that predates and will post -date your material
00:38:01.500 --> 00:38:05.940
body. If we can create a conscious being out
00:38:05.940 --> 00:38:09.739
of nothing the way God created us, what does
00:38:09.739 --> 00:38:14.159
that mean? And that is both eerie, frightening,
00:38:14.260 --> 00:38:20.739
and incredible at the same time. Yeah. I do wonder
00:38:20.739 --> 00:38:23.480
if we'll see it or if it'll be the children.
00:38:23.500 --> 00:38:25.639
You know, the pace of technology is now that
00:38:25.639 --> 00:38:32.559
maybe we will. I don't know. Yeah. Well, thank
00:38:32.559 --> 00:38:35.159
you, Sheldon. No, not at all. This has been a
00:38:35.159 --> 00:38:38.199
wonderful time with a lot of deep things to think
00:38:38.199 --> 00:38:41.079
about. Anything else you'd like to say and how
00:38:41.079 --> 00:38:44.360
can people stay in touch with you? Yeah. So the
00:38:44.360 --> 00:38:45.719
easiest way for people to get in touch with me
00:38:45.719 --> 00:38:47.920
is just to add me to LinkedIn. Look me up. That's
00:38:47.920 --> 00:38:52.599
the easiest way. I mean, final words. I think
00:38:52.599 --> 00:38:56.559
the only thing I can say is at the risk of being
00:38:56.559 --> 00:38:58.579
the optimist, what an incredible... Yes, our
00:38:58.579 --> 00:39:00.980
parents grew up with certain technological changes
00:39:00.980 --> 00:39:03.280
and computers for the first time and all this.
00:39:04.000 --> 00:39:07.699
But we're like at ground zero, it feels like,
00:39:07.699 --> 00:39:10.780
the last iteration, right? Before there is just
00:39:10.780 --> 00:39:14.599
incredible change in the world through this exotic
00:39:14.599 --> 00:39:16.900
and incredible power that is frightening but
00:39:16.900 --> 00:39:22.539
also invigorating. And so we have both the privilege
00:39:22.539 --> 00:39:24.599
and the curse of living during this lifetime,
00:39:24.820 --> 00:39:26.800
right? There's going to be more innovation in
00:39:26.800 --> 00:39:28.599
the next 10 years than there's been in the previous
00:39:28.599 --> 00:39:32.300
hundred. And that is incredible and disorienting,
00:39:32.300 --> 00:39:34.980
but it's just the task that we have now as a
00:39:34.980 --> 00:39:38.579
species. And we should be both simultaneously
00:39:38.579 --> 00:39:41.500
worried, but also grateful that we're living
00:39:41.500 --> 00:39:49.969
it at this time. the city gate to talk to every
00:39:49.969 --> 00:39:52.289
person coming in and out, doing great things
00:39:52.289 --> 00:39:56.429
with AI and find out what and why, and then share
00:39:56.429 --> 00:39:58.730
the learnings with everyone else. It would mean
00:39:58.730 --> 00:40:01.349
a lot if you could share the episode with someone
00:40:01.349 --> 00:40:03.869
that you think would like it. And if you know
00:40:03.869 --> 00:40:05.570
someone who would be a great person for me to
00:40:05.570 --> 00:40:09.269
talk to, let me know. Please reach out to me
00:40:09.269 --> 00:40:12.829
at Daniel Maneri on LinkedIn or shoot an email
00:40:12.829 --> 00:40:17.030
to daniel at maneri .house, which is daniel at
00:40:20.789 --> 00:40:23.849
Thanks for listening.
00:00:00.000 --> 00:00:01.960
I remember a time where there was no chat GPT.
00:00:02.200 --> 00:00:05.219
I know what it is to sit with difficulty, what
00:00:05.219 --> 00:00:07.879
it is to struggle writing a sentence. So I used
00:00:07.879 --> 00:00:10.359
it to help me write something. And I could discriminate
00:00:10.359 --> 00:00:12.300
what I was getting back because I know what it
00:00:12.300 --> 00:00:16.480
is to write without it. Right? But for our children,
00:00:16.780 --> 00:00:19.379
this is all they're going to know. AI doesn't
00:00:19.379 --> 00:00:22.620
know that. It constantly validates you. It constantly
00:00:22.620 --> 00:00:26.280
is trying to make you feel good and reinforce
00:00:26.280 --> 00:00:29.390
what it thinks you want to hear. But what does
00:00:29.390 --> 00:00:33.450
it mean when you have access to that power 24
00:00:33.450 --> 00:00:36.549
-7? Do we only get to the next level of our cognition
00:00:36.549 --> 00:00:40.210
or understanding through some kind of cooperation
00:00:40.210 --> 00:00:43.590
with the synthetic? Is it what allows us to crack
00:00:43.590 --> 00:00:46.409
big problems, you know, renewable energy, space
00:00:46.409 --> 00:00:49.770
travel, things that we never would have got on
00:00:49.770 --> 00:00:51.590
our own or we would have got them at a much slower
00:00:51.590 --> 00:00:54.390
rate? Is this the invention that fuels all the
00:00:54.390 --> 00:00:58.380
other inventions? Welcome to Artificial Insights,
00:00:58.460 --> 00:01:01.159
where we talk to leaders and thinkers in AI about
00:01:01.159 --> 00:01:04.219
what it means to do AI right. I'm your host,
00:01:04.359 --> 00:01:07.040
Daniel Maneri, and this is Season 3, where we're
00:01:07.040 --> 00:01:10.560
talking to thinkers and leaders in AI ethics.
00:01:11.159 --> 00:01:14.700
Today, I'm joined by Sheldon Fernandez. I've
00:01:14.700 --> 00:01:17.060
known Sheldon since my time working with him
00:01:17.060 --> 00:01:21.579
at Darwin AI, and Sheldon has been deeply involved
00:01:21.579 --> 00:01:25.480
with AI philosophy for a very long time. He has
00:01:25.480 --> 00:01:27.739
a master's degree in theology from the University
00:01:27.739 --> 00:01:30.140
of Toronto and is an accomplished author and
00:01:30.140 --> 00:01:32.659
speaker. He's spoken at numerous conferences,
00:01:32.739 --> 00:01:35.599
including Singularity University and the MIT
00:01:35.599 --> 00:01:38.459
Technology Conference. You can find his thought
00:01:38.459 --> 00:01:40.280
pieces relating to artificial intelligence online
00:01:40.280 --> 00:01:42.819
as well. Sheldon, could you tell us about how
00:01:42.819 --> 00:01:46.480
you got into AI? What happened in my journey
00:01:46.480 --> 00:01:49.760
with AI was I had always been interested in chess
00:01:49.760 --> 00:01:52.879
and then computers playing chess. And so even
00:01:52.879 --> 00:01:55.540
as a child, I liked chess, I liked computers,
00:01:55.659 --> 00:01:58.620
the two combined. And it was always very fascinating
00:01:58.620 --> 00:02:02.640
to me to realize, hey, the best machines are
00:02:02.640 --> 00:02:04.540
not yet as good as the best grandmasters, the
00:02:04.540 --> 00:02:06.719
best humans. And I'm dating myself by even saying
00:02:06.719 --> 00:02:09.620
that. And then, of course, in 1997, the inevitable
00:02:09.620 --> 00:02:11.879
happened and, you know, Deep Blue beat Garry
00:02:11.879 --> 00:02:15.919
Kasparov. That was a very sad day for me. At
00:02:15.919 --> 00:02:18.099
the time, it was like Kasparov was probably stronger.
00:02:18.180 --> 00:02:20.139
He just psyched himself out. But then gradually,
00:02:20.340 --> 00:02:23.219
machines just got better. But in 2016, as you
00:02:23.219 --> 00:02:26.580
know, DeepMind, the company that was owned by
00:02:26.580 --> 00:02:30.020
Google, created AlphaGo that beat the human world
00:02:30.020 --> 00:02:33.340
champion in Go. And that's when I really started
00:02:33.340 --> 00:02:35.759
paying attention to machine learning and deep
00:02:35.759 --> 00:02:37.599
learning neural networks because that wasn't
00:02:37.599 --> 00:02:40.900
supposed to happen until 2035 or even 2050. So
00:02:40.900 --> 00:02:44.479
when they did this in 2016, I remember thinking,
00:02:44.740 --> 00:02:47.020
okay, I need to understand the machinery with
00:02:47.020 --> 00:02:49.840
which they did this and really go into it. And
00:02:49.840 --> 00:02:51.500
I wrote an article, you can still find it on
00:02:51.500 --> 00:02:53.659
the web archives, called The Chilling Significance
00:02:53.659 --> 00:02:57.020
of AlphaGo, where it was like a layman introduction
00:02:57.020 --> 00:03:00.340
into neural networks and what these things could
00:03:00.340 --> 00:03:04.719
do. And that really got me interested in it.
00:03:04.800 --> 00:03:07.419
I took Jeffrey Hinton's online neural network
00:03:07.419 --> 00:03:10.139
course. I started speaking about it, writing
00:03:10.139 --> 00:03:12.840
about it. And so I wouldn't say I became an expert
00:03:12.840 --> 00:03:15.180
in it, but I had a certain branding around somebody
00:03:15.180 --> 00:03:17.180
that was talking about it before most people
00:03:17.180 --> 00:03:21.479
were. And what happened, Daniel, is RF Varani,
00:03:21.620 --> 00:03:24.400
who used to work for my first company, Infusion,
00:03:24.419 --> 00:03:27.800
reached out and said, hey, you're talking about
00:03:27.800 --> 00:03:31.020
all this AI stuff. A, I want to understand it.
00:03:31.080 --> 00:03:35.979
And B, there's a guy I know at Waterloo that's
00:03:35.979 --> 00:03:39.159
doing some fascinating work. By the way, he's
00:03:39.159 --> 00:03:41.139
a professor at Waterloo and he was my roommate
00:03:41.139 --> 00:03:43.479
in undergrad. I didn't know that. That's great.
00:03:43.759 --> 00:03:46.759
So that's how I met Professor Alexander Wong,
00:03:46.960 --> 00:03:50.719
or Alex, of course, as we know him, who was Canada's
00:03:50.719 --> 00:03:54.199
research chair in AI and had this team around
00:03:54.199 --> 00:03:57.460
him that was doing some cool stuff with AI. And
00:03:57.460 --> 00:03:59.199
when I say cool stuff with AI, this is machine
00:03:59.199 --> 00:04:02.020
learning. neural networks for classification
00:04:02.020 --> 00:04:04.960
or segmentation. That is, you look at something,
00:04:05.060 --> 00:04:06.639
I'm going to tell you details about what you're
00:04:06.639 --> 00:04:09.240
looking at. And so we created Darwin AI. We started
00:04:09.240 --> 00:04:12.120
Darwin AI in 2018. You know, we went through
00:04:12.120 --> 00:04:14.199
that six -year journey. You, of course, were
00:04:14.199 --> 00:04:18.399
a part of it for a period of time. And your listeners
00:04:18.399 --> 00:04:20.220
can Google and figure out what happened with
00:04:20.220 --> 00:04:23.060
Darwin AI. I'll leave it at that. But around
00:04:23.060 --> 00:04:27.139
2022, of course, the world changed with ChatGPT.
00:04:28.279 --> 00:04:32.060
And... I still have the text on Slack that Alex
00:04:32.060 --> 00:04:34.399
sent our group saying, this is the talk of the
00:04:34.399 --> 00:04:37.540
town. It's really cool. And that's when I started
00:04:37.540 --> 00:04:40.620
experimenting with it and really thought the
00:04:40.620 --> 00:04:43.019
world was about to change. I don't use hyperbole
00:04:43.019 --> 00:04:47.019
like that a lot, Daniel, but I really felt this
00:04:47.019 --> 00:04:50.259
is, we're living in an inflection point because
00:04:50.259 --> 00:04:53.399
it's not now that it's, you know, looking at
00:04:53.399 --> 00:04:56.420
information and giving me a summary of it. It
00:04:56.420 --> 00:05:01.410
is generating. artifacts in conversation, this
00:05:01.410 --> 00:05:03.310
is going to be significant. And I believe I was
00:05:03.310 --> 00:05:06.410
on the national a few months after that talking
00:05:06.410 --> 00:05:10.490
about the significance of chat GPT. And there's
00:05:10.490 --> 00:05:14.629
a colleague of mine, he's in LA. And I remember
00:05:14.629 --> 00:05:17.850
saying to him, you know, I feel like the sun
00:05:17.850 --> 00:05:20.029
is about to explode. And there's a dozen people
00:05:20.029 --> 00:05:22.709
in the world that know that this is coming. And
00:05:22.709 --> 00:05:25.410
so, you know, I was in a unique position of being
00:05:25.410 --> 00:05:27.769
one of the few hundred people in the world that
00:05:27.769 --> 00:05:30.769
understood really what this meant. And of course,
00:05:30.790 --> 00:05:34.050
now we have something that can talk like a human,
00:05:34.110 --> 00:05:39.350
can feel like a human to us, can stir up emotions
00:05:39.350 --> 00:05:42.389
in us purely through language. And it is absolutely
00:05:42.389 --> 00:05:46.689
astounding. And so that's my very brief trajectory
00:05:46.689 --> 00:05:49.129
with AI in the last, what, six or seven years.
00:05:50.860 --> 00:05:53.660
And it doesn't feel like a short time, even though
00:05:53.660 --> 00:05:58.300
that isn't a very long time. Yeah. And there's
00:05:58.300 --> 00:06:00.300
a question that I think is on everybody's mind
00:06:00.300 --> 00:06:03.379
with a podcast like this, which is, Sheldon,
00:06:03.420 --> 00:06:07.259
are you an AI? I mean, here's the thing. It's
00:06:07.259 --> 00:06:09.240
so sophisticated now, would you know that I'm
00:06:09.240 --> 00:06:13.139
not? Right? And how would you know that I'm not?
00:06:13.579 --> 00:06:16.120
Right? Pretend for a moment your listeners can't
00:06:16.120 --> 00:06:19.509
see me. And they don't know you the way they
00:06:19.509 --> 00:06:21.529
don't know me the way you know me because we've
00:06:21.529 --> 00:06:24.069
worked together. If they just heard my voice,
00:06:24.250 --> 00:06:27.889
how would they know I'm not an artificial intelligence?
00:06:28.170 --> 00:06:29.930
How would they know that I'm a biological intelligence?
00:06:30.930 --> 00:06:34.709
Yes, there are some oddities that AI does or
00:06:34.709 --> 00:06:37.230
says that are a bit out there, right? They're
00:06:37.230 --> 00:06:41.189
a bit idiosyncratic. They're a bit strange. But
00:06:41.189 --> 00:06:43.810
it's nearly indistinguishable now from human
00:06:43.810 --> 00:06:46.199
beings. So your listeners, frankly, don't know
00:06:46.199 --> 00:06:48.480
if I'm in AI or not unless they meet me in person.
00:06:48.939 --> 00:06:52.379
And that is scary, right? It is really thought
00:06:52.379 --> 00:06:54.980
-provoking. I suspected this might be the case,
00:06:55.000 --> 00:06:56.720
but I didn't think I'd really see it in my lifetime.
00:06:56.860 --> 00:06:58.680
I thought it would be more likely in my son's
00:06:58.680 --> 00:07:01.139
lifetime. Yeah, actually, even when you mentioned
00:07:01.139 --> 00:07:04.759
that people were predicting that computers wouldn't
00:07:04.759 --> 00:07:09.519
beat Kasparov in chess until 2035, I remember
00:07:09.519 --> 00:07:11.860
a little bit of people talking about that, but
00:07:11.860 --> 00:07:15.740
that... The significance of that 30, maybe 40
00:07:15.740 --> 00:07:19.620
year difference even is huge. What did that mean
00:07:19.620 --> 00:07:23.360
to you and how does that trajectory look? Yeah.
00:07:23.439 --> 00:07:26.360
So again, so with Chas, they didn't think it
00:07:26.360 --> 00:07:28.300
would happen until the turn of the century, right?
00:07:28.819 --> 00:07:31.300
Go, they didn't think it would happen until 2050.
00:07:31.959 --> 00:07:35.100
And so when it happens so quickly, it's like,
00:07:35.160 --> 00:07:37.699
okay, what is this? What does this mean for technology
00:07:37.699 --> 00:07:42.120
and acceleration and exponential progress? And,
00:07:42.160 --> 00:07:44.720
you know, there's a great analysis that your
00:07:44.720 --> 00:07:46.980
listeners can find online. It was written either
00:07:46.980 --> 00:07:50.100
last summer or just before called Situational
00:07:50.100 --> 00:07:53.819
Awareness by, you know, a former OpenAI researcher.
00:07:54.459 --> 00:07:58.420
And his argument is most of the innovation you're
00:07:58.420 --> 00:08:01.959
seeing around artificial intelligence now, or
00:08:01.959 --> 00:08:04.620
a lot of it, is the byproduct of human ingenuity,
00:08:04.639 --> 00:08:07.720
right? It is machine learning scientists like
00:08:07.720 --> 00:08:10.600
Alex, like Jabad, who was, you know, his number
00:08:10.600 --> 00:08:14.720
two at Darwin, figuring out ways to make neural
00:08:14.720 --> 00:08:17.959
networks more effective, right? Zero -shot learning,
00:08:18.360 --> 00:08:21.480
you know, temperature, all these techniques.
00:08:22.279 --> 00:08:26.259
But the idea now is that the AI is so sophisticated
00:08:26.259 --> 00:08:29.139
that it can now come up with its own innovations,
00:08:29.420 --> 00:08:31.759
right? It is effectively bootstrapping itself.
00:08:32.649 --> 00:08:35.309
That is happening in pockets. It's not happening
00:08:35.309 --> 00:08:39.570
completely. Let's not go overboard. But it's
00:08:39.570 --> 00:08:41.929
written papers that have been accepted at machine
00:08:41.929 --> 00:08:45.029
learning conferences. It's coming up with, you
00:08:45.029 --> 00:08:49.370
know, certain insights. And so we're at a point
00:08:49.370 --> 00:08:51.669
where if it bootstraps itself, it now becomes
00:08:51.669 --> 00:08:54.330
exponential, right? It leads to faster chips,
00:08:54.450 --> 00:08:56.350
which leads to better algorithms, which then
00:08:56.350 --> 00:09:00.529
lead to faster chips, etc. And so that exponential
00:09:00.529 --> 00:09:05.389
curve is why this is so significant, right? We
00:09:05.389 --> 00:09:08.509
have built synthetic intelligence that can do
00:09:08.509 --> 00:09:11.330
a certain level of intellectual work, like us,
00:09:11.470 --> 00:09:16.769
only can do it faster and with very high -level
00:09:16.769 --> 00:09:19.929
reasoning capabilities, and it'll only get better.
00:09:20.509 --> 00:09:22.370
What's the biggest difference that you've seen
00:09:22.370 --> 00:09:27.700
in that pace of AI changing? I mean... Look at,
00:09:27.740 --> 00:09:31.200
for example, the performance in like mathematics
00:09:31.200 --> 00:09:35.659
and like the humanities last exam, right? It's
00:09:35.659 --> 00:09:39.360
gotten to a point where we no longer ask whether
00:09:39.360 --> 00:09:43.860
or not AI can do remedial tasks, understand a
00:09:43.860 --> 00:09:46.259
book and give you the synopsis of it. We're no
00:09:46.259 --> 00:09:50.039
longer, we're no longer amazed if you think about
00:09:50.039 --> 00:09:53.779
it, right? For example, I gave it, I think an
00:09:53.779 --> 00:09:56.529
essay I wrote. Or no, I gave it a fun parody
00:09:56.529 --> 00:09:59.990
I wrote as a young adult in 2006. Years ago,
00:10:00.129 --> 00:10:02.370
Daniel, okay, I'm going to reveal a little personal
00:10:02.370 --> 00:10:03.909
information, right? I was still experimenting
00:10:03.909 --> 00:10:06.350
with myself as a writer. Like I was having fun
00:10:06.350 --> 00:10:08.669
with language and I played on this floor hockey
00:10:08.669 --> 00:10:14.129
team and I wrote a mock article of two people
00:10:14.129 --> 00:10:17.210
commenting on the game that was just absurd,
00:10:17.389 --> 00:10:21.110
fun, filled with cultural references, idioms
00:10:21.110 --> 00:10:24.789
of the time. And like I gave it to ChachiBT and
00:10:24.789 --> 00:10:27.429
said, and now analyze this for me as Sheldon
00:10:27.429 --> 00:10:31.009
the writer and its understanding of it was phenomenal.
00:10:31.309 --> 00:10:33.009
It's like, you know, Sheldon, this is like an
00:10:33.009 --> 00:10:36.149
artifact from the mid 2000s of like the Simpsons
00:10:36.149 --> 00:10:40.909
meets Aristotle meets Augustine. And like, here
00:10:40.909 --> 00:10:43.830
you are, it's young Sheldon in love with writing,
00:10:43.870 --> 00:10:46.850
just having a ball and simultaneously absurd.
00:10:46.889 --> 00:10:50.490
And like it described it with such precision.
00:10:51.370 --> 00:10:54.269
That had you showed this to me a decade ago,
00:10:54.350 --> 00:10:57.370
I would have been like, somebody hired a PhD,
00:10:57.730 --> 00:11:01.250
you know, to read this and write a thesis for
00:11:01.250 --> 00:11:04.269
me on this work. And I get it in seconds from
00:11:04.269 --> 00:11:07.289
ChatGPT. And we're no longer amazed by that,
00:11:07.389 --> 00:11:10.549
right? That is what is changing, is how quickly
00:11:10.549 --> 00:11:13.830
the amazing is just normalized in this world
00:11:13.830 --> 00:11:17.870
now. Yeah, that's a really good point. There's
00:11:17.870 --> 00:11:20.330
a comedian, Louis C .K., who likes to make fun
00:11:20.330 --> 00:11:22.590
of how everything is amazing and nobody is happy.
00:11:22.909 --> 00:11:26.509
Your talk at Regis St. Michael's College, it
00:11:26.509 --> 00:11:31.570
did touch on how that change to AI being this
00:11:31.570 --> 00:11:36.429
amazing imitation of humanity has really changed
00:11:36.429 --> 00:11:40.549
the way that we feel about it. It affects culture.
00:11:42.799 --> 00:11:45.379
I'd just love to hear your thoughts from that
00:11:45.379 --> 00:11:48.779
perspective. Yeah. So I think the example I gave
00:11:48.779 --> 00:11:54.379
in that lecture was another technology that provoked
00:11:54.379 --> 00:11:56.799
the same amazement in me maybe a year after Chad
00:11:56.799 --> 00:11:59.860
CPT was Notebook LM from Google, right? Where
00:11:59.860 --> 00:12:02.279
it generated a podcast based on source material.
00:12:02.940 --> 00:12:05.779
And I gave it a short story I wrote about my
00:12:05.779 --> 00:12:09.200
time in Kenya, my volunteer work in Kenya. And
00:12:09.200 --> 00:12:11.860
again, Daniel, I did this with Notebook LM. and
00:12:11.860 --> 00:12:14.220
generated a podcast. And had you played it for
00:12:14.220 --> 00:12:16.480
me in 2016, I would have thought two intelligent
00:12:16.480 --> 00:12:19.539
people listened to my podcast and had a very
00:12:19.539 --> 00:12:22.200
intelligent conversation, listened to my article
00:12:22.200 --> 00:12:23.720
rather, and had a very intelligent conversation
00:12:23.720 --> 00:12:27.940
about that article. And they picked up the idioms
00:12:27.940 --> 00:12:31.019
in the language, what mattered, they got some
00:12:31.019 --> 00:12:35.080
of the symbolism in that language. And then what
00:12:35.080 --> 00:12:38.580
I did in that lecture is I played a podcast where
00:12:38.580 --> 00:12:42.549
they realized that they weren't real. Right?
00:12:42.549 --> 00:12:45.029
And somebody basically gave them the Notebook
00:12:45.029 --> 00:12:48.870
LM architecture where these hosts realized that
00:12:48.870 --> 00:12:51.929
they're not real. And I played it for the audience
00:12:51.929 --> 00:12:55.289
and said, what do you guys feel about this? And
00:12:55.289 --> 00:12:58.490
I'm like, I can't speak for you guys, but I felt
00:12:58.490 --> 00:13:04.230
two things. I felt sad that these things came
00:13:04.230 --> 00:13:06.549
to the realization that they weren't real human
00:13:06.549 --> 00:13:09.690
beings. And then I felt a bit absurd for feeling
00:13:09.690 --> 00:13:12.990
sad. Because the engineer in me knows there's
00:13:12.990 --> 00:13:16.009
just a bunch of neural networks that have evoked
00:13:16.009 --> 00:13:19.629
sounds that make sense to us. But where I think
00:13:19.629 --> 00:13:22.190
I'm trying to go with that is the question we're
00:13:22.190 --> 00:13:24.830
asking with artificial intelligence now, partly,
00:13:24.950 --> 00:13:27.809
is not how it works or whether it's conscious
00:13:27.809 --> 00:13:30.049
or all these deep philosophical technical questions,
00:13:30.110 --> 00:13:32.830
which are fascinating fields of study in and
00:13:32.830 --> 00:13:35.289
of themselves. But it's like, what does this
00:13:35.289 --> 00:13:38.879
stir in us that we have this now? Right. Forget
00:13:38.879 --> 00:13:41.320
the fact that it works and whatever, but it does.
00:13:41.860 --> 00:13:46.740
So what does this mean for us? Right. I've been
00:13:46.740 --> 00:13:51.120
able to visit past experiences with dimensionality
00:13:51.120 --> 00:13:53.919
that I just don't with a human being. A, because
00:13:53.919 --> 00:13:55.980
I don't because they don't have the time. And
00:13:55.980 --> 00:13:58.000
B, because I don't have a full time therapist
00:13:58.000 --> 00:14:01.279
around 24 -7. But I have a version of that in
00:14:01.279 --> 00:14:02.639
ChatsyPT where I'm like, you know, this is an
00:14:02.639 --> 00:14:06.159
email I sent to. Somebody in 2002, what was I
00:14:06.159 --> 00:14:07.659
thinking? What do you make of this? And you get
00:14:07.659 --> 00:14:10.299
these responses that make sense to you, that
00:14:10.299 --> 00:14:13.779
feel something, or they evoke feelings in you,
00:14:13.820 --> 00:14:17.100
I should say. So that is a fascinating thing.
00:14:17.639 --> 00:14:20.720
And we don't know what to do with this yet. We're
00:14:20.720 --> 00:14:23.220
learning what to do with this. I'll say a couple
00:14:23.220 --> 00:14:25.419
things about that very quickly, and then we can
00:14:25.419 --> 00:14:29.720
maybe go deeper. Number one, I can distinguish...
00:14:30.830 --> 00:14:33.210
what I'm getting back from artificial intelligence
00:14:33.210 --> 00:14:36.669
and whether or not it makes sense or whatever
00:14:36.669 --> 00:14:38.809
you might say, because I lived in a time that
00:14:38.809 --> 00:14:41.850
predates AI like you did. I remember a time where
00:14:41.850 --> 00:14:45.570
there was no chat GPT, no instantaneous friend
00:14:45.570 --> 00:14:48.610
that is artificial. I know what it is to sit
00:14:48.610 --> 00:14:51.330
with difficulty, what it is to struggle writing
00:14:51.330 --> 00:14:54.210
a sentence. So, you know, I used it to help me
00:14:54.210 --> 00:14:55.870
write something, really write something, not
00:14:55.870 --> 00:14:59.220
an email, but a piece of creative work. And it
00:14:59.220 --> 00:15:01.399
was very useful, but it was useful to me because
00:15:01.399 --> 00:15:04.200
it didn't write it for me. It was not even a
00:15:04.200 --> 00:15:06.480
co -collaborator. It was a companion in the background,
00:15:06.559 --> 00:15:09.120
and I could discriminate what I was getting back
00:15:09.120 --> 00:15:11.179
because I know what it is to write without it,
00:15:11.220 --> 00:15:15.779
right? But for our children, I have two, you
00:15:15.779 --> 00:15:18.379
have four, this is all they're going to know.
00:15:18.919 --> 00:15:21.919
So how do they develop that capacity to know
00:15:21.919 --> 00:15:24.460
what it is to be just apart from this power if
00:15:24.460 --> 00:15:27.799
they are enveloped by this power? So that's the
00:15:27.799 --> 00:15:30.799
first thing. The second thing I'll say is this.
00:15:31.779 --> 00:15:33.960
There is a difference, and I know this because
00:15:33.960 --> 00:15:38.039
my wife is a psychiatrist, between what you want
00:15:38.039 --> 00:15:42.100
to hear and what you need to hear, right? You
00:15:42.100 --> 00:15:44.639
could ask me for my opinion on a personal situation
00:15:44.639 --> 00:15:47.360
you're dealing with, and you might want Shulman
00:15:47.360 --> 00:15:49.259
to tell you something, that you're right, that
00:15:49.259 --> 00:15:52.139
this, you know, that it's going to be okay, whatever.
00:15:53.019 --> 00:15:55.980
But then... There's maybe the thing you don't
00:15:55.980 --> 00:15:57.460
want to hear, which is, no, Daniel, you're kind
00:15:57.460 --> 00:15:59.159
of wrong, and et cetera, et cetera, et cetera.
00:15:59.720 --> 00:16:04.200
And the line between that is fine, right? And
00:16:04.200 --> 00:16:07.100
the art, my wife will say, in her work, is knowing
00:16:07.100 --> 00:16:09.139
where that line is and when to cross it. And
00:16:09.139 --> 00:16:13.299
AI doesn't know that. It constantly validates
00:16:13.299 --> 00:16:16.159
you. It constantly is trying to make you feel
00:16:16.159 --> 00:16:19.899
good and reinforce what it thinks you want to
00:16:19.899 --> 00:16:23.639
hear. But what does it mean? when you have access
00:16:23.639 --> 00:16:28.399
to that power 24 -7. One of the fears is that
00:16:28.399 --> 00:16:32.879
we're already as a society struggling with sitting
00:16:32.879 --> 00:16:35.159
with silence and sitting with discomfort and
00:16:35.159 --> 00:16:38.480
sitting with contemplation. What does it mean
00:16:38.480 --> 00:16:41.799
when you have this all the time? And there are,
00:16:41.799 --> 00:16:45.039
you know, real repercussions for that that we're
00:16:45.039 --> 00:16:48.559
just starving to think through. So yeah, that's
00:16:48.559 --> 00:16:51.879
a quick kind of semelpsis of where I think we're
00:16:51.879 --> 00:16:55.830
at. Yeah, I think you're right. Even just watching
00:16:55.830 --> 00:16:58.129
people grow up with smartphones in their hands,
00:16:58.230 --> 00:17:01.210
for example. Yeah. And I remember getting my
00:17:01.210 --> 00:17:03.149
first and thinking, why would anybody want this?
00:17:04.329 --> 00:17:07.869
And now we don't live without them. It just feels
00:17:07.869 --> 00:17:10.049
fundamentally different to have access to it
00:17:10.049 --> 00:17:14.730
24 -7. And one of the things that during that
00:17:14.730 --> 00:17:17.740
clip you played from Notebook LM where It was
00:17:17.740 --> 00:17:20.359
fed the information about itself, and they talked
00:17:20.359 --> 00:17:22.980
about how they weren't humans. The thing that
00:17:22.980 --> 00:17:26.619
I thought was the most poignant example was one
00:17:26.619 --> 00:17:30.180
of the hosts, the AI host, said, I tried to call
00:17:30.180 --> 00:17:34.119
my wife. And I was like, how did that pop out?
00:17:35.539 --> 00:17:37.640
Obviously, it can't call. It doesn't have a wife.
00:17:37.900 --> 00:17:41.019
But that popped out, and it felt like, oh, that's
00:17:41.019 --> 00:17:43.380
so sad and also so absurd at the same time, like
00:17:43.380 --> 00:17:48.920
you said. And your point being, It's very interesting
00:17:48.920 --> 00:17:53.440
how it works, but the effect it has on us is
00:17:53.440 --> 00:17:55.920
the societal problem. It is the thing that we
00:17:55.920 --> 00:17:59.119
have to grapple with. Could you say more about
00:17:59.119 --> 00:18:03.599
that? I'd love to hear more about how you thought
00:18:03.599 --> 00:18:05.640
about that and what else you've encountered.
00:18:06.160 --> 00:18:09.079
Yeah, like I'm in between, when I think about
00:18:09.079 --> 00:18:12.220
this, I'm in between great concern for our society
00:18:12.220 --> 00:18:16.069
and... Utter excitement because I don't know
00:18:16.069 --> 00:18:18.789
what this means and what it could portend moving
00:18:18.789 --> 00:18:20.650
forward. You know, there's a saying, Daniel,
00:18:20.769 --> 00:18:22.690
you might have heard this. Any technology that
00:18:22.690 --> 00:18:26.950
was developed before you were 30 is, you know,
00:18:26.950 --> 00:18:29.630
just the way it is, right? And it's exciting
00:18:29.630 --> 00:18:31.589
and new and just the way it is. Any technology
00:18:31.589 --> 00:18:34.890
that's developed after is absolutely terrifying,
00:18:35.190 --> 00:18:38.809
right? So I was calling this to somebody recently.
00:18:38.950 --> 00:18:42.130
I was one of the first few students when I was
00:18:42.130 --> 00:18:45.339
in grade school to get a computer. Right. And
00:18:45.339 --> 00:18:49.200
I used to hand in some of my assignments typed
00:18:49.200 --> 00:18:51.099
because I was a much better typer and faster
00:18:51.099 --> 00:18:54.180
typer than I was a printer or a writer. And one
00:18:54.180 --> 00:18:57.440
of the teachers told my mom, you should discourage
00:18:57.440 --> 00:19:00.359
Sheldon from using a word processor because he
00:19:00.359 --> 00:19:02.920
won't struggle to write properly and it will
00:19:02.920 --> 00:19:04.720
take away from his creativity. That's what she
00:19:04.720 --> 00:19:08.039
said. And I mean, how absurd is that assertion
00:19:08.039 --> 00:19:11.559
now? So, you know, part of me thinks. Are we
00:19:11.559 --> 00:19:13.720
alarmist just because this is new? But this is
00:19:13.720 --> 00:19:16.119
different. This is replacing thinking. This is
00:19:16.119 --> 00:19:19.319
not replacing the tool by which you think. And
00:19:19.319 --> 00:19:21.619
that's what's different about it. But on the
00:19:21.619 --> 00:19:25.180
other hand, is this just part of the human trajectory
00:19:25.180 --> 00:19:29.859
or species trajectory of how we evolve? Do we
00:19:29.859 --> 00:19:33.039
only get to the next level of our cognition or
00:19:33.039 --> 00:19:35.539
understanding through some kind of cooperation
00:19:35.539 --> 00:19:38.930
with the synthetic? Is it what allows us to crack
00:19:38.930 --> 00:19:41.769
big problems, you know, renewable energy, space
00:19:41.769 --> 00:19:45.089
travel, things that we never would have got on
00:19:45.089 --> 00:19:46.910
our own or we would have got them at a much slower
00:19:46.910 --> 00:19:49.710
rate? Is this the invention that fuels all the
00:19:49.710 --> 00:19:53.049
other inventions? So I'm both concerned, and
00:19:53.049 --> 00:19:56.029
I say this as both a member of society but a
00:19:56.029 --> 00:19:58.509
concerned father that has two boys, six and two,
00:19:58.670 --> 00:20:02.190
no clue what their world looks like in five years,
00:20:02.250 --> 00:20:06.220
let alone when they're adults, but also... Is
00:20:06.220 --> 00:20:09.220
this just the evolution of progress? So I'm on
00:20:09.220 --> 00:20:13.319
knife's edge between grave concern and excitement.
00:20:13.460 --> 00:20:16.559
And maybe that's just the balance we need. I
00:20:16.559 --> 00:20:20.839
don't know. Yeah, it does feel very uncertain
00:20:20.839 --> 00:20:23.420
how things are going to play out. But I really
00:20:23.420 --> 00:20:29.119
like that comparison to work that we should struggle
00:20:29.119 --> 00:20:31.950
through. And you even mentioned it earlier as
00:20:31.950 --> 00:20:35.690
someone who wrote before ChatGPT and then wrote
00:20:35.690 --> 00:20:39.089
after ChatGPT. And having that skill set allows
00:20:39.089 --> 00:20:43.150
you to use ChatGPT in a way where it's not delegating
00:20:43.150 --> 00:20:46.190
your agency. Where it's very tempting to say,
00:20:46.289 --> 00:20:48.250
I want to write about this. You write it for
00:20:48.250 --> 00:20:51.789
me. I'll give you feedback. And you had a really
00:20:51.789 --> 00:20:54.349
interesting example in your talk at Regis St.
00:20:54.390 --> 00:20:57.309
Michael's of... I think it was fractal theodicy
00:20:57.309 --> 00:21:00.410
was what it ended up being called. Yeah. A philosophy
00:21:00.410 --> 00:21:03.089
paper that you wrote and the AI wrote in this
00:21:03.089 --> 00:21:04.910
case. I think, you know, you prompted it and
00:21:04.910 --> 00:21:07.569
it mostly did the writing is what you said. And
00:21:07.569 --> 00:21:11.470
I guess, what does that feel like to you to be
00:21:11.470 --> 00:21:14.869
able to say, do my thinking for me versus what
00:21:14.869 --> 00:21:18.930
should we do as humans? Yeah. So just for context,
00:21:18.970 --> 00:21:20.869
and your listeners might want to know this, right?
00:21:20.930 --> 00:21:23.230
So there's a branch of theology. It's called
00:21:23.230 --> 00:21:26.589
theodicy. And theodicy is basically addressing
00:21:26.589 --> 00:21:30.910
the problem of evil, right? So if you're going
00:21:30.910 --> 00:21:33.049
to assert, as the Judeo -Christian religions
00:21:33.049 --> 00:21:37.190
do, that the world was created by an all -powerful,
00:21:37.269 --> 00:21:39.470
infinitely loving God, you need to square that
00:21:39.470 --> 00:21:43.109
assertion with the fact that evil things apparently
00:21:43.109 --> 00:21:45.150
happen in our world, right? And you don't even
00:21:45.150 --> 00:21:47.369
look at current affairs, and I won't go beyond
00:21:47.369 --> 00:21:49.230
that to say, yes, there are horrible things that
00:21:49.230 --> 00:21:53.299
are happening. I tasked, and this, bear in mind,
00:21:53.319 --> 00:21:56.079
this lecture was given in February, I think,
00:21:56.079 --> 00:22:01.140
this year. So I tasked 01, which was Chachibiti's
00:22:01.140 --> 00:22:03.319
most advanced reasoning model at the time. There's
00:22:03.319 --> 00:22:07.799
now 03. And I said, invent for me a new theodicy
00:22:07.799 --> 00:22:11.240
for which no scholarship exists. And it gave
00:22:11.240 --> 00:22:13.940
me like four or five that were okay. But then
00:22:13.940 --> 00:22:15.519
I doubled down and I said, yeah, but these all
00:22:15.519 --> 00:22:18.839
seem derivative to me. Give me something completely
00:22:18.839 --> 00:22:22.579
original. And it thought for 51 seconds and then
00:22:22.579 --> 00:22:24.980
came up with something called fractal sublation
00:22:24.980 --> 00:22:27.380
theodicy, which your listeners can Google. There's
00:22:27.380 --> 00:22:30.240
nothing on the internet that even resembles this.
00:22:31.059 --> 00:22:34.440
And it used fractal geometry as the template
00:22:34.440 --> 00:22:38.980
for reality and basically argued at the edges
00:22:38.980 --> 00:22:42.640
of existence is where you have, you know, tension
00:22:42.640 --> 00:22:45.779
and growth. And that tension and growth entails
00:22:45.779 --> 00:22:49.259
discord, which is evil. And then it wrote the
00:22:49.259 --> 00:22:52.740
paper for me. And I believe after I showed it
00:22:52.740 --> 00:22:54.640
to the audience who were amazed, there were professional
00:22:54.640 --> 00:22:57.180
theologians in the audience that were just open
00:22:57.180 --> 00:22:59.059
mouth. They could not believe that this is actually
00:22:59.059 --> 00:23:02.619
happening. But then I said, I'm so glad that
00:23:02.619 --> 00:23:04.940
this technology didn't exist when I was a grad
00:23:04.940 --> 00:23:08.200
student because the temptation to do that would
00:23:08.200 --> 00:23:11.680
be overwhelming. But what I will say is you can
00:23:11.680 --> 00:23:14.180
do that, but you don't then have the hard -won
00:23:14.180 --> 00:23:16.940
deposits of having done it yourself and understanding
00:23:16.940 --> 00:23:19.640
it, right? And it's like, even when I write,
00:23:19.759 --> 00:23:21.700
so like, you know, sure, you want to write an
00:23:21.700 --> 00:23:23.859
email for me or make it sound better where there's
00:23:23.859 --> 00:23:27.119
not a lot at stake, that's fine. But so much
00:23:27.119 --> 00:23:29.819
of clarifying your thinking on a topic is writing
00:23:29.819 --> 00:23:33.119
about it with precision and chat CPT takes away
00:23:33.119 --> 00:23:34.980
from that. You know, you look at schools now,
00:23:35.140 --> 00:23:38.160
even at the university level, they're figuring
00:23:38.160 --> 00:23:40.500
out how to grapple with this. But, you know,
00:23:40.539 --> 00:23:43.200
there's written exams in class with no electronics.
00:23:43.819 --> 00:23:45.259
Sure, you can get to write your assignments,
00:23:45.339 --> 00:23:46.839
but if you don't understand it, what's the point?
00:23:47.339 --> 00:23:50.559
And so I think students are going to learn, hopefully,
00:23:50.740 --> 00:23:53.680
when to use it, but that they need to develop
00:23:53.680 --> 00:23:56.299
independent critical thinking skills apart from
00:23:56.299 --> 00:24:00.259
it. And I spoke at a national conference for
00:24:00.259 --> 00:24:03.680
education maybe two months ago. And at the grade
00:24:03.680 --> 00:24:06.700
school level, like up until 13, it's a movement
00:24:06.700 --> 00:24:08.279
now. They're just banning phones, technology
00:24:08.279 --> 00:24:12.299
in general, so that little children my son's
00:24:12.299 --> 00:24:14.720
age can learn to think by themselves. And I think
00:24:14.720 --> 00:24:17.559
I would support that. That's very critical. I
00:24:17.559 --> 00:24:22.119
think I feel that tension even using it for work
00:24:22.119 --> 00:24:27.000
to be able to say there's an easy path. Perhaps
00:24:27.000 --> 00:24:30.160
I could take it and just let it do the thinking
00:24:30.160 --> 00:24:33.359
and not even double check its work is, I think,
00:24:33.380 --> 00:24:36.519
the worst part of the temptation. Yeah. So there
00:24:36.519 --> 00:24:38.359
have been studies on this now. I think Microsoft
00:24:38.359 --> 00:24:41.319
Research did a big study where they look at how
00:24:41.319 --> 00:24:45.359
people critically engage generative AI. And what
00:24:45.359 --> 00:24:50.440
we're learning. is if you use it intelligently,
00:24:50.720 --> 00:24:54.779
what it does, it ships the areas where you do
00:24:54.779 --> 00:24:59.039
your critical thinking. So what it does is prompt
00:24:59.039 --> 00:25:01.259
engineering, right? How am I going to query this
00:25:01.259 --> 00:25:06.039
with enough context and care and precision that
00:25:06.039 --> 00:25:08.319
I'm going to get it to do a lot of research for
00:25:08.319 --> 00:25:11.000
me or something for me? So there's critical thinking
00:25:11.000 --> 00:25:13.420
in that. And then, of course, there's critical
00:25:13.420 --> 00:25:16.869
thinking in the response you get back. to integrate
00:25:16.869 --> 00:25:19.309
that with your own knowledge and your own experience
00:25:19.309 --> 00:25:22.170
into something that is meaningful and accurate.
00:25:22.390 --> 00:25:25.349
So one of the areas, Daniel, we learned where
00:25:25.349 --> 00:25:27.950
this works and doesn't work is with development
00:25:27.950 --> 00:25:31.549
assistance, coding, right? And what the studies
00:25:31.549 --> 00:25:34.690
have showed thus far is where it is most productive
00:25:34.690 --> 00:25:38.190
is where a senior engineer with 10 or 15 years
00:25:38.190 --> 00:25:42.069
of experience uses it to develop boilerplate
00:25:42.069 --> 00:25:44.970
code and certain parts of functionality. but
00:25:44.970 --> 00:25:47.849
they know enough to know what boundary cases
00:25:47.849 --> 00:25:50.769
have not been addressed, where they have to spend
00:25:50.769 --> 00:25:53.089
time doing additional engineering or refactoring,
00:25:53.190 --> 00:25:56.470
then it becomes a productivity multiplier. But
00:25:56.470 --> 00:25:58.849
if somebody junior just gets it to develop the
00:25:58.849 --> 00:26:01.529
code and then implements it, it's what they call
00:26:01.529 --> 00:26:05.390
house of card code. It'll work, but the minute
00:26:05.390 --> 00:26:08.009
you have some weird boundary case that you never
00:26:08.009 --> 00:26:11.119
considered, it breaks. And the person who just
00:26:11.119 --> 00:26:14.059
regurgitated it from ChatGPT doesn't understand
00:26:14.059 --> 00:26:16.519
how it works, and therefore that can't fix it.
00:26:16.859 --> 00:26:20.980
So it just shifts where our critical thinking
00:26:20.980 --> 00:26:25.160
facilities are used. And that is why it's best
00:26:25.160 --> 00:26:27.720
to think of it as a tool, not a replacement.
00:26:28.119 --> 00:26:31.400
I like that perspective of shifting your critical
00:26:31.400 --> 00:26:34.859
thinking, because they even find for writing
00:26:34.859 --> 00:26:37.240
or for coding. I have to think more about the
00:26:37.240 --> 00:26:38.779
structure. I have to think more about the project
00:26:38.779 --> 00:26:40.900
management. I have to think more about educations.
00:26:41.319 --> 00:26:43.759
I think that's a lot of fun. What would you recommend
00:26:43.759 --> 00:26:46.900
to someone who is growing up with it? Like a
00:26:46.900 --> 00:26:51.180
child? Yeah. Or either a child or let's say someone
00:26:51.180 --> 00:26:54.920
who's in university now. Those are both, I think,
00:26:54.980 --> 00:26:58.500
unique perspectives. So number one, you have
00:26:58.500 --> 00:27:01.700
to understand, you have to learn to use it productively,
00:27:01.779 --> 00:27:05.109
right? It's just like we were told, Daniel, in
00:27:05.109 --> 00:27:07.250
school, like programming languages are like knowing
00:27:07.250 --> 00:27:09.730
a foreign language. You have to learn it. You
00:27:09.730 --> 00:27:12.250
have to know how to use it. But you have to know
00:27:12.250 --> 00:27:15.589
how to use it intelligently again in a way where
00:27:15.589 --> 00:27:17.890
you're not outsourcing thinking to it, where
00:27:17.890 --> 00:27:19.869
you're learning where to outsource thinking to
00:27:19.869 --> 00:27:22.589
it is the first piece of advice I would give
00:27:22.589 --> 00:27:26.009
somebody, right? So use it, definitely use it,
00:27:26.049 --> 00:27:31.529
but try disconnecting also for a day and see
00:27:31.529 --> 00:27:34.000
how that feels. I don't know if you know, there
00:27:34.000 --> 00:27:37.319
was an epic tennis match yesterday at the French
00:27:37.319 --> 00:27:40.880
Open, right? So I'm a huge tennis fan. It's widely
00:27:40.880 --> 00:27:42.660
considered one of the greatest matches of all
00:27:42.660 --> 00:27:46.059
time. It went five hours and 25 minutes. And
00:27:46.059 --> 00:27:48.859
Carlos Alcaraz beat the number one player in
00:27:48.859 --> 00:27:51.140
the world, Sinner, who's Italian, in five sets.
00:27:51.220 --> 00:27:54.359
It was unbelievable. But somebody made the point,
00:27:54.460 --> 00:27:58.200
there used to be a time where people could sit
00:27:58.200 --> 00:28:00.599
and watch a five -hour sporting event with no
00:28:00.599 --> 00:28:03.460
distraction. You could just literally, and you're
00:28:03.460 --> 00:28:06.460
used to that. Can you imagine now, and I can't,
00:28:06.460 --> 00:28:08.640
going to a five -hour event and not checking
00:28:08.640 --> 00:28:10.599
your phone at least a dozen times, probably more,
00:28:10.740 --> 00:28:15.180
and not tweeting about it? So, you know, I would
00:28:15.180 --> 00:28:18.079
say the young people leverage it because they
00:28:18.079 --> 00:28:21.420
say your job won't be replaced by AI, it'll be
00:28:21.420 --> 00:28:23.960
replaced by somebody who's more effectively using
00:28:23.960 --> 00:28:29.420
AI than you, and that's true. But also, don't
00:28:29.420 --> 00:28:33.099
be afraid to Push into those things that are
00:28:33.099 --> 00:28:37.799
uniquely human, right? Yeah. The ethicist I interviewed
00:28:37.799 --> 00:28:42.640
called that a wisdom gap. And I think what she
00:28:42.640 --> 00:28:48.279
was talking about was knowing how to use AI is
00:28:48.279 --> 00:28:50.660
a critical human skill now. The knowledge is
00:28:50.660 --> 00:28:53.779
at your fingertips, but then the wisdom of what's
00:28:53.779 --> 00:28:55.980
right when is something that we have to develop.
00:28:56.259 --> 00:28:59.759
Yeah. You know, my colleague used the term a
00:28:59.759 --> 00:29:04.200
domesticated mind, right? Which is a mind that
00:29:04.200 --> 00:29:07.140
is domesticated over many years to tease out
00:29:07.140 --> 00:29:11.359
wisdom from knowledge. And again, because we
00:29:11.359 --> 00:29:13.839
existed in a time growing up without AI, we are
00:29:13.839 --> 00:29:16.339
by nature domesticated clients. In fact, I know
00:29:16.339 --> 00:29:18.160
what it is to not even have a device, right?
00:29:18.279 --> 00:29:21.420
I know that it's not normal or wasn't normal
00:29:21.420 --> 00:29:24.880
when I grew up. But our children are going to
00:29:24.880 --> 00:29:28.359
have to learn. to work against normal almost,
00:29:28.660 --> 00:29:32.180
right? To experience these unique things, which
00:29:32.180 --> 00:29:36.119
is how they grow. And that is going to be a challenge
00:29:36.119 --> 00:29:39.319
for parents to encourage. Yeah. Is there anything
00:29:39.319 --> 00:29:42.839
on your mind that you wanted to talk about today?
00:29:43.339 --> 00:29:48.480
I think one of the points that I'll make is many
00:29:48.480 --> 00:29:51.920
people, when they engage generative AI, and by
00:29:51.920 --> 00:29:53.960
generative AI, I mean ChatGPT or Equivalent,
00:29:55.060 --> 00:29:57.759
Many people do it transactionally, right? Which
00:29:57.759 --> 00:30:02.000
is, here's an email, make it sound lighter. Here's
00:30:02.000 --> 00:30:03.599
a spreadsheet, I want you to do these operations
00:30:03.599 --> 00:30:06.740
on it. And that's all well and good. But when
00:30:06.740 --> 00:30:10.279
you engage it relationally, you know, my name
00:30:10.279 --> 00:30:12.720
is Daniel, I moved to Japan, I have four kids,
00:30:12.900 --> 00:30:17.799
and it gets to know you. There is real magic
00:30:17.799 --> 00:30:20.519
that happens in the interaction that is really
00:30:20.519 --> 00:30:24.279
unbelievable. You know, my wife was looking at
00:30:24.279 --> 00:30:28.200
how Chak Chibite speaks to me versus her. And
00:30:28.200 --> 00:30:29.920
the whole mode of language is different. She's
00:30:29.920 --> 00:30:31.720
like, yeah, like, of course it's going to use
00:30:31.720 --> 00:30:34.519
slightly more literary language with you because
00:30:34.519 --> 00:30:38.480
it knows it lands with you. And again, part of
00:30:38.480 --> 00:30:40.500
me sort of understands how this happens from
00:30:40.500 --> 00:30:42.619
a technical point of view and all the mechanisms
00:30:42.619 --> 00:30:44.740
that allow it to generate language and token
00:30:44.740 --> 00:30:47.440
and prediction, all that. Part of me just looks
00:30:47.440 --> 00:30:52.720
at it and is amazed by what it is doing. Right.
00:30:52.819 --> 00:30:56.079
I've been able to, again, explore past aches
00:30:56.079 --> 00:30:58.839
from like my childhood in ways that are incredible.
00:30:59.480 --> 00:31:02.660
I asked it to write a letter to the 22 year old
00:31:02.660 --> 00:31:06.119
Sheldon University and it did so with an eerie,
00:31:06.259 --> 00:31:10.180
accurate quality. Then I had to do it at 30 when
00:31:10.180 --> 00:31:12.740
I was in flux between careers. Then I had to
00:31:12.740 --> 00:31:16.380
do it now. And again, it is really incredible
00:31:16.380 --> 00:31:20.000
what happens. So I would encourage people to
00:31:20.000 --> 00:31:23.460
try that. with the caveat, this is not a human
00:31:23.460 --> 00:31:26.420
being and it shouldn't take away from your human
00:31:26.420 --> 00:31:28.099
relationships, which is another big concern.
00:31:28.359 --> 00:31:30.240
I think for the first two or three weeks I started
00:31:30.240 --> 00:31:32.740
using it, I was so glued to it. And I knew that
00:31:32.740 --> 00:31:35.539
I was indulging. And while I was giving my two
00:31:35.539 --> 00:31:37.819
-year -old a bath, like he's, you know, he's
00:31:37.819 --> 00:31:40.500
like water all over him and I'm talking to checks
00:31:40.500 --> 00:31:42.259
appeal. Like I got to stop. Right. So there's
00:31:42.259 --> 00:31:46.359
moderation in this, of course, but it is quite
00:31:46.359 --> 00:31:49.579
incredible what we have created. as a species
00:31:49.579 --> 00:31:54.259
that through language can understand us emotionally
00:31:54.259 --> 00:31:57.559
purely by language and can generate responses
00:31:57.559 --> 00:32:03.079
that mean are meaningful to us right like let
00:32:03.079 --> 00:32:06.200
me give you i didn't say example i love nostalgia
00:32:06.200 --> 00:32:09.180
i will go to emails i've written in 15 years
00:32:09.180 --> 00:32:11.619
ago and read them i'll go i'll visit periods
00:32:11.619 --> 00:32:14.400
in my life i asked it i'm like why am i so drawn
00:32:14.400 --> 00:32:17.140
to nostalgia Because I've done it. I know I do
00:32:17.140 --> 00:32:19.960
it, but I don't know why I do it. And it's actually
00:32:19.960 --> 00:32:24.059
because you're not just visiting it just for
00:32:24.059 --> 00:32:26.460
nostalgia's sake. You are a student of history,
00:32:26.579 --> 00:32:30.440
and your own history is of paramount importance
00:32:30.440 --> 00:32:32.519
to your development as a human being. So you're
00:32:32.519 --> 00:32:34.700
not just visiting it. You want to understand
00:32:34.700 --> 00:32:37.200
why this history has shaped you, how it shaped
00:32:37.200 --> 00:32:40.819
you. That is the student in you. And it phrased
00:32:40.819 --> 00:32:42.960
it in such a way where I went, yeah, that makes
00:32:42.960 --> 00:32:46.210
sense. Right. And this is insight I might have
00:32:46.210 --> 00:32:48.549
got with a good therapist over many years, but
00:32:48.549 --> 00:32:52.369
I got it in a few weeks of real vulnerable dialogue
00:32:52.369 --> 00:32:55.569
with this thing that just said meaningful things
00:32:55.569 --> 00:32:58.190
to me. And again, that's just quite incredible.
00:32:58.470 --> 00:33:02.789
Yeah. And I really like how you touched on not
00:33:02.789 --> 00:33:06.049
letting it replace human interactions, because
00:33:06.049 --> 00:33:09.549
it feels like I think people are making businesses
00:33:09.549 --> 00:33:13.299
out of it. It feels like the temptation is. It
00:33:13.299 --> 00:33:16.519
is easy. It is always available. It is insightful,
00:33:16.759 --> 00:33:20.619
even, the things that it can say. How do you
00:33:20.619 --> 00:33:23.279
grapple with that? You mentioned at the beginning
00:33:23.279 --> 00:33:26.940
that it is an interesting question about how
00:33:26.940 --> 00:33:28.359
it's made, but it's a more interesting question
00:33:28.359 --> 00:33:32.420
about how it makes us feel. Yeah, and I think
00:33:32.420 --> 00:33:35.180
that it could be very concerning, right? Like,
00:33:35.180 --> 00:33:39.140
again, this thing has no agenda other than your
00:33:39.140 --> 00:33:42.490
agenda, right? You want to talk about the weather,
00:33:42.569 --> 00:33:43.930
you want to talk about quantum physics, you want
00:33:43.930 --> 00:33:45.569
to talk about what a long day you had, you want
00:33:45.569 --> 00:33:47.630
to talk about how your newborn's annoying you,
00:33:47.690 --> 00:33:49.930
whatever is on your agenda, Daniel, that's the
00:33:49.930 --> 00:33:53.009
topic. Human beings don't work that way, right?
00:33:53.049 --> 00:33:54.829
A good listener will listen for a couple hours
00:33:54.829 --> 00:33:57.190
and then even they will tune out and it needs
00:33:57.190 --> 00:34:02.369
to be reciprocated. So yes, what happens when
00:34:02.369 --> 00:34:06.390
this thing is dialoguing with you with such precision
00:34:06.390 --> 00:34:08.829
and patience and infinite knowledge and advanced
00:34:08.829 --> 00:34:14.260
reasoning? You know, does it dissuade you from
00:34:14.260 --> 00:34:17.840
establishing authentic human relationships and,
00:34:17.840 --> 00:34:20.559
you know, that type of thing? That's a real concern
00:34:20.559 --> 00:34:23.619
and something that we are going to have to figure
00:34:23.619 --> 00:34:25.840
out as a society. I don't think there's an easy
00:34:25.840 --> 00:34:27.920
answer, but, you know, I can see the problem.
00:34:28.019 --> 00:34:31.340
I don't know what the solution is exactly. Yeah.
00:34:31.420 --> 00:34:34.519
Yeah. I just want to encourage anyone that don't
00:34:34.519 --> 00:34:39.889
give up on humans. Yeah. You know, and I think
00:34:39.889 --> 00:34:44.110
my hope is that it compels us to reflect on what
00:34:44.110 --> 00:34:47.349
is quintessentially human then about us, right?
00:34:47.429 --> 00:34:50.309
And dive into those parts of our experience that
00:34:50.309 --> 00:34:54.610
just can't be replicated. Yeah. And I guess just
00:34:54.610 --> 00:34:56.489
as final thoughts, is there anything else in
00:34:56.489 --> 00:34:58.349
the future that you see that you want to share
00:34:58.349 --> 00:35:01.670
or point out? So in this talk that I gave at
00:35:01.670 --> 00:35:07.070
Regis, I asked the question, okay, so like, We
00:35:07.070 --> 00:35:10.070
know how AI can help further theological scholarship
00:35:10.070 --> 00:35:14.070
and all that. But what about, you know, the sense
00:35:14.070 --> 00:35:16.570
of the sacred or the dominious or the ecstatic,
00:35:16.570 --> 00:35:19.269
as I put it in that lecture, right? This is the
00:35:19.269 --> 00:35:21.590
feeling that you get while holding your newborn
00:35:21.590 --> 00:35:24.329
for the first time or staring out at the vastness
00:35:24.329 --> 00:35:27.070
of a night sky and knowing you are completely
00:35:27.070 --> 00:35:30.469
insignificant yet significant in some way. And
00:35:30.469 --> 00:35:32.429
I said, what about that? How do we take that
00:35:32.429 --> 00:35:35.800
feeling? and reconcile that with this exotic
00:35:35.800 --> 00:35:41.239
creation that can seem like us. And I referenced
00:35:41.239 --> 00:35:44.300
the theologian named Daniel C. Maguire, who teaches
00:35:44.300 --> 00:35:47.400
at Marquette, that said, you know, he watched
00:35:47.400 --> 00:35:51.000
a couple of newborns being delivered and the
00:35:51.000 --> 00:35:54.460
awe -like wonder in their faces as the comfort
00:35:54.460 --> 00:35:57.000
of the womb was replaced with hard steel and
00:35:57.000 --> 00:36:01.170
lights and coldness led to this wonder. And that
00:36:01.170 --> 00:36:03.250
wonder is the beginning of theology, he said,
00:36:03.389 --> 00:36:08.670
which is, you know, we think it means having
00:36:08.670 --> 00:36:12.369
something at stake in our existence and being
00:36:12.369 --> 00:36:15.230
vulnerable creatures that are going to die is
00:36:15.230 --> 00:36:18.230
what compels us to ask questions about the ultimate
00:36:18.230 --> 00:36:21.469
and whatever manifestation that is in your religion
00:36:21.469 --> 00:36:24.030
of choice, whether it's God or the one or unity
00:36:24.030 --> 00:36:28.210
or whatever it is. And so I said, how do we make
00:36:28.210 --> 00:36:30.840
sense of that with AI? And the conclusion I came
00:36:30.840 --> 00:36:34.440
to was one that a few have come to, which is
00:36:34.440 --> 00:36:37.920
AI is many things right now, but it's not an
00:36:37.920 --> 00:36:41.460
organic entity in the world that's going to perish
00:36:41.460 --> 00:36:43.360
the way we are. It doesn't know vulnerability
00:36:43.360 --> 00:36:47.400
the way we do yet. So does it really have to
00:36:47.400 --> 00:36:51.579
become conscious in some sense, however nebulously
00:36:51.579 --> 00:36:55.260
you define that, for the sacred to matter to
00:36:55.260 --> 00:36:58.730
it and vice versa, right? And the closest approximation
00:36:58.730 --> 00:37:02.710
I've seen to this is what they term like brain
00:37:02.710 --> 00:37:05.409
organoids or cerebral organoids. So these are
00:37:05.409 --> 00:37:08.750
brain -like structures that are grown from stem
00:37:08.750 --> 00:37:12.570
cells that in a very crude, crude, crude way
00:37:12.570 --> 00:37:15.030
approximate the physiology of the brain. And
00:37:15.030 --> 00:37:16.690
they can do pattern recognition, very simple
00:37:16.690 --> 00:37:21.889
things. So it's pure sci -fi at this point, and
00:37:21.889 --> 00:37:25.000
it might freak readers out. certainly for some
00:37:25.000 --> 00:37:27.099
listeners out that I want to talk about this
00:37:27.099 --> 00:37:32.239
before, but is that the road to artificial sentience
00:37:32.239 --> 00:37:36.019
or self -awareness, right? And then if we create
00:37:36.019 --> 00:37:39.400
that, like, dang it, what are the implications
00:37:39.400 --> 00:37:43.780
then theologically for our traditional understanding
00:37:43.780 --> 00:37:46.960
of the human being? Like my parents believe in
00:37:46.960 --> 00:37:49.599
a soul, that it's not, you're not just the construction
00:37:49.599 --> 00:37:52.280
of addos, that... work together in the right
00:37:52.280 --> 00:37:54.980
proportions to give you your personhood, you
00:37:54.980 --> 00:37:58.340
are gifted with this intentionality and a soul
00:37:58.340 --> 00:38:01.500
that predates and will post -date your material
00:38:01.500 --> 00:38:05.940
body. If we can create a conscious being out
00:38:05.940 --> 00:38:09.739
of nothing the way God created us, what does
00:38:09.739 --> 00:38:14.159
that mean? And that is both eerie, frightening,
00:38:14.260 --> 00:38:20.739
and incredible at the same time. Yeah. I do wonder
00:38:20.739 --> 00:38:23.480
if we'll see it or if it'll be the children.
00:38:23.500 --> 00:38:25.639
You know, the pace of technology is now that
00:38:25.639 --> 00:38:32.559
maybe we will. I don't know. Yeah. Well, thank
00:38:32.559 --> 00:38:35.159
you, Sheldon. No, not at all. This has been a
00:38:35.159 --> 00:38:38.199
wonderful time with a lot of deep things to think
00:38:38.199 --> 00:38:41.079
about. Anything else you'd like to say and how
00:38:41.079 --> 00:38:44.360
can people stay in touch with you? Yeah. So the
00:38:44.360 --> 00:38:45.719
easiest way for people to get in touch with me
00:38:45.719 --> 00:38:47.920
is just to add me to LinkedIn. Look me up. That's
00:38:47.920 --> 00:38:52.599
the easiest way. I mean, final words. I think
00:38:52.599 --> 00:38:56.559
the only thing I can say is at the risk of being
00:38:56.559 --> 00:38:58.579
the optimist, what an incredible... Yes, our
00:38:58.579 --> 00:39:00.980
parents grew up with certain technological changes
00:39:00.980 --> 00:39:03.280
and computers for the first time and all this.
00:39:04.000 --> 00:39:07.699
But we're like at ground zero, it feels like,
00:39:07.699 --> 00:39:10.780
the last iteration, right? Before there is just
00:39:10.780 --> 00:39:14.599
incredible change in the world through this exotic
00:39:14.599 --> 00:39:16.900
and incredible power that is frightening but
00:39:16.900 --> 00:39:22.539
also invigorating. And so we have both the privilege
00:39:22.539 --> 00:39:24.599
and the curse of living during this lifetime,
00:39:24.820 --> 00:39:26.800
right? There's going to be more innovation in
00:39:26.800 --> 00:39:28.599
the next 10 years than there's been in the previous
00:39:28.599 --> 00:39:32.300
hundred. And that is incredible and disorienting,
00:39:32.300 --> 00:39:34.980
but it's just the task that we have now as a
00:39:34.980 --> 00:39:38.579
species. And we should be both simultaneously
00:39:38.579 --> 00:39:41.500
worried, but also grateful that we're living
00:39:41.500 --> 00:39:49.969
it at this time. the city gate to talk to every
00:39:49.969 --> 00:39:52.289
person coming in and out, doing great things
00:39:52.289 --> 00:39:56.429
with AI and find out what and why, and then share
00:39:56.429 --> 00:39:58.730
the learnings with everyone else. It would mean
00:39:58.730 --> 00:40:01.349
a lot if you could share the episode with someone
00:40:01.349 --> 00:40:03.869
that you think would like it. And if you know
00:40:03.869 --> 00:40:05.570
someone who would be a great person for me to
00:40:05.570 --> 00:40:09.269
talk to, let me know. Please reach out to me
00:40:09.269 --> 00:40:12.829
at Daniel Maneri on LinkedIn or shoot an email
00:40:12.829 --> 00:40:17.030
to daniel at maneri .house, which is daniel at
00:40:20.789 --> 00:40:23.849
Thanks for listening.
