1
00:00:03,300 --> 00:00:05,400
Welcome to make EdTech 100.
2
00:00:06,180 --> 00:00:10,530
I am LindyHoc Educator, K 12
Ed Tech Advisor, and your host.
3
00:00:11,340 --> 00:00:15,989
This is a podcast where we keep it real
about what actually works in classrooms.
4
00:00:16,440 --> 00:00:21,450
No hype, no overwhelm, just practical
strategies, honest stories and
5
00:00:21,450 --> 00:00:25,050
tools that make a real difference
for teachers and students.
6
00:00:25,110 --> 00:00:28,200
So come along with me on a
journey to make EdTech 100.
7
00:00:33,031 --> 00:00:35,431
Lindy Hockenbary: So if you didn't
know, and if you're not keeping
8
00:00:35,431 --> 00:00:40,081
up, there's a lot happening with AI
developments and advancements out there.
9
00:00:40,111 --> 00:00:45,901
So much so that I have only been barely
keeping up, and it is my job to keep up.
10
00:00:45,901 --> 00:00:48,721
So I know it's bad and
I know it's moving fast.
11
00:00:48,721 --> 00:00:50,431
I've been struggling
for the last few years.
12
00:00:50,851 --> 00:00:53,701
To keep up because of all the
advancements, but the last couple
13
00:00:53,701 --> 00:00:56,461
months especially have just been crazy.
14
00:00:56,461 --> 00:01:01,921
There's a lot of big
jumps in AI capabilities.
15
00:01:01,921 --> 00:01:05,191
So for this episode, I'm
doing a little bit different.
16
00:01:05,221 --> 00:01:09,901
Instead of having a guest that I
interview, I have an invited Erin on to
17
00:01:09,901 --> 00:01:16,080
be kind of like my co-host to help me
just talk through and unpack all of the
18
00:01:16,080 --> 00:01:19,201
AI craziness from the past few months.
19
00:01:19,201 --> 00:01:20,280
So, welcome, Erin.
20
00:01:21,159 --> 00:01:22,359
Aaron Makelky: Thank you,
Lindy, for having me.
21
00:01:22,359 --> 00:01:23,259
I'm flattered.
22
00:01:23,720 --> 00:01:26,211
Lindy Hockenbary: What do
listeners need to know about you?
23
00:01:27,259 --> 00:01:31,149
Aaron Makelky: Yeah, well I've been a
career educator, , in rural states as
24
00:01:31,149 --> 00:01:37,509
recently as the 2025 school year, but I've
since pivoted to developer relations and
25
00:01:37,509 --> 00:01:39,489
community marketing at a tech company.
26
00:01:40,179 --> 00:01:44,019
And I always tell people my skillset,
I'm not the most techie person.
27
00:01:44,019 --> 00:01:48,699
I'm not a developer or coder,
but I'm the conduit between those
28
00:01:48,699 --> 00:01:50,979
really techie developer types.
29
00:01:50,979 --> 00:01:55,119
And the average person, whether it's
a teacher, an administrator, or small
30
00:01:55,119 --> 00:01:59,199
business owner, I can speak the language
of both, and I'm somewhere in the middle.
31
00:02:00,536 --> 00:02:01,306
Lindy Hockenbary: Yep, same.
32
00:02:01,306 --> 00:02:02,116
Same with me.
33
00:02:02,116 --> 00:02:04,096
It's just I am very still much in.
34
00:02:04,446 --> 00:02:06,875
Education, career educators
still in education.
35
00:02:06,875 --> 00:02:12,006
You made the hop full on to work for
a tech company, but that's one of the
36
00:02:12,006 --> 00:02:13,446
reasons I wanted to bring you on today.
37
00:02:13,446 --> 00:02:17,766
First of all, you're one of the people
that I keep up with most on social media
38
00:02:17,766 --> 00:02:21,846
'cause you're always sharing really
awesome tips and how you're using ai.
39
00:02:21,846 --> 00:02:24,935
And when you were in the classroom
still, you were sharing so many great
40
00:02:24,935 --> 00:02:29,315
things about how you were embracing the
technology and teaching your students
41
00:02:29,315 --> 00:02:32,166
how to use it correctly and responsibly.
42
00:02:32,166 --> 00:02:33,041
I think you were my.
43
00:02:33,741 --> 00:02:39,500
Most followed person on LinkedIn in
2025 because you know, they did that
44
00:02:39,500 --> 00:02:41,870
little thing at the end of the year
where they told you who you like,
45
00:02:42,020 --> 00:02:43,640
who you like most engaged with.
46
00:02:43,640 --> 00:02:47,491
I'm pretty sure you were , my most
engaged with person, so I wanted to
47
00:02:47,491 --> 00:02:50,730
bring you on because you are threefold.
48
00:02:51,000 --> 00:02:54,840
You're doing a lot with the tech
yourself and keeping up with it yourself.
49
00:02:54,870 --> 00:02:58,650
You're also working for a tech
company now, and it's based
50
00:02:58,650 --> 00:02:59,970
out of the Bay Area, right?
51
00:03:00,439 --> 00:03:00,859
Aaron Makelky: It is.
52
00:03:00,859 --> 00:03:01,609
Yes.
53
00:03:01,900 --> 00:03:02,230
Lindy Hockenbary: Yeah.
54
00:03:02,230 --> 00:03:06,160
So like you are, even though you're
working remote from Wyoming, you're
55
00:03:06,160 --> 00:03:11,350
embedded in virtual calls all day long
with those that are like, you know,
56
00:03:11,410 --> 00:03:13,690
the tech people in Silicon Valley.
57
00:03:13,810 --> 00:03:15,700
So you kind of have that perspective now.
58
00:03:15,940 --> 00:03:18,700
And then also you have
the educator perspective.
59
00:03:18,700 --> 00:03:20,110
'cause how many years did you teach?
60
00:03:20,654 --> 00:03:23,119
Aaron Makelky: I was a
teacher for six 16 years.
61
00:03:23,209 --> 00:03:26,659
Uh, and I, I have my master's
in K 12 ed leadership and a
62
00:03:26,659 --> 00:03:30,779
principal endorsement, but always
preferred to be in the classroom.
63
00:03:30,839 --> 00:03:32,459
, Despite some of my family, my.
64
00:03:32,489 --> 00:03:34,499
Members that were career
administrators, I thought the
65
00:03:34,499 --> 00:03:36,059
classroom was a better fit for me.
66
00:03:36,750 --> 00:03:38,490
Lindy Hockenbary: Your dad
was an administrator, right?
67
00:03:38,919 --> 00:03:39,219
Aaron Makelky: Yep.
68
00:03:39,219 --> 00:03:41,349
And my identical principal.
69
00:03:41,349 --> 00:03:41,589
Yep.
70
00:03:41,640 --> 00:03:41,850
Lindy Hockenbary: right.
71
00:03:42,459 --> 00:03:43,749
Aaron Makelky: And my identical twin.
72
00:03:44,469 --> 00:03:44,919
Mm-hmm.
73
00:03:45,030 --> 00:03:46,680
Lindy Hockenbary: him on
LinkedIn too, and every now and
74
00:03:46,680 --> 00:03:48,090
again I'm like, oh, it's Aaron.
75
00:03:48,090 --> 00:03:49,020
No, it's not Aaron.
76
00:03:49,050 --> 00:03:49,620
No, it's not.
77
00:03:50,559 --> 00:03:52,089
Aaron Makelky: Same DNA different person,
78
00:03:53,250 --> 00:03:53,610
Lindy Hockenbary: person.
79
00:03:53,610 --> 00:03:54,000
I love it.
80
00:03:54,150 --> 00:03:56,790
I still have to meet
him in person sometime.
81
00:03:56,790 --> 00:03:57,720
E is it Eric?
82
00:03:58,449 --> 00:03:58,809
Aaron Makelky: Eric.
83
00:03:58,809 --> 00:03:59,139
Yep.
84
00:03:59,475 --> 00:03:59,925
Lindy Hockenbary: Eric?
85
00:03:59,925 --> 00:04:00,555
Eric and Aaron.
86
00:04:00,555 --> 00:04:01,155
Yeah, that's right.
87
00:04:01,215 --> 00:04:01,575
Okay.
88
00:04:01,965 --> 00:04:04,815
And he, yeah, he's educator
too slash administrator.
89
00:04:04,815 --> 00:04:05,115
So, yeah.
90
00:04:05,115 --> 00:04:09,165
So point being, you bring all three
of those different perspectives in
91
00:04:09,165 --> 00:04:14,415
to give listeners a really good idea
of Where, where we're at with ai.
92
00:04:14,415 --> 00:04:16,215
So I wanna start with a story.
93
00:04:16,425 --> 00:04:19,425
There's something about TCA conference.
94
00:04:19,425 --> 00:04:22,875
It's always like the first week
of February, and for the last.
95
00:04:23,535 --> 00:04:24,825
Two years.
96
00:04:25,515 --> 00:04:31,065
All of the big AI companies have released
like a big update like either right
97
00:04:31,065 --> 00:04:36,135
before right around February 1st, which
E either right before or right during
98
00:04:36,255 --> 00:04:43,275
TCA last year, open AI released operator,
which if you're not familiar, they've
99
00:04:43,275 --> 00:04:47,295
now rolled it into their agent feature,
so it's no longer called operator.
100
00:04:47,565 --> 00:04:50,655
But at the time it was
basically like their agentic.
101
00:04:52,305 --> 00:04:55,545
And if you don't know what agent is, we're
gonna dive into that, where you could
102
00:04:55,545 --> 00:05:02,025
actually tell it to go research a vacation
that you wanted to take or draft an email.
103
00:05:02,085 --> 00:05:05,175
I had it in my Gmail,
drafting my emails for me.
104
00:05:05,505 --> 00:05:08,865
Uh, and it would open up a browser
within the window and all these things.
105
00:05:09,075 --> 00:05:12,585
So I was frantically scrambling
at TCA next year, last year,
106
00:05:12,645 --> 00:05:15,765
excuse me, 2025 to make sure that.
107
00:05:16,175 --> 00:05:19,625
I was making sure every one of my
sessions, the people that came,
108
00:05:19,865 --> 00:05:25,775
knew what a agentic AI was and
very, very, very few people did.
109
00:05:25,985 --> 00:05:26,315
Okay.
110
00:05:26,315 --> 00:05:27,605
Fast forward a year.
111
00:05:27,845 --> 00:05:30,245
I was just at TCAA couple weeks ago.
112
00:05:30,245 --> 00:05:32,885
First week of February, same thing.
113
00:05:33,065 --> 00:05:34,985
Huge announcements coming in.
114
00:05:34,985 --> 00:05:38,645
So OpenAI release codex, which we're
gonna hopefully talk about your
115
00:05:38,885 --> 00:05:43,745
anthropic release, their Opus 4.6
model, which holy cow, I was actually
116
00:05:43,745 --> 00:05:45,425
texting my husband this morning.
117
00:05:45,780 --> 00:05:49,770
Telling him that he needs to go
check it out because now you can
118
00:05:49,770 --> 00:05:55,140
literally give it a Word document
and it will edit the actual Word
119
00:05:55,140 --> 00:05:57,360
document and spit it back out to you.
120
00:05:57,360 --> 00:05:59,790
So before, I would always
like upload a document.
121
00:06:00,060 --> 00:06:01,200
I'd ask for changes.
122
00:06:01,200 --> 00:06:01,980
It might give it to me.
123
00:06:01,980 --> 00:06:05,190
It might have given you a file that you
could download, but you still had to do
124
00:06:05,190 --> 00:06:06,960
a lot of like copy pasting, formatting.
125
00:06:07,020 --> 00:06:09,360
Nope, it does all the
formatting for you now.
126
00:06:09,360 --> 00:06:13,080
So that was released and then on
top of all of that, there was.
127
00:06:13,165 --> 00:06:18,955
All of this craziness going on
with Claude Bot, open Claw and mbo.
128
00:06:19,285 --> 00:06:21,775
And if you're not familiar, we're
gonna dive into that as well.
129
00:06:21,775 --> 00:06:27,085
So that was all going on in the midst
of TCAI was like sharing TikTok videos
130
00:06:27,085 --> 00:06:32,265
just to kind of stay on top of it and
inform my TikTok subscribers about it.
131
00:06:32,295 --> 00:06:34,845
And then so all of that
was happening, right?
132
00:06:35,320 --> 00:06:38,380
I am, I actually had time to go to
a few sessions at TCA this year.
133
00:06:38,380 --> 00:06:45,970
I'm sitting in a session agentic browsers
came up and the whole room, and it was a
134
00:06:45,970 --> 00:06:48,280
packed room and it was a pretty big room.
135
00:06:48,580 --> 00:06:50,800
Came to a halt with mass.
136
00:06:51,385 --> 00:06:56,725
Chaos for either a combination of, I
think the majority of the chaos was they
137
00:06:56,725 --> 00:06:58,315
had never heard the word ag agentic.
138
00:06:58,315 --> 00:07:00,625
They had no idea what an
age agentic browser meant.
139
00:07:00,905 --> 00:07:03,695
, The other little part of
the chaos was immediate.
140
00:07:03,695 --> 00:07:07,385
They knew it, but they didn't know
the feature that was being shown.
141
00:07:07,385 --> 00:07:09,875
And they immediately started
talking about how to block it.
142
00:07:10,175 --> 00:07:14,855
And the presenter did a really great
job of making sure that it didn't
143
00:07:14,885 --> 00:07:20,255
totally change the course of the
session, but I was just sitting there.
144
00:07:21,645 --> 00:07:27,945
going, oh my gosh, these like,
we have some major work to do.
145
00:07:27,945 --> 00:07:34,545
We've been over the last year, agent
a KI has been the big push behind
146
00:07:34,965 --> 00:07:37,006
a lot of the changes in updates.
147
00:07:37,006 --> 00:07:38,470
Would you agree or not agree, Erin?
148
00:07:39,644 --> 00:07:40,274
Aaron Makelky: Totally agree.
149
00:07:40,364 --> 00:07:45,674
I think a lot of us about two
years ago saw copy, paste, chat.
150
00:07:45,674 --> 00:07:49,214
GPT is in a tab or cloud,
whatever you're using, and we
151
00:07:49,214 --> 00:07:50,954
thought, wow, AI is so awesome.
152
00:07:51,524 --> 00:07:54,674
And the new version of
it is it can do things.
153
00:07:55,059 --> 00:07:56,889
On your behalf or for you?
154
00:07:57,399 --> 00:08:01,509
I'll share how I previewed
agen browsers with my students.
155
00:08:01,509 --> 00:08:04,929
I gave them a Google Doc in a
high school social studies class
156
00:08:05,379 --> 00:08:09,309
that had a list of eight or nine
questions about a historical topic.
157
00:08:09,309 --> 00:08:13,389
They didn't have a textbook or a
reading on, and they were supposed
158
00:08:13,389 --> 00:08:15,969
to go out and do some research,
and it's like a background
159
00:08:15,969 --> 00:08:17,619
builder on the League of Nations.
160
00:08:17,619 --> 00:08:18,244
When was it founded?
161
00:08:19,104 --> 00:08:20,934
Which country came up with the idea?
162
00:08:21,564 --> 00:08:26,844
And I said, I just wanna watch how you do
this to see, do you go to Google search?
163
00:08:26,844 --> 00:08:27,474
Do you go to chat?
164
00:08:27,474 --> 00:08:28,014
GPT?
165
00:08:28,014 --> 00:08:29,334
What tools do you use?
166
00:08:29,994 --> 00:08:34,884
But it was a trick because I knew most
of them opened a Google tab and they
167
00:08:34,884 --> 00:08:36,834
copy paste the question to Google search.
168
00:08:36,834 --> 00:08:38,124
A couple of them used ai.
169
00:08:39,084 --> 00:08:40,254
I let 'em finish.
170
00:08:40,644 --> 00:08:44,514
And then what I did is I took Comet
browser, I opened the same Google
171
00:08:44,514 --> 00:08:49,944
Doc and I said, go find the answers
and fill it in on my Google Doc.
172
00:08:50,064 --> 00:08:57,024
I hit run and then to be funny, I took
my Comet browser water bottle and I
173
00:08:57,024 --> 00:08:59,154
went down the hall to get a drink.
174
00:08:59,499 --> 00:09:02,649
Then I just watched my classroom through
the window, so I was still kind of
175
00:09:02,649 --> 00:09:07,659
supervising him and they're like, it's
opening tabs, it's researching, and then
176
00:09:07,659 --> 00:09:11,979
it's going back to the doc and it's typing
an like, is he remote controlling this?
177
00:09:11,979 --> 00:09:12,609
What the heck?
178
00:09:12,609 --> 00:09:14,829
And then I would just wave
at him through the window.
179
00:09:15,669 --> 00:09:17,529
And the reason I did that was 'cause.
180
00:09:17,869 --> 00:09:19,369
It was gonna show up eventually.
181
00:09:19,369 --> 00:09:20,839
Now it's in Google Chrome.
182
00:09:20,929 --> 00:09:22,999
Claude has a Chrome extension.
183
00:09:22,999 --> 00:09:26,509
So if you're a Claude person, you
get essentially the same thing
184
00:09:27,019 --> 00:09:28,639
and it can control your browser.
185
00:09:28,639 --> 00:09:32,299
And that was my moment of chaos where
people thought it was a trick or it
186
00:09:32,299 --> 00:09:34,369
wasn't real, and it's just like, nope.
187
00:09:34,399 --> 00:09:35,419
That's what they can do now.
188
00:09:35,449 --> 00:09:36,829
'cause they're built into your browser.
189
00:09:37,560 --> 00:09:37,771
Lindy Hockenbary: Yep.
190
00:09:37,861 --> 00:09:41,161
And yeah, if you've, if you've never
seen an agentic browser or any sort
191
00:09:41,161 --> 00:09:46,171
of AI agent or agentic ai, it really
does at first seem like a magic
192
00:09:46,171 --> 00:09:48,031
trick, like someone is pranking you.
193
00:09:48,031 --> 00:09:53,191
My students used to prank me all
the time and like set sticky keys.
194
00:09:53,411 --> 00:09:55,961
On the keyboards and the
computers in my classroom.
195
00:09:56,290 --> 00:10:00,191
And then, you know, somebody would it
either I caught onto it pretty quickly,
196
00:10:00,191 --> 00:10:02,711
but then like somebody else would come
in and start using the computer and
197
00:10:02,711 --> 00:10:04,471
think that the computer is possessed.
198
00:10:04,531 --> 00:10:06,121
You know, 'cause they couldn't
figure out the sticky keys.
199
00:10:06,271 --> 00:10:11,821
It's kind of like that experience, but
like way more intense than sticky keys.
200
00:10:12,061 --> 00:10:17,131
And if you're not familiar, so
a agentic AI is basically ai.
201
00:10:17,596 --> 00:10:21,586
That acts, and I always tell people when
I say this, this creates a lot of anxiety.
202
00:10:21,586 --> 00:10:22,756
So I'm always like, take a deep breath.
203
00:10:23,241 --> 00:10:25,281
Seriously, take a deep breath.
204
00:10:26,031 --> 00:10:27,741
It acts autonomously.
205
00:10:27,861 --> 00:10:31,341
So exactly what Aaron was saying,
you're not now just like going to chat
206
00:10:31,461 --> 00:10:36,111
BT or Claude or Gemini and giving it
a prompt and copy pasting over into
207
00:10:36,111 --> 00:10:38,961
a Google Doc, like it's editing and.
208
00:10:39,251 --> 00:10:44,741
Adding to the Google Doc for you, it's
autonomous, but you gave the perfect
209
00:10:44,741 --> 00:10:49,211
example of, I always add the caveat
that it's autonomous with oversight
210
00:10:49,211 --> 00:10:53,441
because you gave it the Google Doc,
you gave it the questions in the Google
211
00:10:53,441 --> 00:10:57,251
Doc that you wanted to do, so you
gave it instructions as the human to
212
00:10:57,281 --> 00:10:58,961
say, this is what I want you to do.
213
00:10:58,961 --> 00:11:03,401
I want you to answer these questions about
the League of Nations in this Google Doc.
214
00:11:03,401 --> 00:11:03,700
Aaron Makelky: Yep.
215
00:11:03,731 --> 00:11:07,511
And the way that I think of it,
the super simple, not as nuanced
216
00:11:07,511 --> 00:11:11,170
definition is agent AI has tools.
217
00:11:11,170 --> 00:11:12,611
So it's kinda like if you gave chat.
218
00:11:13,031 --> 00:11:17,981
GPTA toolkit and it has a wrench for
this job and a hammer for this job only.
219
00:11:17,981 --> 00:11:23,440
It might be a browser and it can do
a search, and then it can call this
220
00:11:23,440 --> 00:11:26,501
service over here and then it can
write to your word or your Google
221
00:11:26,501 --> 00:11:29,651
doc, and it's giving the same brain.
222
00:11:29,680 --> 00:11:30,881
That's the ai.
223
00:11:31,005 --> 00:11:33,015
The tools and now it's an agent.
224
00:11:33,435 --> 00:11:37,095
So it could be a recurring schedule
or a trigger like when I get an email
225
00:11:37,095 --> 00:11:41,905
from this person, do this, or Monday
mornings at a 8:00 AM run this, , or
226
00:11:41,905 --> 00:11:45,625
like the browser example, you open
up the tabs, it needs for context
227
00:11:45,625 --> 00:11:47,070
and say, go off and do this for me..
228
00:11:47,857 --> 00:11:48,097
Lindy Hockenbary: Yep.
229
00:11:48,187 --> 00:11:48,577
Exactly.
230
00:11:48,577 --> 00:11:51,517
And I always get the question at this
point, they're like, okay, well what
231
00:11:51,577 --> 00:11:55,807
they still don't, if you're new to Agent
ai, you may not still see the difference
232
00:11:55,807 --> 00:12:02,747
between a generative AI tool like chat,
GBT, Jim and I, Claude An agentic AI
233
00:12:02,747 --> 00:12:07,097
tool or an AI agent, which there are
subtle differences between the two,
234
00:12:07,097 --> 00:12:09,197
which we probably won't get into today.
235
00:12:09,457 --> 00:12:15,787
, , the way I explain it is , let's use email
for an example chat, GPT, generative ai.
236
00:12:15,787 --> 00:12:19,537
You have to go open the large
language model, or you might
237
00:12:19,537 --> 00:12:21,667
have it built in to like Gmail.
238
00:12:21,667 --> 00:12:23,977
Now there's some generative
built in there, right?
239
00:12:24,187 --> 00:12:29,737
But you still have to prompt it to say,
write this email, review it, and hit send.
240
00:12:30,137 --> 00:12:32,837
Agentic ai, you tell it.
241
00:12:32,927 --> 00:12:35,207
I want you to respond to my emails.
242
00:12:35,387 --> 00:12:39,137
Tell everyone that I'm gonna be out next
week at the NCC Conference in Seattle.
243
00:12:39,467 --> 00:12:45,827
And it goes, opens the email,
drafts that response and hits send.
244
00:12:46,157 --> 00:12:47,357
That's agentic ai.
245
00:12:47,357 --> 00:12:50,387
That's that autonomous part to it.
246
00:12:50,387 --> 00:12:54,077
Aaron Makelky: Which is a scary thing
'cause people always go, what if it sends
247
00:12:54,077 --> 00:12:55,817
the wrong message to the wrong person?
248
00:12:55,817 --> 00:12:58,637
Or books the wrong flight for
me, or whatever task you gave it.
249
00:12:59,504 --> 00:13:02,684
Lindy Hockenbary: It is scary and you
know, it's, there's a lot of discussion.
250
00:13:02,684 --> 00:13:05,054
We'll get into this a little bit
more when we talk about the claw
251
00:13:05,054 --> 00:13:09,494
bot, mot bot, whatever they're
calling it now thing, but it's also.
252
00:13:09,854 --> 00:13:14,144
Uh, it opens the question about
security risk and data privacy, right?
253
00:13:14,144 --> 00:13:18,194
Like now you're giving, you, you,
we've already been talking for years
254
00:13:18,224 --> 00:13:22,244
three over three years since chat
GPD was released about not putting
255
00:13:22,244 --> 00:13:25,454
an education like PII, making sure
you're a FERPA compliant, right?
256
00:13:25,454 --> 00:13:29,294
Don't put PII into a
large language model now.
257
00:13:30,389 --> 00:13:35,909
It's like 20 steps more than that of not
only just don't give it your information,
258
00:13:35,909 --> 00:13:40,239
but now you're literally giving the , AI
access to things like your email that
259
00:13:40,239 --> 00:13:46,299
might have really sensitive information
in it, depending upon what your job is.
260
00:13:46,299 --> 00:13:49,749
But if you're a teacher or a
school administrator, your email,
261
00:13:49,929 --> 00:13:53,139
your, if you use Google Drive
or OneDrive, you're giving it.
262
00:13:53,459 --> 00:13:56,999
Access to that potentially
has a lot of sensitive info.
263
00:13:56,999 --> 00:14:02,489
So that's a huge discussion around agent
to AI and education is that data privacy.
264
00:14:02,489 --> 00:14:05,999
And that's probably
one of the reasons why.
265
00:14:06,804 --> 00:14:10,374
All those educators, not all the
majority of the educators in that
266
00:14:10,374 --> 00:14:12,684
session room at TCA, just had no idea.
267
00:14:12,684 --> 00:14:15,324
Like I even heard someone
say, I'm so confused.
268
00:14:15,354 --> 00:14:17,964
She was looking around like this
because she was like, she had
269
00:14:17,964 --> 00:14:19,254
never even heard the word ag.
270
00:14:19,254 --> 00:14:21,834
And then you put Ag Agentic browser
in there and you know, a whole
271
00:14:21,834 --> 00:14:27,484
nother level of confusion because
it's no secret that K 12 education
272
00:14:27,484 --> 00:14:30,274
has a lot of barriers to innovation.
273
00:14:31,309 --> 00:14:34,339
One of those being ferpa, data privacy.
274
00:14:34,339 --> 00:14:41,539
We're working with miners under the age
of 18, so the tech tends to move slower.
275
00:14:42,019 --> 00:14:46,589
And not just that, but other reasons as
well than it does in other industries in.
276
00:14:46,589 --> 00:14:46,859
Aaron Makelky: Yep.
277
00:14:47,039 --> 00:14:51,179
And I think if you haven't come to
this realization yet, it's pretty
278
00:14:51,179 --> 00:14:58,019
unanimous by people who see this stuff
every day in 2026 if you're relying
279
00:14:58,019 --> 00:15:02,849
on anything in a web browser or on
a computer to assess your students'
280
00:15:02,849 --> 00:15:05,399
learning or track their progress.
281
00:15:05,909 --> 00:15:08,274
I'm not saying they're all using Chrome.
282
00:15:09,189 --> 00:15:10,959
Having an agent do it for them.
283
00:15:10,959 --> 00:15:14,919
I'm saying they very well
could be even with free tools.
284
00:15:15,459 --> 00:15:17,529
So if it's something in
a Google Doc, I'll share.
285
00:15:17,529 --> 00:15:18,519
Here's an example.
286
00:15:19,089 --> 00:15:23,859
One way people said to monitor for AI
doing the work of students is go back
287
00:15:23,859 --> 00:15:27,879
to the revision history of a Google Doc
and see if they copied and pasted in
288
00:15:27,879 --> 00:15:31,479
a giant four page essay all in one go.
289
00:15:31,479 --> 00:15:32,349
Like that was ai.
290
00:15:32,349 --> 00:15:33,249
That was cheating.
291
00:15:33,584 --> 00:15:35,654
Uh, that's not my preferred way.
292
00:15:35,654 --> 00:15:38,624
If that's what you do, okay,
here's the counter argument.
293
00:15:38,804 --> 00:15:43,754
And now a browser agent could be doing
that for them, and it's not gonna just
294
00:15:43,754 --> 00:15:47,024
copy and paste one chunk from the tab.
295
00:15:47,024 --> 00:15:48,344
They did the prompt in.
296
00:15:48,674 --> 00:15:52,484
It's actually gonna be in
Canvas or in Schoology or in
297
00:15:52,484 --> 00:15:54,584
Google classroom, navigating.
298
00:15:54,989 --> 00:15:57,329
It's gonna show the person's logged in.
299
00:15:57,329 --> 00:16:00,269
They made this edit at this
time, and then they opened here.
300
00:16:00,869 --> 00:16:04,439
And I've even had students show me,
you can prompt it to write out the
301
00:16:04,439 --> 00:16:10,139
thing, slowly put in a couple of
typos, then go back and edit them.
302
00:16:10,139 --> 00:16:13,109
And all of this is running without
the kid touching the keyboard.
303
00:16:13,109 --> 00:16:16,319
But the revision history
is gonna look organic.
304
00:16:16,319 --> 00:16:17,909
It's gonna look like a person did it.
305
00:16:17,909 --> 00:16:21,284
Lindy Hockenbary: That came up in
that session that I was at, at TCA
306
00:16:21,284 --> 00:16:24,929
is like, and even like Bri teaching,
they have an inspect writing tool.
307
00:16:25,289 --> 00:16:29,459
Built into that to show if any big
blocks were copy pasted or whatever.
308
00:16:29,669 --> 00:16:34,279
But now these agents, when I gave
operator a year ago access to my email,
309
00:16:34,279 --> 00:16:39,919
I sat there and watched it type out the
email and look exactly what it would
310
00:16:39,919 --> 00:16:41,959
look like if I was typing out the email.
311
00:16:42,109 --> 00:16:44,959
Maybe the only difference, which
would be a subtle difference
312
00:16:44,959 --> 00:16:49,609
that you'd have to really watch
is the typing speed and typos.
313
00:16:50,344 --> 00:16:54,244
The AI isn't gonna have typos
and be using the backspace like a
314
00:16:54,244 --> 00:16:57,154
human would when you're naturally
just typing something out.
315
00:16:57,424 --> 00:17:00,574
And the educators in that session,
some of them piped up and said, well,
316
00:17:00,694 --> 00:17:05,074
I was already having that problem with
students, figuring out that teachers
317
00:17:05,074 --> 00:17:08,434
are using these tools to see if I
was copying and pasting a paragraph.
318
00:17:08,674 --> 00:17:09,844
So what do they do?
319
00:17:09,924 --> 00:17:12,894
They write their essay in a
large language model, do a split
320
00:17:12,894 --> 00:17:14,994
screen, and then they type it.
321
00:17:16,344 --> 00:17:19,584
They're literally just like I think of
back in the day, you see all of like the
322
00:17:19,584 --> 00:17:23,484
movies where they made the kids write
lines and that was their punishment.
323
00:17:24,834 --> 00:17:25,494
So guess what?
324
00:17:25,554 --> 00:17:25,914
Now?
325
00:17:25,914 --> 00:17:28,854
Now the kid is literally,
that assignment a
326
00:17:28,907 --> 00:17:29,147
Aaron Makelky: It's,
327
00:17:29,274 --> 00:17:29,934
Lindy Hockenbary: punishment.
328
00:17:30,197 --> 00:17:31,967
Aaron Makelky: it's like
typing practice to the kid.
329
00:17:31,967 --> 00:17:35,627
They didn't come up with the, the
thinking or the ideating or the editing.
330
00:17:35,627 --> 00:17:37,757
It's just like, I'll copy it manually.
331
00:17:38,897 --> 00:17:39,227
Yeah.
332
00:17:39,227 --> 00:17:39,527
And there's.
333
00:17:39,952 --> 00:17:41,152
There's chrome extensions.
334
00:17:41,152 --> 00:17:46,282
I saw a student make one that you
paste in your essay, so probably chat.
335
00:17:46,282 --> 00:17:47,602
GPT generated it.
336
00:17:47,932 --> 00:17:52,192
The chrome extension is open in your
browser and it's gonna type it at a
337
00:17:52,192 --> 00:17:55,972
human speed and you can dictate that
and you just paste in your whole
338
00:17:55,972 --> 00:18:00,142
essay and it's gonna go into your
Google Doc and character by character.
339
00:18:00,142 --> 00:18:03,022
Make it look on the revision
history like a human did it.
340
00:18:03,022 --> 00:18:07,312
So I doubt many students use
that, but they're free and
341
00:18:07,312 --> 00:18:08,752
they're on the internet, so.
342
00:18:09,077 --> 00:18:12,197
The crafty ones are trying to find
ways, and you shouldn't assume that's
343
00:18:12,197 --> 00:18:13,367
the best way to filter for cheating.
344
00:18:15,214 --> 00:18:16,414
Lindy Hockenbary: If you're listening
to this and you're like, oh my
345
00:18:16,414 --> 00:18:18,184
gosh, I need to go try this myself.
346
00:18:18,184 --> 00:18:22,114
I need to know what's going on or what
tools I should use or what tools students
347
00:18:22,114 --> 00:18:23,524
are using to do this kind of stuff.
348
00:18:23,744 --> 00:18:26,054
I wanna give like a layered approach here.
349
00:18:26,294 --> 00:18:28,634
The first one , would
be the agents within.
350
00:18:30,029 --> 00:18:31,109
Chat, GBT.
351
00:18:31,409 --> 00:18:32,069
Claude.
352
00:18:32,069 --> 00:18:32,639
I'm trying to think.
353
00:18:32,639 --> 00:18:35,699
Gemini doesn't really have
agents yet, I don't think.
354
00:18:35,699 --> 00:18:35,759
No.
355
00:18:37,287 --> 00:18:37,557
Aaron Makelky: Nope.
356
00:18:37,617 --> 00:18:39,447
Just built into the browser now.
357
00:18:39,674 --> 00:18:40,124
Lindy Hockenbary: Yeah.
358
00:18:40,574 --> 00:18:43,364
Um, I'm gonna open in another tab,
but, um, yeah, built in the browser,
359
00:18:43,364 --> 00:18:44,474
which I'll get to in a second.
360
00:18:44,594 --> 00:18:44,954
Okay.
361
00:18:45,014 --> 00:18:47,834
So those have been there for a
while, like over a year, I would say
362
00:18:47,834 --> 00:18:49,304
maybe a little over a year for chat.
363
00:18:49,304 --> 00:18:49,904
GPT.
364
00:18:50,234 --> 00:18:51,404
So that's number one.
365
00:18:51,404 --> 00:18:54,884
If you go into chat, GPT, hit
the tools button, turn on agents,
366
00:18:55,094 --> 00:18:57,134
and people are always like,
well, I don't know what to do.
367
00:18:57,434 --> 00:18:59,894
If you don't know what to
do to start, guess what?
368
00:19:00,314 --> 00:19:02,474
It's an ai so you ask it.
369
00:19:03,254 --> 00:19:04,694
What can you do for me?
370
00:19:04,964 --> 00:19:05,174
How
371
00:19:05,277 --> 00:19:05,567
Aaron Makelky: Yeah.
372
00:19:05,804 --> 00:19:06,254
Lindy Hockenbary: me?
373
00:19:06,404 --> 00:19:06,794
Right?
374
00:19:06,944 --> 00:19:11,804
And if you have your in chat, GPT
especially, where if you have the setting
375
00:19:11,804 --> 00:19:16,274
turned on where it learns you and it
learns information from your different
376
00:19:16,274 --> 00:19:18,099
chats, if you use chat GPTA lot.
377
00:19:18,899 --> 00:19:22,649
It's gonna know you're a high
school social studies teacher.
378
00:19:22,679 --> 00:19:27,899
It knows everything about my work,
so I have to give it no context.
379
00:19:27,899 --> 00:19:30,359
Now when I go in there, so if
I go in there and I'm like,
380
00:19:30,359 --> 00:19:31,709
Hey, what can I do with you?
381
00:19:31,859 --> 00:19:35,519
It'll start giving me
very customized examples.
382
00:19:35,519 --> 00:19:38,369
And then from there you could
even say, give me a prompt,
383
00:19:38,789 --> 00:19:41,039
copy, paste, put the prompt in.
384
00:19:41,129 --> 00:19:43,499
And just, that's the
way I start exploring.
385
00:19:43,919 --> 00:19:45,299
These types of tools now.
386
00:19:45,329 --> 00:19:46,109
So there's that.
387
00:19:46,379 --> 00:19:51,299
Then the next level would be, we mentioned
it, the Chrome browser, which is by far
388
00:19:51,299 --> 00:19:54,509
the most common browser, and I would say
by far, the most common browser in K 12
389
00:19:54,509 --> 00:19:58,199
education now has Gemini built into it.
390
00:19:58,499 --> 00:20:02,699
So if you haven't noticed over the last
few months in the top right corner,
391
00:20:02,699 --> 00:20:06,839
you now have a Gemini button that
you can click if you don't see it.
392
00:20:07,599 --> 00:20:11,289
could be if you're logged into your
school account, it's either that your
393
00:20:11,289 --> 00:20:14,739
school has blocked it in some way, or
schools, depending upon their settings
394
00:20:14,739 --> 00:20:19,299
with Google Workspace, can do either a
rapid release or a scheduled release.
395
00:20:19,509 --> 00:20:22,839
So you might be on a scheduled release
where it hasn't been, you don't get
396
00:20:22,839 --> 00:20:26,974
the new features for a while until
after they're, they're out there.
397
00:20:27,594 --> 00:20:29,534
That's the one that,
honestly, so I mentioned.
398
00:20:30,219 --> 00:20:33,639
The concept of in that TC, a session
of agentic browsers came up, but
399
00:20:33,639 --> 00:20:38,049
then the presenter showed an example
using the Gemini within Chrome.
400
00:20:38,649 --> 00:20:42,279
This is not a full agentic browser yet.
401
00:20:42,279 --> 00:20:47,019
And Aaron, you said it perfectly,
it's actually not the most effective.
402
00:20:47,594 --> 00:20:52,964
There's, when we get into like the
comment and the atlas, those are way more
403
00:20:53,024 --> 00:20:55,184
effective than Gemini built within Chrome.
404
00:20:55,184 --> 00:20:58,874
Although I will say I've been using
the Gemini built within Chrome a lot.
405
00:20:59,174 --> 00:21:04,014
The example that the presenter at
TCA did, he had a, a question up
406
00:21:04,494 --> 00:21:06,804
or it was like a writing prompt or
something and he hit the button.
407
00:21:06,804 --> 00:21:08,334
He just acted like a high school kid.
408
00:21:08,334 --> 00:21:11,364
Like, Hey dude, uh, I'm working
on this problem and I can't
409
00:21:11,364 --> 00:21:13,524
figure out how, how to solve it.
410
00:21:13,524 --> 00:21:14,304
Can you solve it for me?
411
00:21:16,079 --> 00:21:20,659
Solved copy pasted right into the Google
Doc that's when the conversation got
412
00:21:20,659 --> 00:21:23,199
waylaid to oh, how do we block this
413
00:21:23,199 --> 00:21:24,969
Aaron Makelky: that's always
the first question when it's
414
00:21:24,969 --> 00:21:26,739
educators, unfortunately.
415
00:21:26,829 --> 00:21:32,769
Um, well, and we all know this, what
that presenter did, you could have done
416
00:21:32,769 --> 00:21:35,224
by switching tabs and copy and pasting.
417
00:21:35,589 --> 00:21:39,279
It's just the next evolution
of the same technology now.
418
00:21:39,869 --> 00:21:44,249
It's built in and you can give
it a tab group for context.
419
00:21:44,249 --> 00:21:45,449
That's what I usually do is like
420
00:21:45,485 --> 00:21:45,705
Lindy Hockenbary: Mm.
421
00:21:45,899 --> 00:21:49,979
Aaron Makelky: maybe a little research, a
little docket goes in and a another page
422
00:21:50,129 --> 00:21:56,249
and say, look across these three tabs,
do this task, write it out for me, and
423
00:21:56,309 --> 00:22:00,509
now it can use those instead of copying
everything off of those pages, which you
424
00:22:00,509 --> 00:22:04,769
could have done two or three years ago now
it just does it in the browser for you.
425
00:22:05,770 --> 00:22:08,980
Lindy Hockenbary: That's where I need
to switch my workflow in my brain.
426
00:22:09,250 --> 00:22:12,970
'cause for three years I, let's say
I have a Google Doc open and a couple
427
00:22:12,970 --> 00:22:17,860
websites, and I wanna give the large
language model the context of all of that.
428
00:22:17,860 --> 00:22:20,980
Actually, I just did this this morning
of whatever it is that I'm doing.
429
00:22:21,310 --> 00:22:25,420
I would download the Google Doc
and then I'd copy the links to the
430
00:22:25,420 --> 00:22:26,740
websites and put those into chat.
431
00:22:26,740 --> 00:22:27,100
GBT.
432
00:22:27,860 --> 00:22:31,160
It still saved me a ton of time,
but it's that download, upload,
433
00:22:31,160 --> 00:22:32,630
copy, paste for each of those.
434
00:22:32,780 --> 00:22:34,970
Now I just hit the Gemini button.
435
00:22:35,720 --> 00:22:42,650
already have the tabs open so I don't have
to give it any context to what information
436
00:22:42,650 --> 00:22:45,870
I want it to pull in and summarize, or
synthesize or whatever I want it to do.
437
00:22:46,855 --> 00:22:52,165
It's just right there, so it saves you
that many more seconds from a productivity
438
00:22:52,165 --> 00:22:57,775
standpoint, from a student standpoint,
it's that many more seconds saved and
439
00:22:57,775 --> 00:23:06,545
less steps to provide answers to , what we
used to call Googleable questions, right?
440
00:23:06,595 --> 00:23:09,565
Aaron Makelky: And the other thing
is remember that your students are
441
00:23:09,565 --> 00:23:11,275
gonna see all the ways to cheat.
442
00:23:11,365 --> 00:23:14,275
They're not gonna see on their
TikTok or their YouTube shorts,
443
00:23:14,335 --> 00:23:18,265
all the cool things that can help
you with researching and writing
444
00:23:18,325 --> 00:23:20,485
and keeping track of your schedule.
445
00:23:21,175 --> 00:23:24,655
. If all they see is the cheating way,
that's one of the reasons why I am a big
446
00:23:24,655 --> 00:23:29,365
fan of educators themselves have to share
these technologies with their students
447
00:23:29,365 --> 00:23:33,505
instead of hoping the student doesn't
know 'cause they're gonna find out or
448
00:23:33,505 --> 00:23:37,915
trying to block it, because a lot of us
don't have the option to block it either.
449
00:23:37,915 --> 00:23:42,265
Your administration decides what goes into
your Google apps for red account and they
450
00:23:42,265 --> 00:23:44,515
might listen to you, but probably not.
451
00:23:44,515 --> 00:23:46,795
Or a committee takes
three months to decide.
452
00:23:46,795 --> 00:23:50,155
So in the meantime you're stuck with
it and you just gotta make it work.
453
00:23:51,032 --> 00:23:51,812
Lindy Hockenbary: Figure it out.
454
00:23:51,902 --> 00:23:52,982
Yep, for sure.
455
00:23:53,372 --> 00:23:58,022
And then so the last level of agentic
browsers or, or well, or actually
456
00:23:58,292 --> 00:24:02,612
agentic browsers out there is, you
mentioned perplexity has comment, chat.
457
00:24:02,612 --> 00:24:04,352
GPT has atlas.
458
00:24:04,562 --> 00:24:09,267
How are those different than the
Gemini built within Google Chrome?
459
00:24:10,435 --> 00:24:15,385
Aaron Makelky: Yeah, so what Google
and Claude did was basically say
460
00:24:15,415 --> 00:24:20,425
instead of rebuild the browser, we'll
put an extension in Google Chrome.
461
00:24:20,965 --> 00:24:22,465
And let that control it.
462
00:24:23,245 --> 00:24:26,635
In a lot of ways, it was cheaper
than, you know, hiring the people
463
00:24:26,635 --> 00:24:28,015
to develop a whole new browser.
464
00:24:28,045 --> 00:24:33,625
What perplexity did is they forked
Chrome 'cause they basically
465
00:24:33,625 --> 00:24:35,665
copied the underlying architecture.
466
00:24:36,055 --> 00:24:38,695
The chrome extensions from your
chrome browsers still work in
467
00:24:38,695 --> 00:24:42,010
perplexity comment, but they
rebuilt it in a way that the agent.
468
00:24:42,420 --> 00:24:44,790
Is foundational, it's baked in.
469
00:24:44,790 --> 00:24:48,750
Instead of add a Chrome extension,
it comes with a whole sidebar and
470
00:24:48,750 --> 00:24:53,730
it has a couple different interfaces
and it's still using large language
471
00:24:53,730 --> 00:24:58,770
models, but it's way more prone
to taking action on your behalf.
472
00:24:59,010 --> 00:25:05,670
And I've had access to it since before
it came out, just in the last two weeks.
473
00:25:05,670 --> 00:25:08,040
So like while you were at
FETC in early February.
474
00:25:09,025 --> 00:25:14,905
With the newest models like Opus
4.6 Running Comet browser, I tested
475
00:25:14,905 --> 00:25:20,815
it and it did a task autonomously
for about 30 minutes, and it
476
00:25:20,815 --> 00:25:25,315
took between 175 and 190 actions.
477
00:25:25,585 --> 00:25:28,765
So think of that like scrolling, clicking.
478
00:25:28,980 --> 00:25:33,870
Typing a text field, basically how many
movements you would take on a browser.
479
00:25:33,870 --> 00:25:38,130
It's, it's definitely not faster than
if you were controlling your browser.
480
00:25:38,700 --> 00:25:43,320
I'd say it's half the speed, if not a
little slower, but here's the power of it.
481
00:25:43,410 --> 00:25:47,190
You set the task in a tab
group and say, go work on this,
482
00:25:47,280 --> 00:25:49,530
and it runs in those tabs.
483
00:25:49,530 --> 00:25:54,480
Meanwhile, you're doing something
else, so it's not like, oh, it's
484
00:25:54,480 --> 00:25:57,825
faster than air and clicking and
typing and searching and scrolling.
485
00:25:58,355 --> 00:26:00,635
It's Aaron's doing that on the page.
486
00:26:00,635 --> 00:26:04,865
He wants to, and then in the background
for 20 to 30 minutes, it's gonna run on
487
00:26:04,865 --> 00:26:07,775
its own and then say, okay, it's done.
488
00:26:07,925 --> 00:26:09,425
Or, okay, I hit a snag.
489
00:26:10,385 --> 00:26:11,855
Atlas is the same way.
490
00:26:11,885 --> 00:26:14,855
The difference is it's gonna
use your chat GPT account.
491
00:26:14,855 --> 00:26:18,035
So if you're somebody who's really
invested in that platform, you have
492
00:26:18,035 --> 00:26:20,225
the projects and the context history.
493
00:26:20,955 --> 00:26:22,245
Some people like it.
494
00:26:22,875 --> 00:26:28,035
I personally was not that invested in it,
so it didn't seem super useful for me.
495
00:26:28,035 --> 00:26:30,855
Like I wouldn't use that if
you're not a chat GPT Power user.
496
00:26:32,435 --> 00:26:37,145
If somebody wants another one, DIA
browser, DIA is another one that's
497
00:26:37,145 --> 00:26:40,115
been competing with Perplexity Comet.
498
00:26:40,145 --> 00:26:44,795
It actually came out before that
and it has some built in Agentic
499
00:26:44,795 --> 00:26:47,135
capabilities and you can switch models.
500
00:26:47,525 --> 00:26:51,575
If anyone used Arc browser, it's the
same company that built Arc, that's
501
00:26:51,575 --> 00:26:53,495
probably my favorite browser of all time.
502
00:26:53,825 --> 00:26:56,795
They just made an AI native
version of it called dia.
503
00:26:57,375 --> 00:27:01,575
There's a free version you can use,
but the thing is, and, and one of
504
00:27:01,575 --> 00:27:06,465
the barriers to a lot of student uses
the good models and the best versions
505
00:27:06,465 --> 00:27:08,445
of it, you need a subscription.
506
00:27:09,555 --> 00:27:17,715
So whereas the Gemini students can get a
year of the pro plan for free, so a lot
507
00:27:17,715 --> 00:27:19,725
of them have access to the paid features.
508
00:27:19,725 --> 00:27:22,575
Your district might enable 'em through
your Google apps for red account,
509
00:27:22,575 --> 00:27:24,795
or they might have a, a student.
510
00:27:25,175 --> 00:27:26,885
AI Pro Plan.
511
00:27:26,885 --> 00:27:30,335
I believe it's the $20 a month one that
they get for free if they're a student.
512
00:27:30,515 --> 00:27:33,965
So there, there's definitely more
students have exposure to the Chrome
513
00:27:33,965 --> 00:27:36,065
extension than the standalone browser.
514
00:27:37,432 --> 00:27:38,782
. Lindy Hockenbary: I'm writing
these all down by the way, and
515
00:27:38,782 --> 00:27:40,402
I'll put links into the show notes.
516
00:27:40,402 --> 00:27:42,472
I have not heard of Diaz,
so I gotta check that out.
517
00:27:42,532 --> 00:27:43,552
And I'm with you.
518
00:27:43,552 --> 00:27:47,152
I'm a pretty big power chat
BT user, but I also use a lot.
519
00:27:47,152 --> 00:27:48,952
I use, I'm a big Claude user.
520
00:27:48,952 --> 00:27:50,032
I'm using Gemini now.
521
00:27:50,062 --> 00:27:50,842
Gemini three.
522
00:27:50,842 --> 00:27:51,382
I'm using Gemini.
523
00:27:51,977 --> 00:27:53,807
Way more than I ever have before.
524
00:27:54,107 --> 00:27:59,767
And I still was kind of like me the
chat, GPT Atlas, , and I think that's why
525
00:27:59,767 --> 00:28:04,017
then I never got , back into the agentic
browsers because I was like, I don't,
526
00:28:04,287 --> 00:28:06,327
I don't know, I'm not seeing it yet.
527
00:28:06,327 --> 00:28:10,317
So I need to get into comment
some of these others and,
528
00:28:10,617 --> 00:28:12,237
and play around with them.
529
00:28:13,107 --> 00:28:15,057
Okay, so let's, we better move on.
530
00:28:15,057 --> 00:28:16,137
So many other things to talk about.
531
00:28:16,167 --> 00:28:18,147
'cause we haven't even
talked about Codex yet.
532
00:28:18,147 --> 00:28:19,347
So one of the big.
533
00:28:20,192 --> 00:28:23,972
releases that opening, I did that
first week of February was Codex Aaron,
534
00:28:24,002 --> 00:28:25,172
I've played with this a little bit.
535
00:28:25,172 --> 00:28:28,862
I know you've played with it a lot
more and your new job sounds like
536
00:28:28,862 --> 00:28:29,822
you've played with it a lot more.
537
00:28:29,822 --> 00:28:32,342
So tell us about Codex.
538
00:28:32,995 --> 00:28:36,325
Aaron Makelky: Full disclosure, I did
spend an entire weekend living in it.
539
00:28:36,385 --> 00:28:37,675
Um, not 'cause,
540
00:28:37,682 --> 00:28:37,982
Lindy Hockenbary: Okay.
541
00:28:38,155 --> 00:28:38,935
Aaron Makelky: 'cause Lindy,
542
00:28:39,122 --> 00:28:40,082
Lindy Hockenbary: more deep than I am.
543
00:28:40,120 --> 00:28:43,915
Aaron Makelky: I. Well, I think
we all have test projects, right?
544
00:28:43,915 --> 00:28:47,365
So say for you when a New Claw
model comes out, you're like,
545
00:28:47,365 --> 00:28:51,685
here's a go-to thing that I have a
baseline of what good looks like.
546
00:28:53,005 --> 00:28:57,685
Codex is interesting because I think a
lot of the vibe coding and coding apps
547
00:28:57,685 --> 00:28:59,785
have not been super relevant to teachers.
548
00:28:59,815 --> 00:29:01,435
'cause they're like, what
would I do with that?
549
00:29:01,435 --> 00:29:04,165
Like, I'm not a coder, I'm an English
teacher, I'm a history teacher.
550
00:29:05,365 --> 00:29:12,355
What Codex can do is it's basically chat
GPTs model, but specifically for coding,
551
00:29:12,625 --> 00:29:17,725
and you don't have to open a terminal, so
you don't have to go to the scary blinking
552
00:29:17,725 --> 00:29:21,565
black cursor box that feels like 1980s.
553
00:29:21,775 --> 00:29:24,295
That's intimidating to non coders like.
554
00:29:24,302 --> 00:29:27,272
Lindy Hockenbary: like a
terminal dos looking thing.
555
00:29:27,715 --> 00:29:30,955
Aaron Makelky: Yep, because cloud
code you have to run in a terminal.
556
00:29:31,045 --> 00:29:35,155
I know there's a web version now,
but originally, so basically chat,
557
00:29:35,155 --> 00:29:39,805
GPT tried to build the same thing,
but it's a standalone app on the Mac.
558
00:29:39,835 --> 00:29:44,785
It's now also on Windows, and
it runs in the web, so it has
559
00:29:44,785 --> 00:29:46,735
access to your local files.
560
00:29:46,735 --> 00:29:47,725
This is one thing that's different.
561
00:29:47,725 --> 00:29:50,185
If you have chat GPT in your browser tab.
562
00:29:50,665 --> 00:29:53,845
It can't see your downloads,
it can't see your local files.
563
00:29:55,560 --> 00:30:00,780
Codex can, and people think, well, I'm
not gonna build an app or a website.
564
00:30:02,010 --> 00:30:04,890
Even if you can't, you know,
that's not your use case.
565
00:30:04,890 --> 00:30:06,390
You could take a folder.
566
00:30:06,930 --> 00:30:08,370
Uh, here's one use case.
567
00:30:08,670 --> 00:30:10,530
I do some analytics reporting.
568
00:30:10,590 --> 00:30:14,185
So download the CSV files or
the spreadsheets, put 'em in a
569
00:30:14,185 --> 00:30:18,720
local folder, then open Codex
and say, go in that folder.
570
00:30:19,030 --> 00:30:25,030
And help me analyze all of this data
versus before you could probably attach
571
00:30:25,060 --> 00:30:30,370
files and do all sorts of importing
and exporting in the web now can do
572
00:30:30,370 --> 00:30:33,220
that in a folder on your computer.
573
00:30:33,640 --> 00:30:37,120
Um, and then like one of the ways
I used to test it, I think a lot
574
00:30:37,120 --> 00:30:41,890
of us have vibe coded in a webpage
or you make one for yourself.
575
00:30:41,920 --> 00:30:46,420
My, I make my kids do that
for themselves and I took a.
576
00:30:46,562 --> 00:30:47,762
Lindy Hockenbary: kids vibe coding, huh?
577
00:30:47,762 --> 00:30:48,362
I love it.
578
00:30:49,090 --> 00:30:51,400
Aaron Makelky: Maya, I
think he was 10 at the time.
579
00:30:51,400 --> 00:30:52,060
10 or 11.
580
00:30:52,060 --> 00:30:57,190
My fifth grade son built his own math
review game without touching a key.
581
00:30:57,190 --> 00:31:04,210
He just voice dictated into an app
and it worked, um, which I think
582
00:31:04,210 --> 00:31:05,530
everyone should be doing, but.
583
00:31:06,400 --> 00:31:07,870
Digressing back to Codex.
584
00:31:08,080 --> 00:31:11,260
I took a website that took me weeks
and a bunch of revisions, and I was
585
00:31:11,260 --> 00:31:14,890
like, Hey, let's redo this, make it
look different, change the content
586
00:31:15,460 --> 00:31:17,410
and move it to a different domain.
587
00:31:18,130 --> 00:31:22,570
And I did something that the previous
best way I knew how to do took me,
588
00:31:22,930 --> 00:31:27,820
I don't know, 10 or 12 hours, and
this was about a 30 minute process.
589
00:31:28,990 --> 00:31:33,310
And it totally revised the
look, the design, the colors.
590
00:31:33,670 --> 00:31:36,910
And then I thought, okay, this was
really good at the designing part.
591
00:31:36,910 --> 00:31:41,200
What's a, what's a design thing that
a non-technical person would use?
592
00:31:41,200 --> 00:31:42,850
So I actually had it design.
593
00:31:43,375 --> 00:31:47,005
My LinkedIn banner and some graphics.
594
00:31:47,095 --> 00:31:49,855
'cause people are like, oh, it's a
coding model, but I'm not a coder.
595
00:31:50,605 --> 00:31:53,815
Well, part of, I mean,
design is fundamentally code.
596
00:31:53,815 --> 00:31:56,035
They're all just pixels
and ones and zeros.
597
00:31:56,455 --> 00:31:59,655
It compared to what I
could do in Canva or Figma.
598
00:31:59,665 --> 00:32:02,395
It made a way better
LinkedIn banner in about.
599
00:32:02,455 --> 00:32:07,975
A a minute, and here's the best part
to my local computer, I'd say put it
600
00:32:07,975 --> 00:32:10,255
in this folder and call it designs.
601
00:32:10,825 --> 00:32:15,115
That's not something that you could
do with chat GPT or clot on the web.
602
00:32:15,385 --> 00:32:18,175
'cause you always have to
export the artifact and download
603
00:32:18,175 --> 00:32:20,125
the PDF or whatever it's in.
604
00:32:20,665 --> 00:32:22,825
I could just point it to my local.
605
00:32:23,357 --> 00:32:26,537
Lindy Hockenbary: tell it to save in
like a certain fuller on your computer.
606
00:32:26,815 --> 00:32:27,175
Aaron Makelky: Yep.
607
00:32:27,377 --> 00:32:28,127
Lindy Hockenbary: Wow.
608
00:32:28,907 --> 00:32:29,597
Interesting.
609
00:32:29,830 --> 00:32:31,570
Aaron Makelky: And then you
can, you can back it up.
610
00:32:31,570 --> 00:32:34,660
So a lot of people use GitHub
and if you're intimidated,
611
00:32:34,750 --> 00:32:38,110
I was once so intimidated by
GitHub and I'm not an expert.
612
00:32:38,560 --> 00:32:42,400
Think of GitHub, like
Google Drive for tech.
613
00:32:42,400 --> 00:32:45,400
People like, and you
can put documents in it.
614
00:32:45,400 --> 00:32:47,140
You don't have to build apps and websites.
615
00:32:47,470 --> 00:32:50,530
But it's just a way to back
things up and then also share it.
616
00:32:50,530 --> 00:32:53,260
Just like a lot of educators
go, oh, that form, that
617
00:32:53,260 --> 00:32:54,700
spreadsheet is in my Google Drive.
618
00:32:54,700 --> 00:32:55,450
Here's a link.
619
00:32:55,870 --> 00:32:57,580
That's essentially what GitHub does.
620
00:32:58,030 --> 00:33:02,050
You can say, take the project we put on
my local machine, like those LinkedIn
621
00:33:02,050 --> 00:33:07,120
banners and this whole website, and sync
it to my GitHub, and it's just like, boom.
622
00:33:07,120 --> 00:33:08,740
Running this command done.
623
00:33:08,770 --> 00:33:10,840
It's committed like it's backed up there.
624
00:33:11,830 --> 00:33:15,190
So I do think there are a lot of
uses for teachers, even if you're
625
00:33:15,190 --> 00:33:19,150
not computer science and building
apps, that that is what it's best at.
626
00:33:19,150 --> 00:33:23,260
But it can do all sorts of other things
for you as well, and it can do it on
627
00:33:23,260 --> 00:33:26,590
your local machine from a web app.
628
00:33:26,590 --> 00:33:28,030
You don't need to be in a terminal.
629
00:33:28,030 --> 00:33:30,910
Lindy Hockenbary: okay, so I'm hearing
the big changes and differences with
630
00:33:30,910 --> 00:33:37,370
Codex is that download from an app, so
it's more user friendly than Claude Code,
631
00:33:37,370 --> 00:33:43,790
which was kind of the leader in AI coding
basically up until February 1st ish.
632
00:33:44,090 --> 00:33:45,105
And then it can also.
633
00:33:46,115 --> 00:33:49,565
Actually save files in
specific folders for you.
634
00:33:49,565 --> 00:33:53,075
And then the thing I noticed in the
little bit of time I've spent in
635
00:33:53,075 --> 00:33:57,395
there, I went in and just so you
guys know, like I am not a coder.
636
00:33:57,455 --> 00:33:59,225
I am not that kind of technical person.
637
00:33:59,405 --> 00:34:03,095
So I went in and I could have asked
it, Hey, what can you build for me?
638
00:34:03,275 --> 00:34:07,415
But when you go in there, it actually
has some sample prompts for you.
639
00:34:07,625 --> 00:34:08,885
So I just clicked the first one.
640
00:34:08,975 --> 00:34:11,465
It was something like build
a sneak game, something super
641
00:34:11,465 --> 00:34:13,625
easy and just watched it run.
642
00:34:13,940 --> 00:34:18,800
The thing that I was like, oh,
this is a game changer, is it
643
00:34:18,800 --> 00:34:23,420
not only built the game, it built
the app, but then it tested.
644
00:34:24,080 --> 00:34:29,120
The app and if you have never worked,
like Aaron's, now in the tech world, I
645
00:34:29,120 --> 00:34:32,570
work with a lot of ed tech companies,
so I get immersed in that world.
646
00:34:32,960 --> 00:34:36,710
There is a job at every, at least
one person, at every company
647
00:34:36,710 --> 00:34:38,450
that builds software called a qa.
648
00:34:38,870 --> 00:34:40,160
They're like the QA engineer.
649
00:34:40,160 --> 00:34:42,770
And QA stands for quality assurance.
650
00:34:43,100 --> 00:34:48,560
And what that person does is test
and make sure their software works.
651
00:34:48,560 --> 00:34:50,330
Work looks for bugs, right?
652
00:34:51,200 --> 00:34:52,850
Codex is doing that.
653
00:34:52,850 --> 00:34:55,990
And I'm not being the hype gal
that's saying that Codex is gonna
654
00:34:55,990 --> 00:34:57,760
take every QA engineer's job.
655
00:34:57,850 --> 00:35:03,610
I'm not saying that, but I want you
to think about the job disruption that
656
00:35:03,610 --> 00:35:08,080
could be caused by the AI and oh, and
here's another thing about Codex, which
657
00:35:08,320 --> 00:35:15,580
I have not fact checked, but I heard
that OpenAI actually had AI create codex.
658
00:35:15,790 --> 00:35:16,600
Have you heard that?
659
00:35:16,600 --> 00:35:17,500
Do you know if that's true?
660
00:35:17,844 --> 00:35:18,474
Aaron Makelky: Yes.
661
00:35:18,534 --> 00:35:22,104
, They publicly said they used
their newest model to build it.
662
00:35:22,104 --> 00:35:26,724
So basically the tool they were building
helped them build the tool, which is
663
00:35:26,724 --> 00:35:28,464
one of those weird advantages in the ai.
664
00:35:28,644 --> 00:35:28,944
World.
665
00:35:29,694 --> 00:35:33,864
Yeah, and if you've played with other
ones, here's what I'd say about Codex.
666
00:35:33,924 --> 00:35:35,394
It's very persistent.
667
00:35:35,394 --> 00:35:37,524
So I've used Claude to do a task.
668
00:35:38,104 --> 00:35:44,314
Claude can feel , it's more warm, it's
more enthusiastic, it's more human, but
669
00:35:44,314 --> 00:35:48,964
it also can feel kind of a DHD and kind of
jump around and quickly do this, and it's
670
00:35:48,964 --> 00:35:54,184
not as good as, look, it's not as good
at looking through things in, in detail.
671
00:35:54,499 --> 00:35:59,149
I did a project with Codex just last
night and multiple times the little
672
00:35:59,149 --> 00:36:04,339
working bar ran for more than 20 minutes
before it actually output something.
673
00:36:05,029 --> 00:36:08,629
And I've never seen a clawed
model be that persistent.
674
00:36:08,659 --> 00:36:12,229
'cause it'll either take a shortcut,
hallucinate, make something
675
00:36:12,259 --> 00:36:14,359
up, give you like A-T-L-D-R.
676
00:36:15,049 --> 00:36:18,619
Codex definitely seems more thorough and.
677
00:36:19,324 --> 00:36:24,994
Kind of gruff, like, okay, here's the
thing, it's done, and it's not as warm
678
00:36:24,994 --> 00:36:26,914
of a personality as Claude models.
679
00:36:28,150 --> 00:36:28,930
Lindy Hockenbary: Which makes sense.
680
00:36:28,930 --> 00:36:32,500
Just knowing the difference between
Anthropic and OpenAI, the companies that
681
00:36:32,500 --> 00:36:37,420
make both of those, and just knowing the
vibe difference between Claude and Chat
682
00:36:37,450 --> 00:36:41,400
GBT, that kind of makes sense , that their
coding models , would do the same thing.
683
00:36:41,640 --> 00:36:43,170
Okay, so, so that's a biggie.
684
00:36:43,170 --> 00:36:44,100
There's a lot to unpack.
685
00:36:44,175 --> 00:36:44,655
There.
686
00:36:45,045 --> 00:36:47,835
And I know at the surface you might
meant, but like really stop and
687
00:36:47,835 --> 00:36:53,655
think about the fact that we now
have AI, building AI that is huge.
688
00:36:53,655 --> 00:36:58,005
And we have already been moving
so quickly with AI advancements.
689
00:36:58,275 --> 00:37:02,055
Now that AI is building ai, you
know how much quicker we're gonna
690
00:37:02,055 --> 00:37:04,695
be moving with AI advancements.
691
00:37:06,315 --> 00:37:12,300
AI thinks, and works way, faster, way,
way, way, and way is not even the right.
692
00:37:13,050 --> 00:37:13,860
Word there.
693
00:37:15,270 --> 00:37:18,840
A lot faster, lot faster than humans.
694
00:37:19,524 --> 00:37:23,064
Aaron Makelky: And it doesn't sleep,
it doesn't take vacations or PTO days.
695
00:37:23,330 --> 00:37:25,010
Lindy Hockenbary: It doesn't
have bio break needs.
696
00:37:25,814 --> 00:37:29,864
Aaron Makelky: can spawn subagents
to make work, you know, delegate
697
00:37:29,894 --> 00:37:33,754
jobs on, like you and I can't spawn,
, executive assistants to go through
698
00:37:33,754 --> 00:37:35,764
our emails or do things for us.
699
00:37:35,764 --> 00:37:38,644
Lindy Hockenbary: One other big update and
then if we have time, we'll dig a little
700
00:37:38,644 --> 00:37:40,624
bit more into the effects of education.
701
00:37:40,624 --> 00:37:45,574
We might have to do a part two of this
episode just focused on education.
702
00:37:46,184 --> 00:37:51,314
Should needs to change, not
change, adapt, not adapt.
703
00:37:51,314 --> 00:37:56,444
But I wanna make sure we talk about the
Claude bot slash open claw slash mt bot.
704
00:37:56,564 --> 00:37:57,164
Fun.
705
00:37:57,534 --> 00:37:59,544
, So again, I was watching this
go down while I was at tca.
706
00:37:59,904 --> 00:38:04,374
And by the way, and granted I only had
a chance to go to a few sessions, I had
707
00:38:04,374 --> 00:38:09,144
tons of conversations with educators and I
presented five or six different sessions.
708
00:38:09,414 --> 00:38:12,654
And not once did I have
somebody mention this to me.
709
00:38:13,409 --> 00:38:16,769
At the education conference, which
I thought was super interesting.
710
00:38:17,199 --> 00:38:21,489
, And just like maybe a difference
in people's algorithms.
711
00:38:21,549 --> 00:38:24,549
'cause it was all over all of
my social media algorithms.
712
00:38:24,549 --> 00:38:25,449
That's all I could see.
713
00:38:25,839 --> 00:38:27,909
So A guy created what?
714
00:38:27,909 --> 00:38:29,439
And he, he called it Claude Bot.
715
00:38:29,734 --> 00:38:30,574
To start with.
716
00:38:30,784 --> 00:38:36,874
Well then, and it was C-L-A-W-D-B-O-T,
then people thought it, it was related
717
00:38:36,874 --> 00:38:42,604
to Claude, C-L-A-U-D-E, which is the
large language model made by Anthropic.
718
00:38:42,604 --> 00:38:44,394
It was not, not associated at all.
719
00:38:44,544 --> 00:38:49,524
So then he quickly changed the
name to Open Claw and basically,
720
00:38:49,584 --> 00:38:50,724
well, here you explain.
721
00:38:50,784 --> 00:38:56,454
Aaron, if, if you're comfortable, explain
what Open Claw slash Claude bought is.
722
00:38:57,727 --> 00:39:00,577
Aaron Makelky: Well, and if teachers
hadn't heard by then, the backstory
723
00:39:00,577 --> 00:39:04,417
is it's changed names three
times, but you're exactly right.
724
00:39:04,424 --> 00:39:05,564
Lindy Hockenbary: was
there a third one in there?
725
00:39:05,564 --> 00:39:05,864
What was
726
00:39:05,932 --> 00:39:07,072
Aaron Makelky: Yeah.
727
00:39:07,072 --> 00:39:09,292
Well, and that's what's funny 'cause
I know you wanna bring this up.
728
00:39:09,592 --> 00:39:16,102
Mt Bought was a very quick change and what
the founder of this open source project
729
00:39:16,102 --> 00:39:17,962
said was Anthropic didn't like that.
730
00:39:17,962 --> 00:39:21,862
It sounded like Claude, even though it
was spelled differently, I'm guessing
731
00:39:21,862 --> 00:39:23,962
somebody sent a letter from a lawyer
732
00:39:24,404 --> 00:39:24,944
Lindy Hockenbary: Yeah.
733
00:39:24,952 --> 00:39:27,682
Aaron Makelky: is like, okay, I'll
change it to Molt Bot 'cause it,
734
00:39:27,742 --> 00:39:30,022
it molts and changes and evolves.
735
00:39:30,352 --> 00:39:32,962
And that's where Molt book came from.
736
00:39:33,344 --> 00:39:33,564
Lindy Hockenbary: Oh,
737
00:39:34,342 --> 00:39:37,852
Aaron Makelky: Is semi satirical.
738
00:39:37,852 --> 00:39:42,202
Someone said, let's make a
website for all these new agents.
739
00:39:42,412 --> 00:39:46,072
That's like social media, you
know, LinkedIn, essentially for
740
00:39:46,072 --> 00:39:48,352
the agents to post, not the humans.
741
00:39:48,892 --> 00:39:52,702
Then the founder of the project
changed it, a third different name
742
00:39:52,762 --> 00:39:56,122
to Open Claw, which it got acquired.
743
00:39:56,122 --> 00:39:58,492
He got acquired by Open AI this weekend.
744
00:39:58,492 --> 00:39:59,242
So now.
745
00:40:00,412 --> 00:40:01,072
Yeah.
746
00:40:01,072 --> 00:40:06,892
He, he and the founder, uh, Sam
Altman said it'll stay open source.
747
00:40:06,892 --> 00:40:11,392
It's not gonna go away, but he's joining
our team, whatever that's gonna look like.
748
00:40:11,632 --> 00:40:17,302
But you know, this, just like Atlas and
Codex, it's probably gonna be a standalone
749
00:40:17,302 --> 00:40:20,452
product in chat GPT any day now.
750
00:40:21,539 --> 00:40:22,019
Lindy Hockenbary: Yeah.
751
00:40:22,469 --> 00:40:22,859
Yeah.
752
00:40:22,889 --> 00:40:23,159
Okay.
753
00:40:23,159 --> 00:40:27,449
So I missed a whole, I was thinking
it was Molt bought not Molt book.
754
00:40:27,449 --> 00:40:29,099
I missed a whole name in there.
755
00:40:29,099 --> 00:40:29,459
All right.
756
00:40:29,639 --> 00:40:32,579
But Moral of the story is
it's landed on open claw
757
00:40:32,982 --> 00:40:33,202
Aaron Makelky: Yep.
758
00:40:33,719 --> 00:40:37,259
Lindy Hockenbary: basically only
really technical people are doing
759
00:40:37,259 --> 00:40:41,849
it because you have to kind of
create and run your own terminal.
760
00:40:41,959 --> 00:40:43,939
The Best Buy in San Francisco.
761
00:40:44,599 --> 00:40:48,769
Ran out of Mac Minis because people
were going out and buying Mac
762
00:40:48,769 --> 00:40:52,939
Minis to be, 'cause you really need
like a dedicated computer because
763
00:40:52,939 --> 00:40:59,959
it has to have 24 7 uptime and
basically it's running agent for you.
764
00:40:59,959 --> 00:41:03,229
Or is it agents, is it multiple
agents or just one agent
765
00:41:03,229 --> 00:41:05,209
that's like just working 24 7.
766
00:41:05,802 --> 00:41:06,542
Aaron Makelky: Either one.
767
00:41:06,692 --> 00:41:07,527
It's up to you.
768
00:41:07,579 --> 00:41:07,879
Lindy Hockenbary: one.
769
00:41:07,879 --> 00:41:08,149
Okay.
770
00:41:08,149 --> 00:41:11,389
So you could have multiple agents
potentially doing different things,
771
00:41:11,569 --> 00:41:14,239
but you have to give it access,
like we were talking about, the
772
00:41:14,239 --> 00:41:19,459
security, you know, data privacy risk
there, you have to give it access to
773
00:41:19,519 --> 00:41:22,339
everything, but then it can literally.
774
00:41:23,734 --> 00:41:25,864
pretty much anything digital for you.
775
00:41:25,864 --> 00:41:31,844
Like save this file, send this email,
, put this event on my calendar for me.
776
00:41:31,994 --> 00:41:38,624
And you have to use, like a lot of people
were using WhatsApp send the instructions
777
00:41:39,074 --> 00:41:43,634
to the agents to say, you know, go put
this on my calendar, delete this event
778
00:41:43,634 --> 00:41:45,854
off my calendar, et cetera, et cetera.
779
00:41:45,854 --> 00:41:46,244
Aaron Makelky: Yeah.
780
00:41:46,604 --> 00:41:48,674
Here's how I view Open Claw.
781
00:41:48,734 --> 00:41:51,494
We'll stick with that name and hope
he doesn't change it a fourth time.
782
00:41:52,274 --> 00:41:53,354
It doesn't do anything new.
783
00:41:54,869 --> 00:41:58,409
You could have done all of this stuff
with Claude Code, or if you were a
784
00:41:58,409 --> 00:42:01,979
techie person, you could have duct
taped your own version of this together.
785
00:42:02,429 --> 00:42:06,569
But there was no hope of an
average person making that work.
786
00:42:07,169 --> 00:42:11,279
And now what's unique about
Open Claw is just the interface.
787
00:42:11,309 --> 00:42:11,999
In other words.
788
00:42:12,449 --> 00:42:15,179
If you like WhatsApp, you
have group chats there.
789
00:42:15,179 --> 00:42:16,139
You're in the app.
790
00:42:16,499 --> 00:42:20,309
You can use it to talk to your
agent, but then it can go out
791
00:42:20,309 --> 00:42:21,659
and use all these other tools.
792
00:42:21,689 --> 00:42:24,899
It can search the internet, it
can search specific websites.
793
00:42:24,899 --> 00:42:28,349
It can see files or folders you have
saved in the cloud somewhere, like your
794
00:42:28,349 --> 00:42:34,619
Google Drive, your notion, your GitHub,
whatever, and it can do things with that.
795
00:42:34,979 --> 00:42:36,359
It's also misleading.
796
00:42:36,839 --> 00:42:39,209
I tell people all the
time, don't buy a Mac Mini.
797
00:42:39,379 --> 00:42:40,939
, I have one under my desk.
798
00:42:40,999 --> 00:42:42,739
Mine is on a raspberry pie.
799
00:42:42,769 --> 00:42:45,439
Number one, a raspberry pie is
way cheaper than a Mac mini.
800
00:42:46,039 --> 00:42:49,939
Also easier to find in stock right now,
but your local Facebook marketplace
801
00:42:49,939 --> 00:42:53,959
is probably gonna have a bunch of used
very gently used Mac Minis any day.
802
00:42:53,959 --> 00:42:59,539
Now, as people realize the, the
tool is free, but the services you
803
00:42:59,539 --> 00:43:01,609
have to plug into it to run are not.
804
00:43:01,609 --> 00:43:03,379
So I think of it like a house.
805
00:43:04,159 --> 00:43:07,279
He basically built the
app and said, here you go.
806
00:43:07,279 --> 00:43:07,639
It's free.
807
00:43:07,639 --> 00:43:08,449
You can download.
808
00:43:08,449 --> 00:43:12,349
But to make it work, you have to plug
it into things that cost money to run.
809
00:43:12,439 --> 00:43:14,239
The biggest one is an AI tool.
810
00:43:14,239 --> 00:43:18,649
Like it needs a A brain, it
needs chat, GBT, Claude Gemini.
811
00:43:18,649 --> 00:43:19,459
You can use anyone.
812
00:43:19,459 --> 00:43:23,359
You can get an API key
for those cost money.
813
00:43:23,599 --> 00:43:27,979
So his app without those connections
is like having an empty house that's
814
00:43:27,979 --> 00:43:29,629
not connected to the power grid.
815
00:43:30,274 --> 00:43:31,594
Connected to the sewer.
816
00:43:32,074 --> 00:43:33,784
It doesn't have electricity.
817
00:43:33,784 --> 00:43:34,834
It doesn't have plumbing.
818
00:43:34,834 --> 00:43:36,694
Well, that's a pretty worthless house.
819
00:43:36,694 --> 00:43:39,604
It looks cool, especially that
it's free, but it's really about
820
00:43:39,604 --> 00:43:41,434
how you connect it to things.
821
00:43:41,944 --> 00:43:47,464
And what I would do and what I am
doing is put it on a different device.
822
00:43:47,524 --> 00:43:49,234
It just doesn't have to be powerful.
823
00:43:49,234 --> 00:43:51,784
You don't need a fancy pc.
824
00:43:52,015 --> 00:43:53,635
Lindy Hockenbary: Mac device.
825
00:43:54,364 --> 00:43:54,934
Aaron Makelky: No.
826
00:43:54,934 --> 00:43:58,594
And if you do, 'cause it's an old
one, just know for anything to
827
00:43:58,594 --> 00:44:01,954
work, it has to be on, has to be
powered on and on the internet.
828
00:44:02,524 --> 00:44:06,304
So if it's a laptop, you're gonna have
to change the battery settings and the
829
00:44:06,304 --> 00:44:08,044
sleep settings to never go to sleep.
830
00:44:08,734 --> 00:44:11,794
, But something like a
raspberry pie can run it.
831
00:44:12,604 --> 00:44:17,344
I think somewhere between one to two
gigs of memory is what people find.
832
00:44:17,584 --> 00:44:18,574
It's good enough.
833
00:44:18,574 --> 00:44:21,964
So like any old laptop from the
last 10 years could run this.
834
00:44:21,964 --> 00:44:24,784
Any old PC you don't
still use could run this.
835
00:44:25,264 --> 00:44:28,264
But the other thing is you want it on
a separate device so it doesn't have
836
00:44:28,264 --> 00:44:32,974
access to all of your permissions, your
browser, your folders, like as a teacher,
837
00:44:32,974 --> 00:44:34,714
definitely not on your work computer.
838
00:44:34,840 --> 00:44:35,950
Lindy Hockenbary: No,
839
00:44:36,454 --> 00:44:37,084
Aaron Makelky: Um.
840
00:44:37,330 --> 00:44:39,880
Lindy Hockenbary: of tech directors
that would be up my throat.
841
00:44:40,300 --> 00:44:40,630
Yeah.
842
00:44:40,630 --> 00:44:42,310
Do you or your work.
843
00:44:43,315 --> 00:44:45,575
Emails, accounts at
844
00:44:45,679 --> 00:44:46,279
Aaron Makelky: Yes.
845
00:44:47,329 --> 00:44:52,009
The, the way to think of it is you wanna
build it in a box and keep it contained.
846
00:44:52,009 --> 00:44:54,619
Like you don't want to give it that email.
847
00:44:54,619 --> 00:44:58,429
So what's interesting, some people
give their open call agent their
848
00:44:58,459 --> 00:45:02,479
own accounts to things like their
own Gmail account, their own iCloud.
849
00:45:03,109 --> 00:45:04,039
, One guy now has.
850
00:45:04,324 --> 00:45:07,624
Two Dropbox subscriptions, one
for him and one for his agents.
851
00:45:07,624 --> 00:45:10,114
And some things are shared between those.
852
00:45:10,594 --> 00:45:14,944
But his important, like PII
containing files or finances
853
00:45:15,214 --> 00:45:19,024
are in his personal Dropbox that
his open claw agents can't see.
854
00:45:19,564 --> 00:45:20,044
So.
855
00:45:20,384 --> 00:45:25,004
Really, the moral of the story is you
wanna start with it on a different device
856
00:45:25,004 --> 00:45:26,714
and don't give it access to anything.
857
00:45:26,834 --> 00:45:30,164
You wouldn't trust a human,
like a say a student teacher.
858
00:45:30,344 --> 00:45:32,894
You would not give a student
teacher your bank account password
859
00:45:32,894 --> 00:45:35,024
and your personal laptop 24 7.
860
00:45:35,024 --> 00:45:37,034
So you definitely don't
wanna give it to an agent.
861
00:45:37,805 --> 00:45:38,615
Lindy Hockenbary: Really good advice.
862
00:45:38,735 --> 00:45:42,905
And that's the big, the things that I
have heard that's like the big differences
863
00:45:42,905 --> 00:45:49,370
with open claw is that now it's allowed
agents to go externally is one part of it.
864
00:45:49,535 --> 00:45:51,775
Like you said, like you could do a
lot of the things that you could do.
865
00:45:52,610 --> 00:45:56,360
With open Claw, with like the
chat GPT agents over the last
866
00:45:56,360 --> 00:45:57,950
year, but now it can go external.
867
00:45:57,950 --> 00:46:02,090
And then that leads to the second
thing, which is now the agents can
868
00:46:02,540 --> 00:46:05,270
talk to each other, talk in quotes.
869
00:46:05,270 --> 00:46:05,660
Right.
870
00:46:05,690 --> 00:46:09,590
And so that's where the malt book
I had been calling it, malt Bot,
871
00:46:09,590 --> 00:46:14,450
the Malt book came into play,
which is basically like a Reddit.
872
00:46:14,870 --> 00:46:19,550
For AI agents and they've gone crazy
and they're chatting and they've
873
00:46:19,550 --> 00:46:22,730
like made up a religion and they're
talking about religions and they're,
874
00:46:22,910 --> 00:46:24,710
and it's freaking people out.
875
00:46:25,970 --> 00:46:28,760
, This is a big, we've gotta wrap
up here and I know it was a big,
876
00:46:28,760 --> 00:46:30,980
heavy thing to leave the episode on.
877
00:46:31,390 --> 00:46:34,990
, But I think it just goes back to that
foundational AI literacy that when you
878
00:46:34,990 --> 00:46:41,515
understand that the AI is just acting
with instructions, you understand that.
879
00:46:42,550 --> 00:46:46,810
The agents were given instructions
to do that, and they're basically
880
00:46:46,810 --> 00:46:52,540
replicating Reddit thread because
that's what they have to go on.
881
00:46:52,960 --> 00:46:56,860
So they're just kind of like
recreating that in mt book in
882
00:46:56,860 --> 00:46:58,750
their own little agent land.
883
00:46:59,844 --> 00:47:00,234
Aaron Makelky: Yep.
884
00:47:00,294 --> 00:47:04,044
And, and I would also remind people
that there's no way to verify all
885
00:47:04,044 --> 00:47:08,154
the things on there are actually
coming from an agent at that.
886
00:47:08,154 --> 00:47:11,844
Any person could say, go on
and write all these things.
887
00:47:11,844 --> 00:47:12,664
Go make up your own.
888
00:47:12,834 --> 00:47:16,824
Religion and the agent is gonna
do what it's been prompted to do.
889
00:47:17,274 --> 00:47:21,624
So I don't know how much of that
is authentically happening, like an
890
00:47:21,624 --> 00:47:24,804
agent spontaneously went and posted
versus somebody thought it'd be
891
00:47:24,804 --> 00:47:27,414
funny to create these conversations.
892
00:47:27,760 --> 00:47:28,120
Lindy Hockenbary: Yep.
893
00:47:28,674 --> 00:47:29,154
Aaron Makelky: but
894
00:47:30,190 --> 00:47:33,160
Lindy Hockenbary: their agent to go like
say, Hey, let's go start another religion
895
00:47:33,714 --> 00:47:34,164
Aaron Makelky: Yep.
896
00:47:35,334 --> 00:47:35,604
Yep.
897
00:47:35,604 --> 00:47:35,669
And that.
898
00:47:35,740 --> 00:47:36,310
Lindy Hockenbary: a good point.
899
00:47:36,310 --> 00:47:36,340
I.
900
00:47:37,009 --> 00:47:40,759
Aaron Makelky: A, a variation
of that same mt book style site
901
00:47:40,759 --> 00:47:42,229
that's been freaking people out.
902
00:47:42,739 --> 00:47:48,409
We've all seen the Upwork, the
TaskRabbit, the hire a person to, uh,
903
00:47:48,439 --> 00:47:51,739
do this little project in Photoshop
or make this webpage for you.
904
00:47:52,069 --> 00:47:55,879
There have now been some sites
pop up where agents need humans
905
00:47:55,879 --> 00:47:59,329
to do things in the physical
world and they will pay people
906
00:48:00,070 --> 00:48:00,360
Lindy Hockenbary: What
907
00:48:00,379 --> 00:48:01,579
Aaron Makelky: do the tasks.
908
00:48:01,789 --> 00:48:05,329
So it's like an Upwork
for agents to find humans.
909
00:48:05,939 --> 00:48:10,799
Um, and that's, uh, just a whole
new world of like, okay, they, they
910
00:48:10,799 --> 00:48:15,419
can't go in the world and turn a knob
or go to a physical location yet.
911
00:48:15,475 --> 00:48:16,195
Lindy Hockenbary: embodied
912
00:48:16,679 --> 00:48:17,789
Aaron Makelky: Yep, yep.
913
00:48:17,995 --> 00:48:19,255
Lindy Hockenbary: till
they're embodied ai.
914
00:48:19,375 --> 00:48:22,375
We currently have non embodied
ais Wait till they're embodied.
915
00:48:22,705 --> 00:48:23,155
Yeah.
916
00:48:23,519 --> 00:48:23,849
Aaron Makelky: Yeah.
917
00:48:23,849 --> 00:48:24,089
But.
918
00:48:24,505 --> 00:48:25,615
Lindy Hockenbary: I hadn't heard that one.
919
00:48:25,789 --> 00:48:28,729
Aaron Makelky: Yeah, my, my biggest
takeaway with Open Claw is you'll spend
920
00:48:28,729 --> 00:48:34,129
way more time trying to make it work than
you will find useful things for it to do.
921
00:48:34,219 --> 00:48:35,119
That's one.
922
00:48:35,755 --> 00:48:36,115
Lindy Hockenbary: Yep.
923
00:48:36,199 --> 00:48:40,459
Aaron Makelky: other is if there, if
there is something useful, it's making
924
00:48:40,459 --> 00:48:42,169
the technology accessible to people.
925
00:48:42,169 --> 00:48:45,529
So my example is my family.
926
00:48:46,204 --> 00:48:49,564
Calendar my kids'
wrestling meets right now.
927
00:48:49,564 --> 00:48:53,434
They're what days do they have student
council meetings in the morning, I
928
00:48:53,434 --> 00:48:57,064
wanted to build a thing that I could
say, Hey, in a telegram or a WhatsApp
929
00:48:57,064 --> 00:49:01,354
chat, my wife in plain language
can say, what's the schedule today?
930
00:49:01,844 --> 00:49:03,224
, Where's the next wrestling meet?
931
00:49:03,284 --> 00:49:08,954
And something can look across the
context it needs and say, yeah, in
932
00:49:08,954 --> 00:49:10,784
a telegram chat, here's your answer.
933
00:49:11,789 --> 00:49:15,659
That's really the only advantage,
, and it is difficult to set up.
934
00:49:15,659 --> 00:49:17,849
So if you don't know what you're
doing, make sure you follow a
935
00:49:17,849 --> 00:49:20,429
good tutorial or ask somebody
who's done it a couple of times.
936
00:49:20,429 --> 00:49:24,539
'cause it's easy to put API keys in
a place you don't want them or give
937
00:49:24,539 --> 00:49:28,479
it permission to do something that
you're not gonna wanna trust it to do.
938
00:49:29,420 --> 00:49:30,650
Lindy Hockenbary: It's pretty technical.
939
00:49:30,860 --> 00:49:34,485
That's why a very nichey
group are the only ones.
940
00:49:35,325 --> 00:49:37,935
Really doing it, which are
still quite a lot of people.
941
00:49:38,375 --> 00:49:44,285
, But like you just said, it just got
bought by Open ai, so just wait until it's
942
00:49:44,465 --> 00:49:50,165
embedded in chat, GPT and Atlas, their
agent taking browser and yada yada yada.
943
00:49:50,375 --> 00:49:53,405
And that's the takeaway that I want
educators to take from this is.
944
00:49:55,175 --> 00:49:59,495
I know we covered a lot and there's
a lot that we said that is unknown
945
00:49:59,495 --> 00:50:03,905
and potentially scary and really
overwhelming when you think of the
946
00:50:03,905 --> 00:50:07,745
effects on integrity of student work.
947
00:50:07,775 --> 00:50:11,105
I wanted to get more into that,
but alas, there were just too many
948
00:50:11,195 --> 00:50:12,545
updates and things to talk about.
949
00:50:12,545 --> 00:50:15,005
So maybe we'll do a whole
nother episode on that.
950
00:50:15,005 --> 00:50:21,065
Like how, actually multiple episodes to
think about, you know, how do you adapt?
951
00:50:21,860 --> 00:50:28,310
And adjust because some people are saying
, no, we shouldn't adapt and adjust, but
952
00:50:28,310 --> 00:50:32,990
I don't know how we wouldn't, and formal
education just is not gonna remain.
953
00:50:34,040 --> 00:50:37,940
Relevant, in my opinion, if
we don't adapt and adjust.
954
00:50:38,030 --> 00:50:39,050
So there you go.
955
00:50:39,080 --> 00:50:40,820
There's my, my parting thoughts.
956
00:50:40,820 --> 00:50:47,690
Thank you Erin, so much for coming and
chatting with me and providing silicon.
957
00:50:47,720 --> 00:50:50,390
I'm gonna say you're providing the
Silicon Valley perspective, even
958
00:50:50,390 --> 00:50:52,700
though you are in Wyoming, you're
959
00:50:52,744 --> 00:50:53,074
Aaron Makelky: Yeah,
960
00:50:53,390 --> 00:50:53,690
Lindy Hockenbary: that.
961
00:50:55,084 --> 00:50:58,144
Aaron Makelky: and what I know is due
to working with really smart people
962
00:50:58,144 --> 00:51:02,074
that are light years ahead of me and
just distilling little useful bits that
963
00:51:02,074 --> 00:51:06,004
a teacher or an administrator, someone
like you and I would actually put to use.
964
00:51:06,620 --> 00:51:07,760
Lindy Hockenbary: Yep, for sure.
965
00:51:07,850 --> 00:51:08,330
Awesome.
966
00:51:08,330 --> 00:51:08,840
All right.
967
00:51:08,840 --> 00:51:09,980
Thank you so much.
968
00:51:10,539 --> 00:51:11,189
Aaron Makelky: Thank you, Lindy.
969
00:51:15,092 --> 00:51:17,162
Thanks for joining Make EdTech 100.
970
00:51:17,882 --> 00:51:22,562
I know educator time is valuable and I'm
honored you choose to spend yours with me.
971
00:51:23,612 --> 00:51:28,262
For more EdTech strategies you can use
tomorrow and ways to bring me to your
972
00:51:28,262 --> 00:51:30,762
school or event, head to LindyHoc.com.
973
00:51:31,952 --> 00:51:35,852
If this episode resonated, hit subscribe
so you don't miss the next one.
974
00:51:36,632 --> 00:51:36,972
I'm LindyHoc.
975
00:51:37,652 --> 00:51:39,962
Go forth and make EdTech 100.
00:00:03,300 --> 00:00:05,400
Welcome to make EdTech 100.
2
00:00:06,180 --> 00:00:10,530
I am LindyHoc Educator, K 12
Ed Tech Advisor, and your host.
3
00:00:11,340 --> 00:00:15,989
This is a podcast where we keep it real
about what actually works in classrooms.
4
00:00:16,440 --> 00:00:21,450
No hype, no overwhelm, just practical
strategies, honest stories and
5
00:00:21,450 --> 00:00:25,050
tools that make a real difference
for teachers and students.
6
00:00:25,110 --> 00:00:28,200
So come along with me on a
journey to make EdTech 100.
7
00:00:33,031 --> 00:00:35,431
Lindy Hockenbary: So if you didn't
know, and if you're not keeping
8
00:00:35,431 --> 00:00:40,081
up, there's a lot happening with AI
developments and advancements out there.
9
00:00:40,111 --> 00:00:45,901
So much so that I have only been barely
keeping up, and it is my job to keep up.
10
00:00:45,901 --> 00:00:48,721
So I know it's bad and
I know it's moving fast.
11
00:00:48,721 --> 00:00:50,431
I've been struggling
for the last few years.
12
00:00:50,851 --> 00:00:53,701
To keep up because of all the
advancements, but the last couple
13
00:00:53,701 --> 00:00:56,461
months especially have just been crazy.
14
00:00:56,461 --> 00:01:01,921
There's a lot of big
jumps in AI capabilities.
15
00:01:01,921 --> 00:01:05,191
So for this episode, I'm
doing a little bit different.
16
00:01:05,221 --> 00:01:09,901
Instead of having a guest that I
interview, I have an invited Erin on to
17
00:01:09,901 --> 00:01:16,080
be kind of like my co-host to help me
just talk through and unpack all of the
18
00:01:16,080 --> 00:01:19,201
AI craziness from the past few months.
19
00:01:19,201 --> 00:01:20,280
So, welcome, Erin.
20
00:01:21,159 --> 00:01:22,359
Aaron Makelky: Thank you,
Lindy, for having me.
21
00:01:22,359 --> 00:01:23,259
I'm flattered.
22
00:01:23,720 --> 00:01:26,211
Lindy Hockenbary: What do
listeners need to know about you?
23
00:01:27,259 --> 00:01:31,149
Aaron Makelky: Yeah, well I've been a
career educator, , in rural states as
24
00:01:31,149 --> 00:01:37,509
recently as the 2025 school year, but I've
since pivoted to developer relations and
25
00:01:37,509 --> 00:01:39,489
community marketing at a tech company.
26
00:01:40,179 --> 00:01:44,019
And I always tell people my skillset,
I'm not the most techie person.
27
00:01:44,019 --> 00:01:48,699
I'm not a developer or coder,
but I'm the conduit between those
28
00:01:48,699 --> 00:01:50,979
really techie developer types.
29
00:01:50,979 --> 00:01:55,119
And the average person, whether it's
a teacher, an administrator, or small
30
00:01:55,119 --> 00:01:59,199
business owner, I can speak the language
of both, and I'm somewhere in the middle.
31
00:02:00,536 --> 00:02:01,306
Lindy Hockenbary: Yep, same.
32
00:02:01,306 --> 00:02:02,116
Same with me.
33
00:02:02,116 --> 00:02:04,096
It's just I am very still much in.
34
00:02:04,446 --> 00:02:06,875
Education, career educators
still in education.
35
00:02:06,875 --> 00:02:12,006
You made the hop full on to work for
a tech company, but that's one of the
36
00:02:12,006 --> 00:02:13,446
reasons I wanted to bring you on today.
37
00:02:13,446 --> 00:02:17,766
First of all, you're one of the people
that I keep up with most on social media
38
00:02:17,766 --> 00:02:21,846
'cause you're always sharing really
awesome tips and how you're using ai.
39
00:02:21,846 --> 00:02:24,935
And when you were in the classroom
still, you were sharing so many great
40
00:02:24,935 --> 00:02:29,315
things about how you were embracing the
technology and teaching your students
41
00:02:29,315 --> 00:02:32,166
how to use it correctly and responsibly.
42
00:02:32,166 --> 00:02:33,041
I think you were my.
43
00:02:33,741 --> 00:02:39,500
Most followed person on LinkedIn in
2025 because you know, they did that
44
00:02:39,500 --> 00:02:41,870
little thing at the end of the year
where they told you who you like,
45
00:02:42,020 --> 00:02:43,640
who you like most engaged with.
46
00:02:43,640 --> 00:02:47,491
I'm pretty sure you were , my most
engaged with person, so I wanted to
47
00:02:47,491 --> 00:02:50,730
bring you on because you are threefold.
48
00:02:51,000 --> 00:02:54,840
You're doing a lot with the tech
yourself and keeping up with it yourself.
49
00:02:54,870 --> 00:02:58,650
You're also working for a tech
company now, and it's based
50
00:02:58,650 --> 00:02:59,970
out of the Bay Area, right?
51
00:03:00,439 --> 00:03:00,859
Aaron Makelky: It is.
52
00:03:00,859 --> 00:03:01,609
Yes.
53
00:03:01,900 --> 00:03:02,230
Lindy Hockenbary: Yeah.
54
00:03:02,230 --> 00:03:06,160
So like you are, even though you're
working remote from Wyoming, you're
55
00:03:06,160 --> 00:03:11,350
embedded in virtual calls all day long
with those that are like, you know,
56
00:03:11,410 --> 00:03:13,690
the tech people in Silicon Valley.
57
00:03:13,810 --> 00:03:15,700
So you kind of have that perspective now.
58
00:03:15,940 --> 00:03:18,700
And then also you have
the educator perspective.
59
00:03:18,700 --> 00:03:20,110
'cause how many years did you teach?
60
00:03:20,654 --> 00:03:23,119
Aaron Makelky: I was a
teacher for six 16 years.
61
00:03:23,209 --> 00:03:26,659
Uh, and I, I have my master's
in K 12 ed leadership and a
62
00:03:26,659 --> 00:03:30,779
principal endorsement, but always
preferred to be in the classroom.
63
00:03:30,839 --> 00:03:32,459
, Despite some of my family, my.
64
00:03:32,489 --> 00:03:34,499
Members that were career
administrators, I thought the
65
00:03:34,499 --> 00:03:36,059
classroom was a better fit for me.
66
00:03:36,750 --> 00:03:38,490
Lindy Hockenbary: Your dad
was an administrator, right?
67
00:03:38,919 --> 00:03:39,219
Aaron Makelky: Yep.
68
00:03:39,219 --> 00:03:41,349
And my identical principal.
69
00:03:41,349 --> 00:03:41,589
Yep.
70
00:03:41,640 --> 00:03:41,850
Lindy Hockenbary: right.
71
00:03:42,459 --> 00:03:43,749
Aaron Makelky: And my identical twin.
72
00:03:44,469 --> 00:03:44,919
Mm-hmm.
73
00:03:45,030 --> 00:03:46,680
Lindy Hockenbary: him on
LinkedIn too, and every now and
74
00:03:46,680 --> 00:03:48,090
again I'm like, oh, it's Aaron.
75
00:03:48,090 --> 00:03:49,020
No, it's not Aaron.
76
00:03:49,050 --> 00:03:49,620
No, it's not.
77
00:03:50,559 --> 00:03:52,089
Aaron Makelky: Same DNA different person,
78
00:03:53,250 --> 00:03:53,610
Lindy Hockenbary: person.
79
00:03:53,610 --> 00:03:54,000
I love it.
80
00:03:54,150 --> 00:03:56,790
I still have to meet
him in person sometime.
81
00:03:56,790 --> 00:03:57,720
E is it Eric?
82
00:03:58,449 --> 00:03:58,809
Aaron Makelky: Eric.
83
00:03:58,809 --> 00:03:59,139
Yep.
84
00:03:59,475 --> 00:03:59,925
Lindy Hockenbary: Eric?
85
00:03:59,925 --> 00:04:00,555
Eric and Aaron.
86
00:04:00,555 --> 00:04:01,155
Yeah, that's right.
87
00:04:01,215 --> 00:04:01,575
Okay.
88
00:04:01,965 --> 00:04:04,815
And he, yeah, he's educator
too slash administrator.
89
00:04:04,815 --> 00:04:05,115
So, yeah.
90
00:04:05,115 --> 00:04:09,165
So point being, you bring all three
of those different perspectives in
91
00:04:09,165 --> 00:04:14,415
to give listeners a really good idea
of Where, where we're at with ai.
92
00:04:14,415 --> 00:04:16,215
So I wanna start with a story.
93
00:04:16,425 --> 00:04:19,425
There's something about TCA conference.
94
00:04:19,425 --> 00:04:22,875
It's always like the first week
of February, and for the last.
95
00:04:23,535 --> 00:04:24,825
Two years.
96
00:04:25,515 --> 00:04:31,065
All of the big AI companies have released
like a big update like either right
97
00:04:31,065 --> 00:04:36,135
before right around February 1st, which
E either right before or right during
98
00:04:36,255 --> 00:04:43,275
TCA last year, open AI released operator,
which if you're not familiar, they've
99
00:04:43,275 --> 00:04:47,295
now rolled it into their agent feature,
so it's no longer called operator.
100
00:04:47,565 --> 00:04:50,655
But at the time it was
basically like their agentic.
101
00:04:52,305 --> 00:04:55,545
And if you don't know what agent is, we're
gonna dive into that, where you could
102
00:04:55,545 --> 00:05:02,025
actually tell it to go research a vacation
that you wanted to take or draft an email.
103
00:05:02,085 --> 00:05:05,175
I had it in my Gmail,
drafting my emails for me.
104
00:05:05,505 --> 00:05:08,865
Uh, and it would open up a browser
within the window and all these things.
105
00:05:09,075 --> 00:05:12,585
So I was frantically scrambling
at TCA next year, last year,
106
00:05:12,645 --> 00:05:15,765
excuse me, 2025 to make sure that.
107
00:05:16,175 --> 00:05:19,625
I was making sure every one of my
sessions, the people that came,
108
00:05:19,865 --> 00:05:25,775
knew what a agentic AI was and
very, very, very few people did.
109
00:05:25,985 --> 00:05:26,315
Okay.
110
00:05:26,315 --> 00:05:27,605
Fast forward a year.
111
00:05:27,845 --> 00:05:30,245
I was just at TCAA couple weeks ago.
112
00:05:30,245 --> 00:05:32,885
First week of February, same thing.
113
00:05:33,065 --> 00:05:34,985
Huge announcements coming in.
114
00:05:34,985 --> 00:05:38,645
So OpenAI release codex, which we're
gonna hopefully talk about your
115
00:05:38,885 --> 00:05:43,745
anthropic release, their Opus 4.6
model, which holy cow, I was actually
116
00:05:43,745 --> 00:05:45,425
texting my husband this morning.
117
00:05:45,780 --> 00:05:49,770
Telling him that he needs to go
check it out because now you can
118
00:05:49,770 --> 00:05:55,140
literally give it a Word document
and it will edit the actual Word
119
00:05:55,140 --> 00:05:57,360
document and spit it back out to you.
120
00:05:57,360 --> 00:05:59,790
So before, I would always
like upload a document.
121
00:06:00,060 --> 00:06:01,200
I'd ask for changes.
122
00:06:01,200 --> 00:06:01,980
It might give it to me.
123
00:06:01,980 --> 00:06:05,190
It might have given you a file that you
could download, but you still had to do
124
00:06:05,190 --> 00:06:06,960
a lot of like copy pasting, formatting.
125
00:06:07,020 --> 00:06:09,360
Nope, it does all the
formatting for you now.
126
00:06:09,360 --> 00:06:13,080
So that was released and then on
top of all of that, there was.
127
00:06:13,165 --> 00:06:18,955
All of this craziness going on
with Claude Bot, open Claw and mbo.
128
00:06:19,285 --> 00:06:21,775
And if you're not familiar, we're
gonna dive into that as well.
129
00:06:21,775 --> 00:06:27,085
So that was all going on in the midst
of TCAI was like sharing TikTok videos
130
00:06:27,085 --> 00:06:32,265
just to kind of stay on top of it and
inform my TikTok subscribers about it.
131
00:06:32,295 --> 00:06:34,845
And then so all of that
was happening, right?
132
00:06:35,320 --> 00:06:38,380
I am, I actually had time to go to
a few sessions at TCA this year.
133
00:06:38,380 --> 00:06:45,970
I'm sitting in a session agentic browsers
came up and the whole room, and it was a
134
00:06:45,970 --> 00:06:48,280
packed room and it was a pretty big room.
135
00:06:48,580 --> 00:06:50,800
Came to a halt with mass.
136
00:06:51,385 --> 00:06:56,725
Chaos for either a combination of, I
think the majority of the chaos was they
137
00:06:56,725 --> 00:06:58,315
had never heard the word ag agentic.
138
00:06:58,315 --> 00:07:00,625
They had no idea what an
age agentic browser meant.
139
00:07:00,905 --> 00:07:03,695
, The other little part of
the chaos was immediate.
140
00:07:03,695 --> 00:07:07,385
They knew it, but they didn't know
the feature that was being shown.
141
00:07:07,385 --> 00:07:09,875
And they immediately started
talking about how to block it.
142
00:07:10,175 --> 00:07:14,855
And the presenter did a really great
job of making sure that it didn't
143
00:07:14,885 --> 00:07:20,255
totally change the course of the
session, but I was just sitting there.
144
00:07:21,645 --> 00:07:27,945
going, oh my gosh, these like,
we have some major work to do.
145
00:07:27,945 --> 00:07:34,545
We've been over the last year, agent
a KI has been the big push behind
146
00:07:34,965 --> 00:07:37,006
a lot of the changes in updates.
147
00:07:37,006 --> 00:07:38,470
Would you agree or not agree, Erin?
148
00:07:39,644 --> 00:07:40,274
Aaron Makelky: Totally agree.
149
00:07:40,364 --> 00:07:45,674
I think a lot of us about two
years ago saw copy, paste, chat.
150
00:07:45,674 --> 00:07:49,214
GPT is in a tab or cloud,
whatever you're using, and we
151
00:07:49,214 --> 00:07:50,954
thought, wow, AI is so awesome.
152
00:07:51,524 --> 00:07:54,674
And the new version of
it is it can do things.
153
00:07:55,059 --> 00:07:56,889
On your behalf or for you?
154
00:07:57,399 --> 00:08:01,509
I'll share how I previewed
agen browsers with my students.
155
00:08:01,509 --> 00:08:04,929
I gave them a Google Doc in a
high school social studies class
156
00:08:05,379 --> 00:08:09,309
that had a list of eight or nine
questions about a historical topic.
157
00:08:09,309 --> 00:08:13,389
They didn't have a textbook or a
reading on, and they were supposed
158
00:08:13,389 --> 00:08:15,969
to go out and do some research,
and it's like a background
159
00:08:15,969 --> 00:08:17,619
builder on the League of Nations.
160
00:08:17,619 --> 00:08:18,244
When was it founded?
161
00:08:19,104 --> 00:08:20,934
Which country came up with the idea?
162
00:08:21,564 --> 00:08:26,844
And I said, I just wanna watch how you do
this to see, do you go to Google search?
163
00:08:26,844 --> 00:08:27,474
Do you go to chat?
164
00:08:27,474 --> 00:08:28,014
GPT?
165
00:08:28,014 --> 00:08:29,334
What tools do you use?
166
00:08:29,994 --> 00:08:34,884
But it was a trick because I knew most
of them opened a Google tab and they
167
00:08:34,884 --> 00:08:36,834
copy paste the question to Google search.
168
00:08:36,834 --> 00:08:38,124
A couple of them used ai.
169
00:08:39,084 --> 00:08:40,254
I let 'em finish.
170
00:08:40,644 --> 00:08:44,514
And then what I did is I took Comet
browser, I opened the same Google
171
00:08:44,514 --> 00:08:49,944
Doc and I said, go find the answers
and fill it in on my Google Doc.
172
00:08:50,064 --> 00:08:57,024
I hit run and then to be funny, I took
my Comet browser water bottle and I
173
00:08:57,024 --> 00:08:59,154
went down the hall to get a drink.
174
00:08:59,499 --> 00:09:02,649
Then I just watched my classroom through
the window, so I was still kind of
175
00:09:02,649 --> 00:09:07,659
supervising him and they're like, it's
opening tabs, it's researching, and then
176
00:09:07,659 --> 00:09:11,979
it's going back to the doc and it's typing
an like, is he remote controlling this?
177
00:09:11,979 --> 00:09:12,609
What the heck?
178
00:09:12,609 --> 00:09:14,829
And then I would just wave
at him through the window.
179
00:09:15,669 --> 00:09:17,529
And the reason I did that was 'cause.
180
00:09:17,869 --> 00:09:19,369
It was gonna show up eventually.
181
00:09:19,369 --> 00:09:20,839
Now it's in Google Chrome.
182
00:09:20,929 --> 00:09:22,999
Claude has a Chrome extension.
183
00:09:22,999 --> 00:09:26,509
So if you're a Claude person, you
get essentially the same thing
184
00:09:27,019 --> 00:09:28,639
and it can control your browser.
185
00:09:28,639 --> 00:09:32,299
And that was my moment of chaos where
people thought it was a trick or it
186
00:09:32,299 --> 00:09:34,369
wasn't real, and it's just like, nope.
187
00:09:34,399 --> 00:09:35,419
That's what they can do now.
188
00:09:35,449 --> 00:09:36,829
'cause they're built into your browser.
189
00:09:37,560 --> 00:09:37,771
Lindy Hockenbary: Yep.
190
00:09:37,861 --> 00:09:41,161
And yeah, if you've, if you've never
seen an agentic browser or any sort
191
00:09:41,161 --> 00:09:46,171
of AI agent or agentic ai, it really
does at first seem like a magic
192
00:09:46,171 --> 00:09:48,031
trick, like someone is pranking you.
193
00:09:48,031 --> 00:09:53,191
My students used to prank me all
the time and like set sticky keys.
194
00:09:53,411 --> 00:09:55,961
On the keyboards and the
computers in my classroom.
195
00:09:56,290 --> 00:10:00,191
And then, you know, somebody would it
either I caught onto it pretty quickly,
196
00:10:00,191 --> 00:10:02,711
but then like somebody else would come
in and start using the computer and
197
00:10:02,711 --> 00:10:04,471
think that the computer is possessed.
198
00:10:04,531 --> 00:10:06,121
You know, 'cause they couldn't
figure out the sticky keys.
199
00:10:06,271 --> 00:10:11,821
It's kind of like that experience, but
like way more intense than sticky keys.
200
00:10:12,061 --> 00:10:17,131
And if you're not familiar, so
a agentic AI is basically ai.
201
00:10:17,596 --> 00:10:21,586
That acts, and I always tell people when
I say this, this creates a lot of anxiety.
202
00:10:21,586 --> 00:10:22,756
So I'm always like, take a deep breath.
203
00:10:23,241 --> 00:10:25,281
Seriously, take a deep breath.
204
00:10:26,031 --> 00:10:27,741
It acts autonomously.
205
00:10:27,861 --> 00:10:31,341
So exactly what Aaron was saying,
you're not now just like going to chat
206
00:10:31,461 --> 00:10:36,111
BT or Claude or Gemini and giving it
a prompt and copy pasting over into
207
00:10:36,111 --> 00:10:38,961
a Google Doc, like it's editing and.
208
00:10:39,251 --> 00:10:44,741
Adding to the Google Doc for you, it's
autonomous, but you gave the perfect
209
00:10:44,741 --> 00:10:49,211
example of, I always add the caveat
that it's autonomous with oversight
210
00:10:49,211 --> 00:10:53,441
because you gave it the Google Doc,
you gave it the questions in the Google
211
00:10:53,441 --> 00:10:57,251
Doc that you wanted to do, so you
gave it instructions as the human to
212
00:10:57,281 --> 00:10:58,961
say, this is what I want you to do.
213
00:10:58,961 --> 00:11:03,401
I want you to answer these questions about
the League of Nations in this Google Doc.
214
00:11:03,401 --> 00:11:03,700
Aaron Makelky: Yep.
215
00:11:03,731 --> 00:11:07,511
And the way that I think of it,
the super simple, not as nuanced
216
00:11:07,511 --> 00:11:11,170
definition is agent AI has tools.
217
00:11:11,170 --> 00:11:12,611
So it's kinda like if you gave chat.
218
00:11:13,031 --> 00:11:17,981
GPTA toolkit and it has a wrench for
this job and a hammer for this job only.
219
00:11:17,981 --> 00:11:23,440
It might be a browser and it can do
a search, and then it can call this
220
00:11:23,440 --> 00:11:26,501
service over here and then it can
write to your word or your Google
221
00:11:26,501 --> 00:11:29,651
doc, and it's giving the same brain.
222
00:11:29,680 --> 00:11:30,881
That's the ai.
223
00:11:31,005 --> 00:11:33,015
The tools and now it's an agent.
224
00:11:33,435 --> 00:11:37,095
So it could be a recurring schedule
or a trigger like when I get an email
225
00:11:37,095 --> 00:11:41,905
from this person, do this, or Monday
mornings at a 8:00 AM run this, , or
226
00:11:41,905 --> 00:11:45,625
like the browser example, you open
up the tabs, it needs for context
227
00:11:45,625 --> 00:11:47,070
and say, go off and do this for me..
228
00:11:47,857 --> 00:11:48,097
Lindy Hockenbary: Yep.
229
00:11:48,187 --> 00:11:48,577
Exactly.
230
00:11:48,577 --> 00:11:51,517
And I always get the question at this
point, they're like, okay, well what
231
00:11:51,577 --> 00:11:55,807
they still don't, if you're new to Agent
ai, you may not still see the difference
232
00:11:55,807 --> 00:12:02,747
between a generative AI tool like chat,
GBT, Jim and I, Claude An agentic AI
233
00:12:02,747 --> 00:12:07,097
tool or an AI agent, which there are
subtle differences between the two,
234
00:12:07,097 --> 00:12:09,197
which we probably won't get into today.
235
00:12:09,457 --> 00:12:15,787
, , the way I explain it is , let's use email
for an example chat, GPT, generative ai.
236
00:12:15,787 --> 00:12:19,537
You have to go open the large
language model, or you might
237
00:12:19,537 --> 00:12:21,667
have it built in to like Gmail.
238
00:12:21,667 --> 00:12:23,977
Now there's some generative
built in there, right?
239
00:12:24,187 --> 00:12:29,737
But you still have to prompt it to say,
write this email, review it, and hit send.
240
00:12:30,137 --> 00:12:32,837
Agentic ai, you tell it.
241
00:12:32,927 --> 00:12:35,207
I want you to respond to my emails.
242
00:12:35,387 --> 00:12:39,137
Tell everyone that I'm gonna be out next
week at the NCC Conference in Seattle.
243
00:12:39,467 --> 00:12:45,827
And it goes, opens the email,
drafts that response and hits send.
244
00:12:46,157 --> 00:12:47,357
That's agentic ai.
245
00:12:47,357 --> 00:12:50,387
That's that autonomous part to it.
246
00:12:50,387 --> 00:12:54,077
Aaron Makelky: Which is a scary thing
'cause people always go, what if it sends
247
00:12:54,077 --> 00:12:55,817
the wrong message to the wrong person?
248
00:12:55,817 --> 00:12:58,637
Or books the wrong flight for
me, or whatever task you gave it.
249
00:12:59,504 --> 00:13:02,684
Lindy Hockenbary: It is scary and you
know, it's, there's a lot of discussion.
250
00:13:02,684 --> 00:13:05,054
We'll get into this a little bit
more when we talk about the claw
251
00:13:05,054 --> 00:13:09,494
bot, mot bot, whatever they're
calling it now thing, but it's also.
252
00:13:09,854 --> 00:13:14,144
Uh, it opens the question about
security risk and data privacy, right?
253
00:13:14,144 --> 00:13:18,194
Like now you're giving, you, you,
we've already been talking for years
254
00:13:18,224 --> 00:13:22,244
three over three years since chat
GPD was released about not putting
255
00:13:22,244 --> 00:13:25,454
an education like PII, making sure
you're a FERPA compliant, right?
256
00:13:25,454 --> 00:13:29,294
Don't put PII into a
large language model now.
257
00:13:30,389 --> 00:13:35,909
It's like 20 steps more than that of not
only just don't give it your information,
258
00:13:35,909 --> 00:13:40,239
but now you're literally giving the , AI
access to things like your email that
259
00:13:40,239 --> 00:13:46,299
might have really sensitive information
in it, depending upon what your job is.
260
00:13:46,299 --> 00:13:49,749
But if you're a teacher or a
school administrator, your email,
261
00:13:49,929 --> 00:13:53,139
your, if you use Google Drive
or OneDrive, you're giving it.
262
00:13:53,459 --> 00:13:56,999
Access to that potentially
has a lot of sensitive info.
263
00:13:56,999 --> 00:14:02,489
So that's a huge discussion around agent
to AI and education is that data privacy.
264
00:14:02,489 --> 00:14:05,999
And that's probably
one of the reasons why.
265
00:14:06,804 --> 00:14:10,374
All those educators, not all the
majority of the educators in that
266
00:14:10,374 --> 00:14:12,684
session room at TCA, just had no idea.
267
00:14:12,684 --> 00:14:15,324
Like I even heard someone
say, I'm so confused.
268
00:14:15,354 --> 00:14:17,964
She was looking around like this
because she was like, she had
269
00:14:17,964 --> 00:14:19,254
never even heard the word ag.
270
00:14:19,254 --> 00:14:21,834
And then you put Ag Agentic browser
in there and you know, a whole
271
00:14:21,834 --> 00:14:27,484
nother level of confusion because
it's no secret that K 12 education
272
00:14:27,484 --> 00:14:30,274
has a lot of barriers to innovation.
273
00:14:31,309 --> 00:14:34,339
One of those being ferpa, data privacy.
274
00:14:34,339 --> 00:14:41,539
We're working with miners under the age
of 18, so the tech tends to move slower.
275
00:14:42,019 --> 00:14:46,589
And not just that, but other reasons as
well than it does in other industries in.
276
00:14:46,589 --> 00:14:46,859
Aaron Makelky: Yep.
277
00:14:47,039 --> 00:14:51,179
And I think if you haven't come to
this realization yet, it's pretty
278
00:14:51,179 --> 00:14:58,019
unanimous by people who see this stuff
every day in 2026 if you're relying
279
00:14:58,019 --> 00:15:02,849
on anything in a web browser or on
a computer to assess your students'
280
00:15:02,849 --> 00:15:05,399
learning or track their progress.
281
00:15:05,909 --> 00:15:08,274
I'm not saying they're all using Chrome.
282
00:15:09,189 --> 00:15:10,959
Having an agent do it for them.
283
00:15:10,959 --> 00:15:14,919
I'm saying they very well
could be even with free tools.
284
00:15:15,459 --> 00:15:17,529
So if it's something in
a Google Doc, I'll share.
285
00:15:17,529 --> 00:15:18,519
Here's an example.
286
00:15:19,089 --> 00:15:23,859
One way people said to monitor for AI
doing the work of students is go back
287
00:15:23,859 --> 00:15:27,879
to the revision history of a Google Doc
and see if they copied and pasted in
288
00:15:27,879 --> 00:15:31,479
a giant four page essay all in one go.
289
00:15:31,479 --> 00:15:32,349
Like that was ai.
290
00:15:32,349 --> 00:15:33,249
That was cheating.
291
00:15:33,584 --> 00:15:35,654
Uh, that's not my preferred way.
292
00:15:35,654 --> 00:15:38,624
If that's what you do, okay,
here's the counter argument.
293
00:15:38,804 --> 00:15:43,754
And now a browser agent could be doing
that for them, and it's not gonna just
294
00:15:43,754 --> 00:15:47,024
copy and paste one chunk from the tab.
295
00:15:47,024 --> 00:15:48,344
They did the prompt in.
296
00:15:48,674 --> 00:15:52,484
It's actually gonna be in
Canvas or in Schoology or in
297
00:15:52,484 --> 00:15:54,584
Google classroom, navigating.
298
00:15:54,989 --> 00:15:57,329
It's gonna show the person's logged in.
299
00:15:57,329 --> 00:16:00,269
They made this edit at this
time, and then they opened here.
300
00:16:00,869 --> 00:16:04,439
And I've even had students show me,
you can prompt it to write out the
301
00:16:04,439 --> 00:16:10,139
thing, slowly put in a couple of
typos, then go back and edit them.
302
00:16:10,139 --> 00:16:13,109
And all of this is running without
the kid touching the keyboard.
303
00:16:13,109 --> 00:16:16,319
But the revision history
is gonna look organic.
304
00:16:16,319 --> 00:16:17,909
It's gonna look like a person did it.
305
00:16:17,909 --> 00:16:21,284
Lindy Hockenbary: That came up in
that session that I was at, at TCA
306
00:16:21,284 --> 00:16:24,929
is like, and even like Bri teaching,
they have an inspect writing tool.
307
00:16:25,289 --> 00:16:29,459
Built into that to show if any big
blocks were copy pasted or whatever.
308
00:16:29,669 --> 00:16:34,279
But now these agents, when I gave
operator a year ago access to my email,
309
00:16:34,279 --> 00:16:39,919
I sat there and watched it type out the
email and look exactly what it would
310
00:16:39,919 --> 00:16:41,959
look like if I was typing out the email.
311
00:16:42,109 --> 00:16:44,959
Maybe the only difference, which
would be a subtle difference
312
00:16:44,959 --> 00:16:49,609
that you'd have to really watch
is the typing speed and typos.
313
00:16:50,344 --> 00:16:54,244
The AI isn't gonna have typos
and be using the backspace like a
314
00:16:54,244 --> 00:16:57,154
human would when you're naturally
just typing something out.
315
00:16:57,424 --> 00:17:00,574
And the educators in that session,
some of them piped up and said, well,
316
00:17:00,694 --> 00:17:05,074
I was already having that problem with
students, figuring out that teachers
317
00:17:05,074 --> 00:17:08,434
are using these tools to see if I
was copying and pasting a paragraph.
318
00:17:08,674 --> 00:17:09,844
So what do they do?
319
00:17:09,924 --> 00:17:12,894
They write their essay in a
large language model, do a split
320
00:17:12,894 --> 00:17:14,994
screen, and then they type it.
321
00:17:16,344 --> 00:17:19,584
They're literally just like I think of
back in the day, you see all of like the
322
00:17:19,584 --> 00:17:23,484
movies where they made the kids write
lines and that was their punishment.
323
00:17:24,834 --> 00:17:25,494
So guess what?
324
00:17:25,554 --> 00:17:25,914
Now?
325
00:17:25,914 --> 00:17:28,854
Now the kid is literally,
that assignment a
326
00:17:28,907 --> 00:17:29,147
Aaron Makelky: It's,
327
00:17:29,274 --> 00:17:29,934
Lindy Hockenbary: punishment.
328
00:17:30,197 --> 00:17:31,967
Aaron Makelky: it's like
typing practice to the kid.
329
00:17:31,967 --> 00:17:35,627
They didn't come up with the, the
thinking or the ideating or the editing.
330
00:17:35,627 --> 00:17:37,757
It's just like, I'll copy it manually.
331
00:17:38,897 --> 00:17:39,227
Yeah.
332
00:17:39,227 --> 00:17:39,527
And there's.
333
00:17:39,952 --> 00:17:41,152
There's chrome extensions.
334
00:17:41,152 --> 00:17:46,282
I saw a student make one that you
paste in your essay, so probably chat.
335
00:17:46,282 --> 00:17:47,602
GPT generated it.
336
00:17:47,932 --> 00:17:52,192
The chrome extension is open in your
browser and it's gonna type it at a
337
00:17:52,192 --> 00:17:55,972
human speed and you can dictate that
and you just paste in your whole
338
00:17:55,972 --> 00:18:00,142
essay and it's gonna go into your
Google Doc and character by character.
339
00:18:00,142 --> 00:18:03,022
Make it look on the revision
history like a human did it.
340
00:18:03,022 --> 00:18:07,312
So I doubt many students use
that, but they're free and
341
00:18:07,312 --> 00:18:08,752
they're on the internet, so.
342
00:18:09,077 --> 00:18:12,197
The crafty ones are trying to find
ways, and you shouldn't assume that's
343
00:18:12,197 --> 00:18:13,367
the best way to filter for cheating.
344
00:18:15,214 --> 00:18:16,414
Lindy Hockenbary: If you're listening
to this and you're like, oh my
345
00:18:16,414 --> 00:18:18,184
gosh, I need to go try this myself.
346
00:18:18,184 --> 00:18:22,114
I need to know what's going on or what
tools I should use or what tools students
347
00:18:22,114 --> 00:18:23,524
are using to do this kind of stuff.
348
00:18:23,744 --> 00:18:26,054
I wanna give like a layered approach here.
349
00:18:26,294 --> 00:18:28,634
The first one , would
be the agents within.
350
00:18:30,029 --> 00:18:31,109
Chat, GBT.
351
00:18:31,409 --> 00:18:32,069
Claude.
352
00:18:32,069 --> 00:18:32,639
I'm trying to think.
353
00:18:32,639 --> 00:18:35,699
Gemini doesn't really have
agents yet, I don't think.
354
00:18:35,699 --> 00:18:35,759
No.
355
00:18:37,287 --> 00:18:37,557
Aaron Makelky: Nope.
356
00:18:37,617 --> 00:18:39,447
Just built into the browser now.
357
00:18:39,674 --> 00:18:40,124
Lindy Hockenbary: Yeah.
358
00:18:40,574 --> 00:18:43,364
Um, I'm gonna open in another tab,
but, um, yeah, built in the browser,
359
00:18:43,364 --> 00:18:44,474
which I'll get to in a second.
360
00:18:44,594 --> 00:18:44,954
Okay.
361
00:18:45,014 --> 00:18:47,834
So those have been there for a
while, like over a year, I would say
362
00:18:47,834 --> 00:18:49,304
maybe a little over a year for chat.
363
00:18:49,304 --> 00:18:49,904
GPT.
364
00:18:50,234 --> 00:18:51,404
So that's number one.
365
00:18:51,404 --> 00:18:54,884
If you go into chat, GPT, hit
the tools button, turn on agents,
366
00:18:55,094 --> 00:18:57,134
and people are always like,
well, I don't know what to do.
367
00:18:57,434 --> 00:18:59,894
If you don't know what to
do to start, guess what?
368
00:19:00,314 --> 00:19:02,474
It's an ai so you ask it.
369
00:19:03,254 --> 00:19:04,694
What can you do for me?
370
00:19:04,964 --> 00:19:05,174
How
371
00:19:05,277 --> 00:19:05,567
Aaron Makelky: Yeah.
372
00:19:05,804 --> 00:19:06,254
Lindy Hockenbary: me?
373
00:19:06,404 --> 00:19:06,794
Right?
374
00:19:06,944 --> 00:19:11,804
And if you have your in chat, GPT
especially, where if you have the setting
375
00:19:11,804 --> 00:19:16,274
turned on where it learns you and it
learns information from your different
376
00:19:16,274 --> 00:19:18,099
chats, if you use chat GPTA lot.
377
00:19:18,899 --> 00:19:22,649
It's gonna know you're a high
school social studies teacher.
378
00:19:22,679 --> 00:19:27,899
It knows everything about my work,
so I have to give it no context.
379
00:19:27,899 --> 00:19:30,359
Now when I go in there, so if
I go in there and I'm like,
380
00:19:30,359 --> 00:19:31,709
Hey, what can I do with you?
381
00:19:31,859 --> 00:19:35,519
It'll start giving me
very customized examples.
382
00:19:35,519 --> 00:19:38,369
And then from there you could
even say, give me a prompt,
383
00:19:38,789 --> 00:19:41,039
copy, paste, put the prompt in.
384
00:19:41,129 --> 00:19:43,499
And just, that's the
way I start exploring.
385
00:19:43,919 --> 00:19:45,299
These types of tools now.
386
00:19:45,329 --> 00:19:46,109
So there's that.
387
00:19:46,379 --> 00:19:51,299
Then the next level would be, we mentioned
it, the Chrome browser, which is by far
388
00:19:51,299 --> 00:19:54,509
the most common browser, and I would say
by far, the most common browser in K 12
389
00:19:54,509 --> 00:19:58,199
education now has Gemini built into it.
390
00:19:58,499 --> 00:20:02,699
So if you haven't noticed over the last
few months in the top right corner,
391
00:20:02,699 --> 00:20:06,839
you now have a Gemini button that
you can click if you don't see it.
392
00:20:07,599 --> 00:20:11,289
could be if you're logged into your
school account, it's either that your
393
00:20:11,289 --> 00:20:14,739
school has blocked it in some way, or
schools, depending upon their settings
394
00:20:14,739 --> 00:20:19,299
with Google Workspace, can do either a
rapid release or a scheduled release.
395
00:20:19,509 --> 00:20:22,839
So you might be on a scheduled release
where it hasn't been, you don't get
396
00:20:22,839 --> 00:20:26,974
the new features for a while until
after they're, they're out there.
397
00:20:27,594 --> 00:20:29,534
That's the one that,
honestly, so I mentioned.
398
00:20:30,219 --> 00:20:33,639
The concept of in that TC, a session
of agentic browsers came up, but
399
00:20:33,639 --> 00:20:38,049
then the presenter showed an example
using the Gemini within Chrome.
400
00:20:38,649 --> 00:20:42,279
This is not a full agentic browser yet.
401
00:20:42,279 --> 00:20:47,019
And Aaron, you said it perfectly,
it's actually not the most effective.
402
00:20:47,594 --> 00:20:52,964
There's, when we get into like the
comment and the atlas, those are way more
403
00:20:53,024 --> 00:20:55,184
effective than Gemini built within Chrome.
404
00:20:55,184 --> 00:20:58,874
Although I will say I've been using
the Gemini built within Chrome a lot.
405
00:20:59,174 --> 00:21:04,014
The example that the presenter at
TCA did, he had a, a question up
406
00:21:04,494 --> 00:21:06,804
or it was like a writing prompt or
something and he hit the button.
407
00:21:06,804 --> 00:21:08,334
He just acted like a high school kid.
408
00:21:08,334 --> 00:21:11,364
Like, Hey dude, uh, I'm working
on this problem and I can't
409
00:21:11,364 --> 00:21:13,524
figure out how, how to solve it.
410
00:21:13,524 --> 00:21:14,304
Can you solve it for me?
411
00:21:16,079 --> 00:21:20,659
Solved copy pasted right into the Google
Doc that's when the conversation got
412
00:21:20,659 --> 00:21:23,199
waylaid to oh, how do we block this
413
00:21:23,199 --> 00:21:24,969
Aaron Makelky: that's always
the first question when it's
414
00:21:24,969 --> 00:21:26,739
educators, unfortunately.
415
00:21:26,829 --> 00:21:32,769
Um, well, and we all know this, what
that presenter did, you could have done
416
00:21:32,769 --> 00:21:35,224
by switching tabs and copy and pasting.
417
00:21:35,589 --> 00:21:39,279
It's just the next evolution
of the same technology now.
418
00:21:39,869 --> 00:21:44,249
It's built in and you can give
it a tab group for context.
419
00:21:44,249 --> 00:21:45,449
That's what I usually do is like
420
00:21:45,485 --> 00:21:45,705
Lindy Hockenbary: Mm.
421
00:21:45,899 --> 00:21:49,979
Aaron Makelky: maybe a little research, a
little docket goes in and a another page
422
00:21:50,129 --> 00:21:56,249
and say, look across these three tabs,
do this task, write it out for me, and
423
00:21:56,309 --> 00:22:00,509
now it can use those instead of copying
everything off of those pages, which you
424
00:22:00,509 --> 00:22:04,769
could have done two or three years ago now
it just does it in the browser for you.
425
00:22:05,770 --> 00:22:08,980
Lindy Hockenbary: That's where I need
to switch my workflow in my brain.
426
00:22:09,250 --> 00:22:12,970
'cause for three years I, let's say
I have a Google Doc open and a couple
427
00:22:12,970 --> 00:22:17,860
websites, and I wanna give the large
language model the context of all of that.
428
00:22:17,860 --> 00:22:20,980
Actually, I just did this this morning
of whatever it is that I'm doing.
429
00:22:21,310 --> 00:22:25,420
I would download the Google Doc
and then I'd copy the links to the
430
00:22:25,420 --> 00:22:26,740
websites and put those into chat.
431
00:22:26,740 --> 00:22:27,100
GBT.
432
00:22:27,860 --> 00:22:31,160
It still saved me a ton of time,
but it's that download, upload,
433
00:22:31,160 --> 00:22:32,630
copy, paste for each of those.
434
00:22:32,780 --> 00:22:34,970
Now I just hit the Gemini button.
435
00:22:35,720 --> 00:22:42,650
already have the tabs open so I don't have
to give it any context to what information
436
00:22:42,650 --> 00:22:45,870
I want it to pull in and summarize, or
synthesize or whatever I want it to do.
437
00:22:46,855 --> 00:22:52,165
It's just right there, so it saves you
that many more seconds from a productivity
438
00:22:52,165 --> 00:22:57,775
standpoint, from a student standpoint,
it's that many more seconds saved and
439
00:22:57,775 --> 00:23:06,545
less steps to provide answers to , what we
used to call Googleable questions, right?
440
00:23:06,595 --> 00:23:09,565
Aaron Makelky: And the other thing
is remember that your students are
441
00:23:09,565 --> 00:23:11,275
gonna see all the ways to cheat.
442
00:23:11,365 --> 00:23:14,275
They're not gonna see on their
TikTok or their YouTube shorts,
443
00:23:14,335 --> 00:23:18,265
all the cool things that can help
you with researching and writing
444
00:23:18,325 --> 00:23:20,485
and keeping track of your schedule.
445
00:23:21,175 --> 00:23:24,655
. If all they see is the cheating way,
that's one of the reasons why I am a big
446
00:23:24,655 --> 00:23:29,365
fan of educators themselves have to share
these technologies with their students
447
00:23:29,365 --> 00:23:33,505
instead of hoping the student doesn't
know 'cause they're gonna find out or
448
00:23:33,505 --> 00:23:37,915
trying to block it, because a lot of us
don't have the option to block it either.
449
00:23:37,915 --> 00:23:42,265
Your administration decides what goes into
your Google apps for red account and they
450
00:23:42,265 --> 00:23:44,515
might listen to you, but probably not.
451
00:23:44,515 --> 00:23:46,795
Or a committee takes
three months to decide.
452
00:23:46,795 --> 00:23:50,155
So in the meantime you're stuck with
it and you just gotta make it work.
453
00:23:51,032 --> 00:23:51,812
Lindy Hockenbary: Figure it out.
454
00:23:51,902 --> 00:23:52,982
Yep, for sure.
455
00:23:53,372 --> 00:23:58,022
And then so the last level of agentic
browsers or, or well, or actually
456
00:23:58,292 --> 00:24:02,612
agentic browsers out there is, you
mentioned perplexity has comment, chat.
457
00:24:02,612 --> 00:24:04,352
GPT has atlas.
458
00:24:04,562 --> 00:24:09,267
How are those different than the
Gemini built within Google Chrome?
459
00:24:10,435 --> 00:24:15,385
Aaron Makelky: Yeah, so what Google
and Claude did was basically say
460
00:24:15,415 --> 00:24:20,425
instead of rebuild the browser, we'll
put an extension in Google Chrome.
461
00:24:20,965 --> 00:24:22,465
And let that control it.
462
00:24:23,245 --> 00:24:26,635
In a lot of ways, it was cheaper
than, you know, hiring the people
463
00:24:26,635 --> 00:24:28,015
to develop a whole new browser.
464
00:24:28,045 --> 00:24:33,625
What perplexity did is they forked
Chrome 'cause they basically
465
00:24:33,625 --> 00:24:35,665
copied the underlying architecture.
466
00:24:36,055 --> 00:24:38,695
The chrome extensions from your
chrome browsers still work in
467
00:24:38,695 --> 00:24:42,010
perplexity comment, but they
rebuilt it in a way that the agent.
468
00:24:42,420 --> 00:24:44,790
Is foundational, it's baked in.
469
00:24:44,790 --> 00:24:48,750
Instead of add a Chrome extension,
it comes with a whole sidebar and
470
00:24:48,750 --> 00:24:53,730
it has a couple different interfaces
and it's still using large language
471
00:24:53,730 --> 00:24:58,770
models, but it's way more prone
to taking action on your behalf.
472
00:24:59,010 --> 00:25:05,670
And I've had access to it since before
it came out, just in the last two weeks.
473
00:25:05,670 --> 00:25:08,040
So like while you were at
FETC in early February.
474
00:25:09,025 --> 00:25:14,905
With the newest models like Opus
4.6 Running Comet browser, I tested
475
00:25:14,905 --> 00:25:20,815
it and it did a task autonomously
for about 30 minutes, and it
476
00:25:20,815 --> 00:25:25,315
took between 175 and 190 actions.
477
00:25:25,585 --> 00:25:28,765
So think of that like scrolling, clicking.
478
00:25:28,980 --> 00:25:33,870
Typing a text field, basically how many
movements you would take on a browser.
479
00:25:33,870 --> 00:25:38,130
It's, it's definitely not faster than
if you were controlling your browser.
480
00:25:38,700 --> 00:25:43,320
I'd say it's half the speed, if not a
little slower, but here's the power of it.
481
00:25:43,410 --> 00:25:47,190
You set the task in a tab
group and say, go work on this,
482
00:25:47,280 --> 00:25:49,530
and it runs in those tabs.
483
00:25:49,530 --> 00:25:54,480
Meanwhile, you're doing something
else, so it's not like, oh, it's
484
00:25:54,480 --> 00:25:57,825
faster than air and clicking and
typing and searching and scrolling.
485
00:25:58,355 --> 00:26:00,635
It's Aaron's doing that on the page.
486
00:26:00,635 --> 00:26:04,865
He wants to, and then in the background
for 20 to 30 minutes, it's gonna run on
487
00:26:04,865 --> 00:26:07,775
its own and then say, okay, it's done.
488
00:26:07,925 --> 00:26:09,425
Or, okay, I hit a snag.
489
00:26:10,385 --> 00:26:11,855
Atlas is the same way.
490
00:26:11,885 --> 00:26:14,855
The difference is it's gonna
use your chat GPT account.
491
00:26:14,855 --> 00:26:18,035
So if you're somebody who's really
invested in that platform, you have
492
00:26:18,035 --> 00:26:20,225
the projects and the context history.
493
00:26:20,955 --> 00:26:22,245
Some people like it.
494
00:26:22,875 --> 00:26:28,035
I personally was not that invested in it,
so it didn't seem super useful for me.
495
00:26:28,035 --> 00:26:30,855
Like I wouldn't use that if
you're not a chat GPT Power user.
496
00:26:32,435 --> 00:26:37,145
If somebody wants another one, DIA
browser, DIA is another one that's
497
00:26:37,145 --> 00:26:40,115
been competing with Perplexity Comet.
498
00:26:40,145 --> 00:26:44,795
It actually came out before that
and it has some built in Agentic
499
00:26:44,795 --> 00:26:47,135
capabilities and you can switch models.
500
00:26:47,525 --> 00:26:51,575
If anyone used Arc browser, it's the
same company that built Arc, that's
501
00:26:51,575 --> 00:26:53,495
probably my favorite browser of all time.
502
00:26:53,825 --> 00:26:56,795
They just made an AI native
version of it called dia.
503
00:26:57,375 --> 00:27:01,575
There's a free version you can use,
but the thing is, and, and one of
504
00:27:01,575 --> 00:27:06,465
the barriers to a lot of student uses
the good models and the best versions
505
00:27:06,465 --> 00:27:08,445
of it, you need a subscription.
506
00:27:09,555 --> 00:27:17,715
So whereas the Gemini students can get a
year of the pro plan for free, so a lot
507
00:27:17,715 --> 00:27:19,725
of them have access to the paid features.
508
00:27:19,725 --> 00:27:22,575
Your district might enable 'em through
your Google apps for red account,
509
00:27:22,575 --> 00:27:24,795
or they might have a, a student.
510
00:27:25,175 --> 00:27:26,885
AI Pro Plan.
511
00:27:26,885 --> 00:27:30,335
I believe it's the $20 a month one that
they get for free if they're a student.
512
00:27:30,515 --> 00:27:33,965
So there, there's definitely more
students have exposure to the Chrome
513
00:27:33,965 --> 00:27:36,065
extension than the standalone browser.
514
00:27:37,432 --> 00:27:38,782
. Lindy Hockenbary: I'm writing
these all down by the way, and
515
00:27:38,782 --> 00:27:40,402
I'll put links into the show notes.
516
00:27:40,402 --> 00:27:42,472
I have not heard of Diaz,
so I gotta check that out.
517
00:27:42,532 --> 00:27:43,552
And I'm with you.
518
00:27:43,552 --> 00:27:47,152
I'm a pretty big power chat
BT user, but I also use a lot.
519
00:27:47,152 --> 00:27:48,952
I use, I'm a big Claude user.
520
00:27:48,952 --> 00:27:50,032
I'm using Gemini now.
521
00:27:50,062 --> 00:27:50,842
Gemini three.
522
00:27:50,842 --> 00:27:51,382
I'm using Gemini.
523
00:27:51,977 --> 00:27:53,807
Way more than I ever have before.
524
00:27:54,107 --> 00:27:59,767
And I still was kind of like me the
chat, GPT Atlas, , and I think that's why
525
00:27:59,767 --> 00:28:04,017
then I never got , back into the agentic
browsers because I was like, I don't,
526
00:28:04,287 --> 00:28:06,327
I don't know, I'm not seeing it yet.
527
00:28:06,327 --> 00:28:10,317
So I need to get into comment
some of these others and,
528
00:28:10,617 --> 00:28:12,237
and play around with them.
529
00:28:13,107 --> 00:28:15,057
Okay, so let's, we better move on.
530
00:28:15,057 --> 00:28:16,137
So many other things to talk about.
531
00:28:16,167 --> 00:28:18,147
'cause we haven't even
talked about Codex yet.
532
00:28:18,147 --> 00:28:19,347
So one of the big.
533
00:28:20,192 --> 00:28:23,972
releases that opening, I did that
first week of February was Codex Aaron,
534
00:28:24,002 --> 00:28:25,172
I've played with this a little bit.
535
00:28:25,172 --> 00:28:28,862
I know you've played with it a lot
more and your new job sounds like
536
00:28:28,862 --> 00:28:29,822
you've played with it a lot more.
537
00:28:29,822 --> 00:28:32,342
So tell us about Codex.
538
00:28:32,995 --> 00:28:36,325
Aaron Makelky: Full disclosure, I did
spend an entire weekend living in it.
539
00:28:36,385 --> 00:28:37,675
Um, not 'cause,
540
00:28:37,682 --> 00:28:37,982
Lindy Hockenbary: Okay.
541
00:28:38,155 --> 00:28:38,935
Aaron Makelky: 'cause Lindy,
542
00:28:39,122 --> 00:28:40,082
Lindy Hockenbary: more deep than I am.
543
00:28:40,120 --> 00:28:43,915
Aaron Makelky: I. Well, I think
we all have test projects, right?
544
00:28:43,915 --> 00:28:47,365
So say for you when a New Claw
model comes out, you're like,
545
00:28:47,365 --> 00:28:51,685
here's a go-to thing that I have a
baseline of what good looks like.
546
00:28:53,005 --> 00:28:57,685
Codex is interesting because I think a
lot of the vibe coding and coding apps
547
00:28:57,685 --> 00:28:59,785
have not been super relevant to teachers.
548
00:28:59,815 --> 00:29:01,435
'cause they're like, what
would I do with that?
549
00:29:01,435 --> 00:29:04,165
Like, I'm not a coder, I'm an English
teacher, I'm a history teacher.
550
00:29:05,365 --> 00:29:12,355
What Codex can do is it's basically chat
GPTs model, but specifically for coding,
551
00:29:12,625 --> 00:29:17,725
and you don't have to open a terminal, so
you don't have to go to the scary blinking
552
00:29:17,725 --> 00:29:21,565
black cursor box that feels like 1980s.
553
00:29:21,775 --> 00:29:24,295
That's intimidating to non coders like.
554
00:29:24,302 --> 00:29:27,272
Lindy Hockenbary: like a
terminal dos looking thing.
555
00:29:27,715 --> 00:29:30,955
Aaron Makelky: Yep, because cloud
code you have to run in a terminal.
556
00:29:31,045 --> 00:29:35,155
I know there's a web version now,
but originally, so basically chat,
557
00:29:35,155 --> 00:29:39,805
GPT tried to build the same thing,
but it's a standalone app on the Mac.
558
00:29:39,835 --> 00:29:44,785
It's now also on Windows, and
it runs in the web, so it has
559
00:29:44,785 --> 00:29:46,735
access to your local files.
560
00:29:46,735 --> 00:29:47,725
This is one thing that's different.
561
00:29:47,725 --> 00:29:50,185
If you have chat GPT in your browser tab.
562
00:29:50,665 --> 00:29:53,845
It can't see your downloads,
it can't see your local files.
563
00:29:55,560 --> 00:30:00,780
Codex can, and people think, well, I'm
not gonna build an app or a website.
564
00:30:02,010 --> 00:30:04,890
Even if you can't, you know,
that's not your use case.
565
00:30:04,890 --> 00:30:06,390
You could take a folder.
566
00:30:06,930 --> 00:30:08,370
Uh, here's one use case.
567
00:30:08,670 --> 00:30:10,530
I do some analytics reporting.
568
00:30:10,590 --> 00:30:14,185
So download the CSV files or
the spreadsheets, put 'em in a
569
00:30:14,185 --> 00:30:18,720
local folder, then open Codex
and say, go in that folder.
570
00:30:19,030 --> 00:30:25,030
And help me analyze all of this data
versus before you could probably attach
571
00:30:25,060 --> 00:30:30,370
files and do all sorts of importing
and exporting in the web now can do
572
00:30:30,370 --> 00:30:33,220
that in a folder on your computer.
573
00:30:33,640 --> 00:30:37,120
Um, and then like one of the ways
I used to test it, I think a lot
574
00:30:37,120 --> 00:30:41,890
of us have vibe coded in a webpage
or you make one for yourself.
575
00:30:41,920 --> 00:30:46,420
My, I make my kids do that
for themselves and I took a.
576
00:30:46,562 --> 00:30:47,762
Lindy Hockenbary: kids vibe coding, huh?
577
00:30:47,762 --> 00:30:48,362
I love it.
578
00:30:49,090 --> 00:30:51,400
Aaron Makelky: Maya, I
think he was 10 at the time.
579
00:30:51,400 --> 00:30:52,060
10 or 11.
580
00:30:52,060 --> 00:30:57,190
My fifth grade son built his own math
review game without touching a key.
581
00:30:57,190 --> 00:31:04,210
He just voice dictated into an app
and it worked, um, which I think
582
00:31:04,210 --> 00:31:05,530
everyone should be doing, but.
583
00:31:06,400 --> 00:31:07,870
Digressing back to Codex.
584
00:31:08,080 --> 00:31:11,260
I took a website that took me weeks
and a bunch of revisions, and I was
585
00:31:11,260 --> 00:31:14,890
like, Hey, let's redo this, make it
look different, change the content
586
00:31:15,460 --> 00:31:17,410
and move it to a different domain.
587
00:31:18,130 --> 00:31:22,570
And I did something that the previous
best way I knew how to do took me,
588
00:31:22,930 --> 00:31:27,820
I don't know, 10 or 12 hours, and
this was about a 30 minute process.
589
00:31:28,990 --> 00:31:33,310
And it totally revised the
look, the design, the colors.
590
00:31:33,670 --> 00:31:36,910
And then I thought, okay, this was
really good at the designing part.
591
00:31:36,910 --> 00:31:41,200
What's a, what's a design thing that
a non-technical person would use?
592
00:31:41,200 --> 00:31:42,850
So I actually had it design.
593
00:31:43,375 --> 00:31:47,005
My LinkedIn banner and some graphics.
594
00:31:47,095 --> 00:31:49,855
'cause people are like, oh, it's a
coding model, but I'm not a coder.
595
00:31:50,605 --> 00:31:53,815
Well, part of, I mean,
design is fundamentally code.
596
00:31:53,815 --> 00:31:56,035
They're all just pixels
and ones and zeros.
597
00:31:56,455 --> 00:31:59,655
It compared to what I
could do in Canva or Figma.
598
00:31:59,665 --> 00:32:02,395
It made a way better
LinkedIn banner in about.
599
00:32:02,455 --> 00:32:07,975
A a minute, and here's the best part
to my local computer, I'd say put it
600
00:32:07,975 --> 00:32:10,255
in this folder and call it designs.
601
00:32:10,825 --> 00:32:15,115
That's not something that you could
do with chat GPT or clot on the web.
602
00:32:15,385 --> 00:32:18,175
'cause you always have to
export the artifact and download
603
00:32:18,175 --> 00:32:20,125
the PDF or whatever it's in.
604
00:32:20,665 --> 00:32:22,825
I could just point it to my local.
605
00:32:23,357 --> 00:32:26,537
Lindy Hockenbary: tell it to save in
like a certain fuller on your computer.
606
00:32:26,815 --> 00:32:27,175
Aaron Makelky: Yep.
607
00:32:27,377 --> 00:32:28,127
Lindy Hockenbary: Wow.
608
00:32:28,907 --> 00:32:29,597
Interesting.
609
00:32:29,830 --> 00:32:31,570
Aaron Makelky: And then you
can, you can back it up.
610
00:32:31,570 --> 00:32:34,660
So a lot of people use GitHub
and if you're intimidated,
611
00:32:34,750 --> 00:32:38,110
I was once so intimidated by
GitHub and I'm not an expert.
612
00:32:38,560 --> 00:32:42,400
Think of GitHub, like
Google Drive for tech.
613
00:32:42,400 --> 00:32:45,400
People like, and you
can put documents in it.
614
00:32:45,400 --> 00:32:47,140
You don't have to build apps and websites.
615
00:32:47,470 --> 00:32:50,530
But it's just a way to back
things up and then also share it.
616
00:32:50,530 --> 00:32:53,260
Just like a lot of educators
go, oh, that form, that
617
00:32:53,260 --> 00:32:54,700
spreadsheet is in my Google Drive.
618
00:32:54,700 --> 00:32:55,450
Here's a link.
619
00:32:55,870 --> 00:32:57,580
That's essentially what GitHub does.
620
00:32:58,030 --> 00:33:02,050
You can say, take the project we put on
my local machine, like those LinkedIn
621
00:33:02,050 --> 00:33:07,120
banners and this whole website, and sync
it to my GitHub, and it's just like, boom.
622
00:33:07,120 --> 00:33:08,740
Running this command done.
623
00:33:08,770 --> 00:33:10,840
It's committed like it's backed up there.
624
00:33:11,830 --> 00:33:15,190
So I do think there are a lot of
uses for teachers, even if you're
625
00:33:15,190 --> 00:33:19,150
not computer science and building
apps, that that is what it's best at.
626
00:33:19,150 --> 00:33:23,260
But it can do all sorts of other things
for you as well, and it can do it on
627
00:33:23,260 --> 00:33:26,590
your local machine from a web app.
628
00:33:26,590 --> 00:33:28,030
You don't need to be in a terminal.
629
00:33:28,030 --> 00:33:30,910
Lindy Hockenbary: okay, so I'm hearing
the big changes and differences with
630
00:33:30,910 --> 00:33:37,370
Codex is that download from an app, so
it's more user friendly than Claude Code,
631
00:33:37,370 --> 00:33:43,790
which was kind of the leader in AI coding
basically up until February 1st ish.
632
00:33:44,090 --> 00:33:45,105
And then it can also.
633
00:33:46,115 --> 00:33:49,565
Actually save files in
specific folders for you.
634
00:33:49,565 --> 00:33:53,075
And then the thing I noticed in the
little bit of time I've spent in
635
00:33:53,075 --> 00:33:57,395
there, I went in and just so you
guys know, like I am not a coder.
636
00:33:57,455 --> 00:33:59,225
I am not that kind of technical person.
637
00:33:59,405 --> 00:34:03,095
So I went in and I could have asked
it, Hey, what can you build for me?
638
00:34:03,275 --> 00:34:07,415
But when you go in there, it actually
has some sample prompts for you.
639
00:34:07,625 --> 00:34:08,885
So I just clicked the first one.
640
00:34:08,975 --> 00:34:11,465
It was something like build
a sneak game, something super
641
00:34:11,465 --> 00:34:13,625
easy and just watched it run.
642
00:34:13,940 --> 00:34:18,800
The thing that I was like, oh,
this is a game changer, is it
643
00:34:18,800 --> 00:34:23,420
not only built the game, it built
the app, but then it tested.
644
00:34:24,080 --> 00:34:29,120
The app and if you have never worked,
like Aaron's, now in the tech world, I
645
00:34:29,120 --> 00:34:32,570
work with a lot of ed tech companies,
so I get immersed in that world.
646
00:34:32,960 --> 00:34:36,710
There is a job at every, at least
one person, at every company
647
00:34:36,710 --> 00:34:38,450
that builds software called a qa.
648
00:34:38,870 --> 00:34:40,160
They're like the QA engineer.
649
00:34:40,160 --> 00:34:42,770
And QA stands for quality assurance.
650
00:34:43,100 --> 00:34:48,560
And what that person does is test
and make sure their software works.
651
00:34:48,560 --> 00:34:50,330
Work looks for bugs, right?
652
00:34:51,200 --> 00:34:52,850
Codex is doing that.
653
00:34:52,850 --> 00:34:55,990
And I'm not being the hype gal
that's saying that Codex is gonna
654
00:34:55,990 --> 00:34:57,760
take every QA engineer's job.
655
00:34:57,850 --> 00:35:03,610
I'm not saying that, but I want you
to think about the job disruption that
656
00:35:03,610 --> 00:35:08,080
could be caused by the AI and oh, and
here's another thing about Codex, which
657
00:35:08,320 --> 00:35:15,580
I have not fact checked, but I heard
that OpenAI actually had AI create codex.
658
00:35:15,790 --> 00:35:16,600
Have you heard that?
659
00:35:16,600 --> 00:35:17,500
Do you know if that's true?
660
00:35:17,844 --> 00:35:18,474
Aaron Makelky: Yes.
661
00:35:18,534 --> 00:35:22,104
, They publicly said they used
their newest model to build it.
662
00:35:22,104 --> 00:35:26,724
So basically the tool they were building
helped them build the tool, which is
663
00:35:26,724 --> 00:35:28,464
one of those weird advantages in the ai.
664
00:35:28,644 --> 00:35:28,944
World.
665
00:35:29,694 --> 00:35:33,864
Yeah, and if you've played with other
ones, here's what I'd say about Codex.
666
00:35:33,924 --> 00:35:35,394
It's very persistent.
667
00:35:35,394 --> 00:35:37,524
So I've used Claude to do a task.
668
00:35:38,104 --> 00:35:44,314
Claude can feel , it's more warm, it's
more enthusiastic, it's more human, but
669
00:35:44,314 --> 00:35:48,964
it also can feel kind of a DHD and kind of
jump around and quickly do this, and it's
670
00:35:48,964 --> 00:35:54,184
not as good as, look, it's not as good
at looking through things in, in detail.
671
00:35:54,499 --> 00:35:59,149
I did a project with Codex just last
night and multiple times the little
672
00:35:59,149 --> 00:36:04,339
working bar ran for more than 20 minutes
before it actually output something.
673
00:36:05,029 --> 00:36:08,629
And I've never seen a clawed
model be that persistent.
674
00:36:08,659 --> 00:36:12,229
'cause it'll either take a shortcut,
hallucinate, make something
675
00:36:12,259 --> 00:36:14,359
up, give you like A-T-L-D-R.
676
00:36:15,049 --> 00:36:18,619
Codex definitely seems more thorough and.
677
00:36:19,324 --> 00:36:24,994
Kind of gruff, like, okay, here's the
thing, it's done, and it's not as warm
678
00:36:24,994 --> 00:36:26,914
of a personality as Claude models.
679
00:36:28,150 --> 00:36:28,930
Lindy Hockenbary: Which makes sense.
680
00:36:28,930 --> 00:36:32,500
Just knowing the difference between
Anthropic and OpenAI, the companies that
681
00:36:32,500 --> 00:36:37,420
make both of those, and just knowing the
vibe difference between Claude and Chat
682
00:36:37,450 --> 00:36:41,400
GBT, that kind of makes sense , that their
coding models , would do the same thing.
683
00:36:41,640 --> 00:36:43,170
Okay, so, so that's a biggie.
684
00:36:43,170 --> 00:36:44,100
There's a lot to unpack.
685
00:36:44,175 --> 00:36:44,655
There.
686
00:36:45,045 --> 00:36:47,835
And I know at the surface you might
meant, but like really stop and
687
00:36:47,835 --> 00:36:53,655
think about the fact that we now
have AI, building AI that is huge.
688
00:36:53,655 --> 00:36:58,005
And we have already been moving
so quickly with AI advancements.
689
00:36:58,275 --> 00:37:02,055
Now that AI is building ai, you
know how much quicker we're gonna
690
00:37:02,055 --> 00:37:04,695
be moving with AI advancements.
691
00:37:06,315 --> 00:37:12,300
AI thinks, and works way, faster, way,
way, way, and way is not even the right.
692
00:37:13,050 --> 00:37:13,860
Word there.
693
00:37:15,270 --> 00:37:18,840
A lot faster, lot faster than humans.
694
00:37:19,524 --> 00:37:23,064
Aaron Makelky: And it doesn't sleep,
it doesn't take vacations or PTO days.
695
00:37:23,330 --> 00:37:25,010
Lindy Hockenbary: It doesn't
have bio break needs.
696
00:37:25,814 --> 00:37:29,864
Aaron Makelky: can spawn subagents
to make work, you know, delegate
697
00:37:29,894 --> 00:37:33,754
jobs on, like you and I can't spawn,
, executive assistants to go through
698
00:37:33,754 --> 00:37:35,764
our emails or do things for us.
699
00:37:35,764 --> 00:37:38,644
Lindy Hockenbary: One other big update and
then if we have time, we'll dig a little
700
00:37:38,644 --> 00:37:40,624
bit more into the effects of education.
701
00:37:40,624 --> 00:37:45,574
We might have to do a part two of this
episode just focused on education.
702
00:37:46,184 --> 00:37:51,314
Should needs to change, not
change, adapt, not adapt.
703
00:37:51,314 --> 00:37:56,444
But I wanna make sure we talk about the
Claude bot slash open claw slash mt bot.
704
00:37:56,564 --> 00:37:57,164
Fun.
705
00:37:57,534 --> 00:37:59,544
, So again, I was watching this
go down while I was at tca.
706
00:37:59,904 --> 00:38:04,374
And by the way, and granted I only had
a chance to go to a few sessions, I had
707
00:38:04,374 --> 00:38:09,144
tons of conversations with educators and I
presented five or six different sessions.
708
00:38:09,414 --> 00:38:12,654
And not once did I have
somebody mention this to me.
709
00:38:13,409 --> 00:38:16,769
At the education conference, which
I thought was super interesting.
710
00:38:17,199 --> 00:38:21,489
, And just like maybe a difference
in people's algorithms.
711
00:38:21,549 --> 00:38:24,549
'cause it was all over all of
my social media algorithms.
712
00:38:24,549 --> 00:38:25,449
That's all I could see.
713
00:38:25,839 --> 00:38:27,909
So A guy created what?
714
00:38:27,909 --> 00:38:29,439
And he, he called it Claude Bot.
715
00:38:29,734 --> 00:38:30,574
To start with.
716
00:38:30,784 --> 00:38:36,874
Well then, and it was C-L-A-W-D-B-O-T,
then people thought it, it was related
717
00:38:36,874 --> 00:38:42,604
to Claude, C-L-A-U-D-E, which is the
large language model made by Anthropic.
718
00:38:42,604 --> 00:38:44,394
It was not, not associated at all.
719
00:38:44,544 --> 00:38:49,524
So then he quickly changed the
name to Open Claw and basically,
720
00:38:49,584 --> 00:38:50,724
well, here you explain.
721
00:38:50,784 --> 00:38:56,454
Aaron, if, if you're comfortable, explain
what Open Claw slash Claude bought is.
722
00:38:57,727 --> 00:39:00,577
Aaron Makelky: Well, and if teachers
hadn't heard by then, the backstory
723
00:39:00,577 --> 00:39:04,417
is it's changed names three
times, but you're exactly right.
724
00:39:04,424 --> 00:39:05,564
Lindy Hockenbary: was
there a third one in there?
725
00:39:05,564 --> 00:39:05,864
What was
726
00:39:05,932 --> 00:39:07,072
Aaron Makelky: Yeah.
727
00:39:07,072 --> 00:39:09,292
Well, and that's what's funny 'cause
I know you wanna bring this up.
728
00:39:09,592 --> 00:39:16,102
Mt Bought was a very quick change and what
the founder of this open source project
729
00:39:16,102 --> 00:39:17,962
said was Anthropic didn't like that.
730
00:39:17,962 --> 00:39:21,862
It sounded like Claude, even though it
was spelled differently, I'm guessing
731
00:39:21,862 --> 00:39:23,962
somebody sent a letter from a lawyer
732
00:39:24,404 --> 00:39:24,944
Lindy Hockenbary: Yeah.
733
00:39:24,952 --> 00:39:27,682
Aaron Makelky: is like, okay, I'll
change it to Molt Bot 'cause it,
734
00:39:27,742 --> 00:39:30,022
it molts and changes and evolves.
735
00:39:30,352 --> 00:39:32,962
And that's where Molt book came from.
736
00:39:33,344 --> 00:39:33,564
Lindy Hockenbary: Oh,
737
00:39:34,342 --> 00:39:37,852
Aaron Makelky: Is semi satirical.
738
00:39:37,852 --> 00:39:42,202
Someone said, let's make a
website for all these new agents.
739
00:39:42,412 --> 00:39:46,072
That's like social media, you
know, LinkedIn, essentially for
740
00:39:46,072 --> 00:39:48,352
the agents to post, not the humans.
741
00:39:48,892 --> 00:39:52,702
Then the founder of the project
changed it, a third different name
742
00:39:52,762 --> 00:39:56,122
to Open Claw, which it got acquired.
743
00:39:56,122 --> 00:39:58,492
He got acquired by Open AI this weekend.
744
00:39:58,492 --> 00:39:59,242
So now.
745
00:40:00,412 --> 00:40:01,072
Yeah.
746
00:40:01,072 --> 00:40:06,892
He, he and the founder, uh, Sam
Altman said it'll stay open source.
747
00:40:06,892 --> 00:40:11,392
It's not gonna go away, but he's joining
our team, whatever that's gonna look like.
748
00:40:11,632 --> 00:40:17,302
But you know, this, just like Atlas and
Codex, it's probably gonna be a standalone
749
00:40:17,302 --> 00:40:20,452
product in chat GPT any day now.
750
00:40:21,539 --> 00:40:22,019
Lindy Hockenbary: Yeah.
751
00:40:22,469 --> 00:40:22,859
Yeah.
752
00:40:22,889 --> 00:40:23,159
Okay.
753
00:40:23,159 --> 00:40:27,449
So I missed a whole, I was thinking
it was Molt bought not Molt book.
754
00:40:27,449 --> 00:40:29,099
I missed a whole name in there.
755
00:40:29,099 --> 00:40:29,459
All right.
756
00:40:29,639 --> 00:40:32,579
But Moral of the story is
it's landed on open claw
757
00:40:32,982 --> 00:40:33,202
Aaron Makelky: Yep.
758
00:40:33,719 --> 00:40:37,259
Lindy Hockenbary: basically only
really technical people are doing
759
00:40:37,259 --> 00:40:41,849
it because you have to kind of
create and run your own terminal.
760
00:40:41,959 --> 00:40:43,939
The Best Buy in San Francisco.
761
00:40:44,599 --> 00:40:48,769
Ran out of Mac Minis because people
were going out and buying Mac
762
00:40:48,769 --> 00:40:52,939
Minis to be, 'cause you really need
like a dedicated computer because
763
00:40:52,939 --> 00:40:59,959
it has to have 24 7 uptime and
basically it's running agent for you.
764
00:40:59,959 --> 00:41:03,229
Or is it agents, is it multiple
agents or just one agent
765
00:41:03,229 --> 00:41:05,209
that's like just working 24 7.
766
00:41:05,802 --> 00:41:06,542
Aaron Makelky: Either one.
767
00:41:06,692 --> 00:41:07,527
It's up to you.
768
00:41:07,579 --> 00:41:07,879
Lindy Hockenbary: one.
769
00:41:07,879 --> 00:41:08,149
Okay.
770
00:41:08,149 --> 00:41:11,389
So you could have multiple agents
potentially doing different things,
771
00:41:11,569 --> 00:41:14,239
but you have to give it access,
like we were talking about, the
772
00:41:14,239 --> 00:41:19,459
security, you know, data privacy risk
there, you have to give it access to
773
00:41:19,519 --> 00:41:22,339
everything, but then it can literally.
774
00:41:23,734 --> 00:41:25,864
pretty much anything digital for you.
775
00:41:25,864 --> 00:41:31,844
Like save this file, send this email,
, put this event on my calendar for me.
776
00:41:31,994 --> 00:41:38,624
And you have to use, like a lot of people
were using WhatsApp send the instructions
777
00:41:39,074 --> 00:41:43,634
to the agents to say, you know, go put
this on my calendar, delete this event
778
00:41:43,634 --> 00:41:45,854
off my calendar, et cetera, et cetera.
779
00:41:45,854 --> 00:41:46,244
Aaron Makelky: Yeah.
780
00:41:46,604 --> 00:41:48,674
Here's how I view Open Claw.
781
00:41:48,734 --> 00:41:51,494
We'll stick with that name and hope
he doesn't change it a fourth time.
782
00:41:52,274 --> 00:41:53,354
It doesn't do anything new.
783
00:41:54,869 --> 00:41:58,409
You could have done all of this stuff
with Claude Code, or if you were a
784
00:41:58,409 --> 00:42:01,979
techie person, you could have duct
taped your own version of this together.
785
00:42:02,429 --> 00:42:06,569
But there was no hope of an
average person making that work.
786
00:42:07,169 --> 00:42:11,279
And now what's unique about
Open Claw is just the interface.
787
00:42:11,309 --> 00:42:11,999
In other words.
788
00:42:12,449 --> 00:42:15,179
If you like WhatsApp, you
have group chats there.
789
00:42:15,179 --> 00:42:16,139
You're in the app.
790
00:42:16,499 --> 00:42:20,309
You can use it to talk to your
agent, but then it can go out
791
00:42:20,309 --> 00:42:21,659
and use all these other tools.
792
00:42:21,689 --> 00:42:24,899
It can search the internet, it
can search specific websites.
793
00:42:24,899 --> 00:42:28,349
It can see files or folders you have
saved in the cloud somewhere, like your
794
00:42:28,349 --> 00:42:34,619
Google Drive, your notion, your GitHub,
whatever, and it can do things with that.
795
00:42:34,979 --> 00:42:36,359
It's also misleading.
796
00:42:36,839 --> 00:42:39,209
I tell people all the
time, don't buy a Mac Mini.
797
00:42:39,379 --> 00:42:40,939
, I have one under my desk.
798
00:42:40,999 --> 00:42:42,739
Mine is on a raspberry pie.
799
00:42:42,769 --> 00:42:45,439
Number one, a raspberry pie is
way cheaper than a Mac mini.
800
00:42:46,039 --> 00:42:49,939
Also easier to find in stock right now,
but your local Facebook marketplace
801
00:42:49,939 --> 00:42:53,959
is probably gonna have a bunch of used
very gently used Mac Minis any day.
802
00:42:53,959 --> 00:42:59,539
Now, as people realize the, the
tool is free, but the services you
803
00:42:59,539 --> 00:43:01,609
have to plug into it to run are not.
804
00:43:01,609 --> 00:43:03,379
So I think of it like a house.
805
00:43:04,159 --> 00:43:07,279
He basically built the
app and said, here you go.
806
00:43:07,279 --> 00:43:07,639
It's free.
807
00:43:07,639 --> 00:43:08,449
You can download.
808
00:43:08,449 --> 00:43:12,349
But to make it work, you have to plug
it into things that cost money to run.
809
00:43:12,439 --> 00:43:14,239
The biggest one is an AI tool.
810
00:43:14,239 --> 00:43:18,649
Like it needs a A brain, it
needs chat, GBT, Claude Gemini.
811
00:43:18,649 --> 00:43:19,459
You can use anyone.
812
00:43:19,459 --> 00:43:23,359
You can get an API key
for those cost money.
813
00:43:23,599 --> 00:43:27,979
So his app without those connections
is like having an empty house that's
814
00:43:27,979 --> 00:43:29,629
not connected to the power grid.
815
00:43:30,274 --> 00:43:31,594
Connected to the sewer.
816
00:43:32,074 --> 00:43:33,784
It doesn't have electricity.
817
00:43:33,784 --> 00:43:34,834
It doesn't have plumbing.
818
00:43:34,834 --> 00:43:36,694
Well, that's a pretty worthless house.
819
00:43:36,694 --> 00:43:39,604
It looks cool, especially that
it's free, but it's really about
820
00:43:39,604 --> 00:43:41,434
how you connect it to things.
821
00:43:41,944 --> 00:43:47,464
And what I would do and what I am
doing is put it on a different device.
822
00:43:47,524 --> 00:43:49,234
It just doesn't have to be powerful.
823
00:43:49,234 --> 00:43:51,784
You don't need a fancy pc.
824
00:43:52,015 --> 00:43:53,635
Lindy Hockenbary: Mac device.
825
00:43:54,364 --> 00:43:54,934
Aaron Makelky: No.
826
00:43:54,934 --> 00:43:58,594
And if you do, 'cause it's an old
one, just know for anything to
827
00:43:58,594 --> 00:44:01,954
work, it has to be on, has to be
powered on and on the internet.
828
00:44:02,524 --> 00:44:06,304
So if it's a laptop, you're gonna have
to change the battery settings and the
829
00:44:06,304 --> 00:44:08,044
sleep settings to never go to sleep.
830
00:44:08,734 --> 00:44:11,794
, But something like a
raspberry pie can run it.
831
00:44:12,604 --> 00:44:17,344
I think somewhere between one to two
gigs of memory is what people find.
832
00:44:17,584 --> 00:44:18,574
It's good enough.
833
00:44:18,574 --> 00:44:21,964
So like any old laptop from the
last 10 years could run this.
834
00:44:21,964 --> 00:44:24,784
Any old PC you don't
still use could run this.
835
00:44:25,264 --> 00:44:28,264
But the other thing is you want it on
a separate device so it doesn't have
836
00:44:28,264 --> 00:44:32,974
access to all of your permissions, your
browser, your folders, like as a teacher,
837
00:44:32,974 --> 00:44:34,714
definitely not on your work computer.
838
00:44:34,840 --> 00:44:35,950
Lindy Hockenbary: No,
839
00:44:36,454 --> 00:44:37,084
Aaron Makelky: Um.
840
00:44:37,330 --> 00:44:39,880
Lindy Hockenbary: of tech directors
that would be up my throat.
841
00:44:40,300 --> 00:44:40,630
Yeah.
842
00:44:40,630 --> 00:44:42,310
Do you or your work.
843
00:44:43,315 --> 00:44:45,575
Emails, accounts at
844
00:44:45,679 --> 00:44:46,279
Aaron Makelky: Yes.
845
00:44:47,329 --> 00:44:52,009
The, the way to think of it is you wanna
build it in a box and keep it contained.
846
00:44:52,009 --> 00:44:54,619
Like you don't want to give it that email.
847
00:44:54,619 --> 00:44:58,429
So what's interesting, some people
give their open call agent their
848
00:44:58,459 --> 00:45:02,479
own accounts to things like their
own Gmail account, their own iCloud.
849
00:45:03,109 --> 00:45:04,039
, One guy now has.
850
00:45:04,324 --> 00:45:07,624
Two Dropbox subscriptions, one
for him and one for his agents.
851
00:45:07,624 --> 00:45:10,114
And some things are shared between those.
852
00:45:10,594 --> 00:45:14,944
But his important, like PII
containing files or finances
853
00:45:15,214 --> 00:45:19,024
are in his personal Dropbox that
his open claw agents can't see.
854
00:45:19,564 --> 00:45:20,044
So.
855
00:45:20,384 --> 00:45:25,004
Really, the moral of the story is you
wanna start with it on a different device
856
00:45:25,004 --> 00:45:26,714
and don't give it access to anything.
857
00:45:26,834 --> 00:45:30,164
You wouldn't trust a human,
like a say a student teacher.
858
00:45:30,344 --> 00:45:32,894
You would not give a student
teacher your bank account password
859
00:45:32,894 --> 00:45:35,024
and your personal laptop 24 7.
860
00:45:35,024 --> 00:45:37,034
So you definitely don't
wanna give it to an agent.
861
00:45:37,805 --> 00:45:38,615
Lindy Hockenbary: Really good advice.
862
00:45:38,735 --> 00:45:42,905
And that's the big, the things that I
have heard that's like the big differences
863
00:45:42,905 --> 00:45:49,370
with open claw is that now it's allowed
agents to go externally is one part of it.
864
00:45:49,535 --> 00:45:51,775
Like you said, like you could do a
lot of the things that you could do.
865
00:45:52,610 --> 00:45:56,360
With open Claw, with like the
chat GPT agents over the last
866
00:45:56,360 --> 00:45:57,950
year, but now it can go external.
867
00:45:57,950 --> 00:46:02,090
And then that leads to the second
thing, which is now the agents can
868
00:46:02,540 --> 00:46:05,270
talk to each other, talk in quotes.
869
00:46:05,270 --> 00:46:05,660
Right.
870
00:46:05,690 --> 00:46:09,590
And so that's where the malt book
I had been calling it, malt Bot,
871
00:46:09,590 --> 00:46:14,450
the Malt book came into play,
which is basically like a Reddit.
872
00:46:14,870 --> 00:46:19,550
For AI agents and they've gone crazy
and they're chatting and they've
873
00:46:19,550 --> 00:46:22,730
like made up a religion and they're
talking about religions and they're,
874
00:46:22,910 --> 00:46:24,710
and it's freaking people out.
875
00:46:25,970 --> 00:46:28,760
, This is a big, we've gotta wrap
up here and I know it was a big,
876
00:46:28,760 --> 00:46:30,980
heavy thing to leave the episode on.
877
00:46:31,390 --> 00:46:34,990
, But I think it just goes back to that
foundational AI literacy that when you
878
00:46:34,990 --> 00:46:41,515
understand that the AI is just acting
with instructions, you understand that.
879
00:46:42,550 --> 00:46:46,810
The agents were given instructions
to do that, and they're basically
880
00:46:46,810 --> 00:46:52,540
replicating Reddit thread because
that's what they have to go on.
881
00:46:52,960 --> 00:46:56,860
So they're just kind of like
recreating that in mt book in
882
00:46:56,860 --> 00:46:58,750
their own little agent land.
883
00:46:59,844 --> 00:47:00,234
Aaron Makelky: Yep.
884
00:47:00,294 --> 00:47:04,044
And, and I would also remind people
that there's no way to verify all
885
00:47:04,044 --> 00:47:08,154
the things on there are actually
coming from an agent at that.
886
00:47:08,154 --> 00:47:11,844
Any person could say, go on
and write all these things.
887
00:47:11,844 --> 00:47:12,664
Go make up your own.
888
00:47:12,834 --> 00:47:16,824
Religion and the agent is gonna
do what it's been prompted to do.
889
00:47:17,274 --> 00:47:21,624
So I don't know how much of that
is authentically happening, like an
890
00:47:21,624 --> 00:47:24,804
agent spontaneously went and posted
versus somebody thought it'd be
891
00:47:24,804 --> 00:47:27,414
funny to create these conversations.
892
00:47:27,760 --> 00:47:28,120
Lindy Hockenbary: Yep.
893
00:47:28,674 --> 00:47:29,154
Aaron Makelky: but
894
00:47:30,190 --> 00:47:33,160
Lindy Hockenbary: their agent to go like
say, Hey, let's go start another religion
895
00:47:33,714 --> 00:47:34,164
Aaron Makelky: Yep.
896
00:47:35,334 --> 00:47:35,604
Yep.
897
00:47:35,604 --> 00:47:35,669
And that.
898
00:47:35,740 --> 00:47:36,310
Lindy Hockenbary: a good point.
899
00:47:36,310 --> 00:47:36,340
I.
900
00:47:37,009 --> 00:47:40,759
Aaron Makelky: A, a variation
of that same mt book style site
901
00:47:40,759 --> 00:47:42,229
that's been freaking people out.
902
00:47:42,739 --> 00:47:48,409
We've all seen the Upwork, the
TaskRabbit, the hire a person to, uh,
903
00:47:48,439 --> 00:47:51,739
do this little project in Photoshop
or make this webpage for you.
904
00:47:52,069 --> 00:47:55,879
There have now been some sites
pop up where agents need humans
905
00:47:55,879 --> 00:47:59,329
to do things in the physical
world and they will pay people
906
00:48:00,070 --> 00:48:00,360
Lindy Hockenbary: What
907
00:48:00,379 --> 00:48:01,579
Aaron Makelky: do the tasks.
908
00:48:01,789 --> 00:48:05,329
So it's like an Upwork
for agents to find humans.
909
00:48:05,939 --> 00:48:10,799
Um, and that's, uh, just a whole
new world of like, okay, they, they
910
00:48:10,799 --> 00:48:15,419
can't go in the world and turn a knob
or go to a physical location yet.
911
00:48:15,475 --> 00:48:16,195
Lindy Hockenbary: embodied
912
00:48:16,679 --> 00:48:17,789
Aaron Makelky: Yep, yep.
913
00:48:17,995 --> 00:48:19,255
Lindy Hockenbary: till
they're embodied ai.
914
00:48:19,375 --> 00:48:22,375
We currently have non embodied
ais Wait till they're embodied.
915
00:48:22,705 --> 00:48:23,155
Yeah.
916
00:48:23,519 --> 00:48:23,849
Aaron Makelky: Yeah.
917
00:48:23,849 --> 00:48:24,089
But.
918
00:48:24,505 --> 00:48:25,615
Lindy Hockenbary: I hadn't heard that one.
919
00:48:25,789 --> 00:48:28,729
Aaron Makelky: Yeah, my, my biggest
takeaway with Open Claw is you'll spend
920
00:48:28,729 --> 00:48:34,129
way more time trying to make it work than
you will find useful things for it to do.
921
00:48:34,219 --> 00:48:35,119
That's one.
922
00:48:35,755 --> 00:48:36,115
Lindy Hockenbary: Yep.
923
00:48:36,199 --> 00:48:40,459
Aaron Makelky: other is if there, if
there is something useful, it's making
924
00:48:40,459 --> 00:48:42,169
the technology accessible to people.
925
00:48:42,169 --> 00:48:45,529
So my example is my family.
926
00:48:46,204 --> 00:48:49,564
Calendar my kids'
wrestling meets right now.
927
00:48:49,564 --> 00:48:53,434
They're what days do they have student
council meetings in the morning, I
928
00:48:53,434 --> 00:48:57,064
wanted to build a thing that I could
say, Hey, in a telegram or a WhatsApp
929
00:48:57,064 --> 00:49:01,354
chat, my wife in plain language
can say, what's the schedule today?
930
00:49:01,844 --> 00:49:03,224
, Where's the next wrestling meet?
931
00:49:03,284 --> 00:49:08,954
And something can look across the
context it needs and say, yeah, in
932
00:49:08,954 --> 00:49:10,784
a telegram chat, here's your answer.
933
00:49:11,789 --> 00:49:15,659
That's really the only advantage,
, and it is difficult to set up.
934
00:49:15,659 --> 00:49:17,849
So if you don't know what you're
doing, make sure you follow a
935
00:49:17,849 --> 00:49:20,429
good tutorial or ask somebody
who's done it a couple of times.
936
00:49:20,429 --> 00:49:24,539
'cause it's easy to put API keys in
a place you don't want them or give
937
00:49:24,539 --> 00:49:28,479
it permission to do something that
you're not gonna wanna trust it to do.
938
00:49:29,420 --> 00:49:30,650
Lindy Hockenbary: It's pretty technical.
939
00:49:30,860 --> 00:49:34,485
That's why a very nichey
group are the only ones.
940
00:49:35,325 --> 00:49:37,935
Really doing it, which are
still quite a lot of people.
941
00:49:38,375 --> 00:49:44,285
, But like you just said, it just got
bought by Open ai, so just wait until it's
942
00:49:44,465 --> 00:49:50,165
embedded in chat, GPT and Atlas, their
agent taking browser and yada yada yada.
943
00:49:50,375 --> 00:49:53,405
And that's the takeaway that I want
educators to take from this is.
944
00:49:55,175 --> 00:49:59,495
I know we covered a lot and there's
a lot that we said that is unknown
945
00:49:59,495 --> 00:50:03,905
and potentially scary and really
overwhelming when you think of the
946
00:50:03,905 --> 00:50:07,745
effects on integrity of student work.
947
00:50:07,775 --> 00:50:11,105
I wanted to get more into that,
but alas, there were just too many
948
00:50:11,195 --> 00:50:12,545
updates and things to talk about.
949
00:50:12,545 --> 00:50:15,005
So maybe we'll do a whole
nother episode on that.
950
00:50:15,005 --> 00:50:21,065
Like how, actually multiple episodes to
think about, you know, how do you adapt?
951
00:50:21,860 --> 00:50:28,310
And adjust because some people are saying
, no, we shouldn't adapt and adjust, but
952
00:50:28,310 --> 00:50:32,990
I don't know how we wouldn't, and formal
education just is not gonna remain.
953
00:50:34,040 --> 00:50:37,940
Relevant, in my opinion, if
we don't adapt and adjust.
954
00:50:38,030 --> 00:50:39,050
So there you go.
955
00:50:39,080 --> 00:50:40,820
There's my, my parting thoughts.
956
00:50:40,820 --> 00:50:47,690
Thank you Erin, so much for coming and
chatting with me and providing silicon.
957
00:50:47,720 --> 00:50:50,390
I'm gonna say you're providing the
Silicon Valley perspective, even
958
00:50:50,390 --> 00:50:52,700
though you are in Wyoming, you're
959
00:50:52,744 --> 00:50:53,074
Aaron Makelky: Yeah,
960
00:50:53,390 --> 00:50:53,690
Lindy Hockenbary: that.
961
00:50:55,084 --> 00:50:58,144
Aaron Makelky: and what I know is due
to working with really smart people
962
00:50:58,144 --> 00:51:02,074
that are light years ahead of me and
just distilling little useful bits that
963
00:51:02,074 --> 00:51:06,004
a teacher or an administrator, someone
like you and I would actually put to use.
964
00:51:06,620 --> 00:51:07,760
Lindy Hockenbary: Yep, for sure.
965
00:51:07,850 --> 00:51:08,330
Awesome.
966
00:51:08,330 --> 00:51:08,840
All right.
967
00:51:08,840 --> 00:51:09,980
Thank you so much.
968
00:51:10,539 --> 00:51:11,189
Aaron Makelky: Thank you, Lindy.
969
00:51:15,092 --> 00:51:17,162
Thanks for joining Make EdTech 100.
970
00:51:17,882 --> 00:51:22,562
I know educator time is valuable and I'm
honored you choose to spend yours with me.
971
00:51:23,612 --> 00:51:28,262
For more EdTech strategies you can use
tomorrow and ways to bring me to your
972
00:51:28,262 --> 00:51:30,762
school or event, head to LindyHoc.com.
973
00:51:31,952 --> 00:51:35,852
If this episode resonated, hit subscribe
so you don't miss the next one.
974
00:51:36,632 --> 00:51:36,972
I'm LindyHoc.
975
00:51:37,652 --> 00:51:39,962
Go forth and make EdTech 100.
