1
00:00:00,960 --> 00:00:05,117
Richard, I'm I'm curious.
Where are we at the moment with RSI

2
00:00:05,200 --> 00:00:09,677
>> in various weak forms? We already
have RSI. We're not quite there yet, but

3
00:00:09,760 --> 00:00:14,237
we're very close.
>> When do you believe we reach ASI?

4
00:00:14,320 --> 00:00:16,637
>> I think we'll take us probably several

5
00:00:16,720 --> 00:00:18,237
decades.

6
00:00:18,320 --> 00:00:21,437
>> On Terminal Bench 4.0,
Sonnet 5.5 jumped

7
00:00:21,520 --> 00:00:24,637
from 10% to 70%. Pretty extraordinary.

8
00:00:24,720 --> 00:00:28,477
My theory would be that they're trying
to compete with, you know, China, which

9
00:00:28,560 --> 00:00:32,557
is about three months behind, and fill
that gap before a lot of enterprises go

10
00:00:32,640 --> 00:00:35,997
to to open source models. >> It's
not at all obvious to me why anyone

11
00:00:36,080 --> 00:00:38,877
should be using Sonnet 5.5 over Opus 5.5

12
00:00:38,960 --> 00:00:43,517
unless you have some token or latency
or other consideration. I do not plan to

13
00:00:43,600 --> 00:00:47,437
use Sonnet 5.5. >> The Defense
Secretary announced Project

14
00:00:47,520 --> 00:00:49,757
Meridian. It's a new Pentagon effort on

15
00:00:49,840 --> 00:00:52,877
the future of warfare. It's co-led by

16
00:00:52,960 --> 00:00:54,637
Elon Musk and Palmer Lucky.

17
00:00:54,720 --> 00:00:57,197
>> This I think is a
transformative moment.

18
00:00:57,280 --> 00:00:59,917
This is a major major step forward. The

19
00:01:00,000 --> 00:01:03,597
big question is going to be.

20
00:01:03,680 --> 00:01:08,077
>> Now that's a moonshot.
Ladies and gentlemen,

21
00:01:08,160 --> 00:01:09,757
>> this episode is brought to you by the

22
00:01:09,840 --> 00:01:13,997
Abundance Summit and Link Ventures.

23
00:01:14,080 --> 00:01:17,277
>> Welcome to Moonshots, everyone.
Your number one podcast and help you

24
00:01:17,360 --> 00:01:19,357
understand the singularity. What's going

25
00:01:19,440 --> 00:01:21,917
on? What does it mean to you and your

26
00:01:22,000 --> 00:01:24,637
family and what the breaking news is? We

27
00:01:24,720 --> 00:01:29,037
publish this twice a week to help you
keep up with the extraordinary rate of

28
00:01:29,120 --> 00:01:31,517
change. With me today is the Fantastic

29
00:01:31,600 --> 00:01:35,117
Four, Alex Weer Gross, our in-house ASI,

30
00:01:35,200 --> 00:01:37,517
Dave Blendon, our improario of AI

31
00:01:37,600 --> 00:01:42,877
investing, Salem Ismael, our Globe
Troder who's home today. Amazing Sem.

32
00:01:42,960 --> 00:01:46,637
>> Woohoo. >> And the and the father of
the organizational singularity. I'm Peter

33
00:01:46,720 --> 00:01:48,557
Diamandis, your host and datadriven

34
00:01:48,640 --> 00:01:51,197
optimist. Our mission here, keep you

35
00:01:51,280 --> 00:01:53,837
optimistic about the future. Uh, and

36
00:01:53,920 --> 00:01:55,677
here's the numbers from this week. In 7

37
00:01:55,760 --> 00:01:58,317
days, Google shipped Gemini for Argon.

38
00:01:58,400 --> 00:02:01,117
Anthropic shipped Sonnet 5.5. Open eye

39
00:02:01,200 --> 00:02:03,837
shipped GPT 6.1.

40
00:02:03,920 --> 00:02:06,957
Saul and the Frontier Intelligence has

41
00:02:07,040 --> 00:02:09,437
gotten cheaper by three-fold. SpaceX

42
00:02:09,520 --> 00:02:11,197
launched the next batch of astronauts,

43
00:02:11,280 --> 00:02:14,557
the ISS, and also launched Google's TPUs

44
00:02:14,640 --> 00:02:19,117
into orbit for Project Suncatcher. Uh,
if you're new to the Moonshots podcast,

45
00:02:19,200 --> 00:02:23,037
please hit subscribe. We publish twice
a week and you don't want to miss any of

46
00:02:23,120 --> 00:02:25,197
this news during this hypersonic

47
00:02:25,280 --> 00:02:28,317
tsunami. Our moonshot here is to 20x our

48
00:02:28,400 --> 00:02:33,357
subscriber base to get to 10 million to
help spread the word of optimism and the

49
00:02:33,440 --> 00:02:37,517
extraordinary future that we're
building. Today on Moonshot, we're

50
00:02:37,600 --> 00:02:39,357
joined by one of the architects of

51
00:02:39,440 --> 00:02:43,037
modern AI, Richard. Uh, born in Germany

52
00:02:43,120 --> 00:02:47,037
and trained at Stanford. Richard
is among the world's most cited natural

53
00:02:47,120 --> 00:02:51,277
language processing researchers,
a pioneer in deep learning and prompt

54
00:02:51,360 --> 00:02:53,517
engineering and a serial founder who's

55
00:02:53,600 --> 00:02:55,677
repeatedly turned frontier research into

56
00:02:55,760 --> 00:02:58,477
category defining companies. First, he

57
00:02:58,560 --> 00:03:03,037
founded Metammind uh which was acquired
by Salesforce where he then became the

58
00:03:03,120 --> 00:03:05,277
chief scientist leading AI research for

59
00:03:05,360 --> 00:03:07,597
friend of the pod Mark Beni off. Next,

60
00:03:07,680 --> 00:03:10,637
Richard founded you.com, recognized by

61
00:03:10,720 --> 00:03:13,197
Time in the World Economic Forum and now

62
00:03:13,280 --> 00:03:16,637
valued at north of 1.5 billion as well

63
00:03:16,720 --> 00:03:19,837
as his fund AO aix

64
00:03:19,920 --> 00:03:22,237
ventures. Today, Richard is co-founder

65
00:03:22,320 --> 00:03:25,597
and CEO of Recursive, perhaps uh the

66
00:03:25,680 --> 00:03:28,317
biggest moonshot he's ever taken, uh

67
00:03:28,400 --> 00:03:30,317
focused on recursively self-improving

68
00:03:30,400 --> 00:03:33,197
super intelligence. Richard, congrats on

69
00:03:33,280 --> 00:03:35,837
a $670 million fund raise from Google

70
00:03:35,920 --> 00:03:38,317
Ventures. Graycraft, Nvidia, and AMD and

71
00:03:38,400 --> 00:03:41,117
410 million in compute from AWS.

72
00:03:41,200 --> 00:03:43,437
Personal disclosure, I'm very proud to

73
00:03:43,520 --> 00:03:46,477
be a seed investor in Recursive. And of

74
00:03:46,560 --> 00:03:48,477
course, Richard, uh, important to

75
00:03:48,560 --> 00:03:51,197
mention your new book just got released,

76
00:03:51,280 --> 00:03:54,317
The Eureka Machine: Why AI is the key to

77
00:03:54,400 --> 00:03:59,357
unlocking a new era of scientific
discoveries. Welcome to the pod, Richard.

78
00:03:59,440 --> 00:04:02,637
>> So great to be back.
>> And let me just add, a truly truly

79
00:04:02,720 --> 00:04:06,877
awesome guy. A lot of people who listen
to the pot are worried about the the

80
00:04:06,960 --> 00:04:11,117
ethics and the the risks and everything,
but boy, if you want a a person to

81
00:04:11,200 --> 00:04:15,517
conquer recursive self-improvement that
you can like and trust, Richard is the

82
00:04:15,600 --> 00:04:18,397
man. >> Yeah. >> Thank you
so much. So great to be back

83
00:04:18,480 --> 00:04:22,397
here. I love your guys. Constructive
optimism in the world right now.

84
00:04:22,480 --> 00:04:24,557
>> I have I have a bold
prediction for this episode.

85
00:04:24,640 --> 00:04:28,077
>> What's that? >> This one
episode will prove our thesis

86
00:04:28,160 --> 00:04:32,237
for this whole podcast more than
any other episode we ever record.

87
00:04:32,320 --> 00:04:34,557
>> Awesome. Let's go. >> Let's go.

88
00:04:34,640 --> 00:04:37,437
>> And uh uh Alex, do you know Richard?

89
00:04:37,520 --> 00:04:41,117
>> Yeah, Richard, you and I have
chatted quite a bit. I don't think we've

90
00:04:41,200 --> 00:04:43,117
actually ever met in person, though.

91
00:04:43,200 --> 00:04:44,957
>> Really? >> We have not.
And it's the first time on

92
00:04:45,040 --> 00:04:46,477
a podcast together.
It's going to be awesome.

93
00:04:46,560 --> 00:04:48,397
>> Amazing. Well, let's make history.

94
00:04:48,480 --> 00:04:51,517
>> Yeah, for sure. So, uh Richard, let's

95
00:04:51,600 --> 00:04:55,917
start with your book. Uh the central
claim of the Eureka machine is that AI

96
00:04:56,000 --> 00:05:00,477
will deliver a century of scientific
breakthroughs in the next decade.

97
00:05:00,560 --> 00:05:02,877
uh you know this maps directly onto what

98
00:05:02,960 --> 00:05:05,117
Alex and I wrote in solve everything. So

99
00:05:05,200 --> 00:05:07,757
we're fans of that prediction. Uh your

100
00:05:07,840 --> 00:05:10,317
thesis in the book is that every stage

101
00:05:10,400 --> 00:05:13,197
of scientific process gets connected and

102
00:05:13,280 --> 00:05:15,517
transformed simultaneously. Hypothesis,

103
00:05:15,600 --> 00:05:18,557
experiment, data theory, you call it the

104
00:05:18,640 --> 00:05:21,757
full stack AI. Uh you also say that

105
00:05:21,840 --> 00:05:24,317
scientific progress has slowed and

106
00:05:24,400 --> 00:05:26,477
that's been caused not by underfunding

107
00:05:26,560 --> 00:05:29,517
but by fragmentation. So let's start

108
00:05:29,600 --> 00:05:31,917
with those two items. Could you take a

109
00:05:32,000 --> 00:05:33,997
moment and you know talk to us about

110
00:05:34,080 --> 00:05:36,797
full stack AI for science and why you

111
00:05:36,880 --> 00:05:39,677
think progress has slowed?

112
00:05:39,760 --> 00:05:43,437
>> Yeah. So progress has slowed uh largely

113
00:05:43,520 --> 00:05:47,437
because we have so many different subd
disccipline and uh disciplines and even

114
00:05:47,520 --> 00:05:51,677
sus of lem uh realized like wow that
like if there are only so many people

115
00:05:51,760 --> 00:05:55,437
and we have more and more fragmentation
of more and more subd disciplines and

116
00:05:55,520 --> 00:05:57,997
niches like you can not just do AI right

117
00:05:58,080 --> 00:06:01,117
you're doing often like optimization uh

118
00:06:01,200 --> 00:06:03,357
and gradient descent methods and second

119
00:06:03,440 --> 00:06:06,397
order derivative methods uh and so on of

120
00:06:06,480 --> 00:06:10,557
uh these neural networks that is which
is one category of AI and so on. And the

121
00:06:10,640 --> 00:06:13,917
same is true in biology. You study
biology, you're going to be either a

122
00:06:14,000 --> 00:06:16,877
cell biologist or a molecular biologist

123
00:06:16,960 --> 00:06:22,077
uh or you're like in medicine and like
these like there's so much separation

124
00:06:22,160 --> 00:06:23,677
that it's hard to weave it all together.

125
00:06:23,760 --> 00:06:26,317
And that is the perfect time for AI to

126
00:06:26,400 --> 00:06:30,957
come in and help us weave back together
all these separate pieces and also help

127
00:06:31,040 --> 00:06:33,037
us understand how these large complex

128
00:06:33,120 --> 00:06:38,317
systems actually work. like we can't
have one model where humans say here are

129
00:06:38,400 --> 00:06:42,157
all the things I know about natural
language and now like we can put all

130
00:06:42,240 --> 00:06:45,837
these rules together and then we have
a conversation. It required a large neural

131
00:06:45,920 --> 00:06:47,917
network with a ton of data. And guess

132
00:06:48,000 --> 00:06:50,797
what? We're going to get a lot of data

133
00:06:50,880 --> 00:06:55,357
uh about biology. And so we can make
more and more disciplines and move them

134
00:06:55,440 --> 00:06:59,117
transfer them from being traditionally
natural sciences where we just try to

135
00:06:59,200 --> 00:07:01,757
understand biology and nature to being

136
00:07:01,840 --> 00:07:04,557
programmable engineering sciences. And

137
00:07:04,640 --> 00:07:06,797
that I think is one of the many

138
00:07:06,880 --> 00:07:12,157
different exciting aspects of things
to come. And you know why why am I so

139
00:07:12,240 --> 00:07:15,197
excited that that this is happening
right now and and what are the major

140
00:07:15,280 --> 00:07:18,557
ingredients? It's essentially that we
now have world knowledge in the form of

141
00:07:18,640 --> 00:07:22,477
LMS. We have more and more scientific
data that's getting digitized that we

142
00:07:22,560 --> 00:07:26,477
can sit on top of. We have better and
better simulations of things and we have

143
00:07:26,560 --> 00:07:30,157
robotic process automation that will
soon be possible. And on top of that

144
00:07:30,240 --> 00:07:31,517
we're going to have an agent swarm and

145
00:07:31,600 --> 00:07:34,717
that is how uh this the full scientific

146
00:07:34,800 --> 00:07:38,237
stack uh can be automated. Alex.

147
00:07:38,320 --> 00:07:42,157
>> Yeah, I can say I disagree. As Peter,
you and I wrote in solve everything,

148
00:07:42,240 --> 00:07:45,437
math, science, and engineering are cooked.
I'm curious, Richard, what your

149
00:07:45,520 --> 00:07:48,317
latest timelines are.

150
00:07:48,400 --> 00:07:51,117
>> I mean, the I think
every field is going

151
00:07:51,200 --> 00:07:54,317
to get to higher and higher levels
of abstraction. Computer science has been

152
00:07:54,400 --> 00:07:57,757
very good at that, right? No one
is programming in zeros and ones anymore.

153
00:07:57,840 --> 00:07:59,997
No one like very few people have to

154
00:08:00,080 --> 00:08:02,397
still know C++ and complex pointers and

155
00:08:02,480 --> 00:08:04,957
and memory uh architectures and so on.

156
00:08:05,040 --> 00:08:07,037
we can get a larger and larger and

157
00:08:07,120 --> 00:08:08,797
computer science has basically

158
00:08:08,880 --> 00:08:12,157
abstracted enough so that it met the

159
00:08:12,240 --> 00:08:14,237
rest of humanity with English, right?

160
00:08:14,320 --> 00:08:16,237
And natural language that now can be

161
00:08:16,320 --> 00:08:18,797
used to do computer science. And so I

162
00:08:18,880 --> 00:08:21,517
think that um gives me great hope uh

163
00:08:21,600 --> 00:08:24,077
that we can get other fields into

164
00:08:24,160 --> 00:08:27,037
similar uh stages of abstraction and

165
00:08:27,120 --> 00:08:31,677
then all meet and natural language
to do science. But to pin you down, as I

166
00:08:31,760 --> 00:08:35,997
recall from my understanding of your book,
3 to five years for the physical

167
00:08:36,080 --> 00:08:39,277
sciences, >> you know,
like there's no like single

168
00:08:39,360 --> 00:08:42,877
sort of this is the threshold and now we
solved all of physical sciences, right?

169
00:08:42,960 --> 00:08:47,357
But like multiple different diseases
will get cured in the next 12 months.

170
00:08:47,440 --> 00:08:51,277
They're going to be the simpler diseases
maybe where there's one gene that needs

171
00:08:51,360 --> 00:08:52,957
to be fixed for that disease to be

172
00:08:53,040 --> 00:08:55,997
cured. Um but there are a lot of single

173
00:08:56,080 --> 00:09:00,717
gene diseases uh that that are out there
in aggregate, right? like and so those

174
00:09:00,800 --> 00:09:03,837
diseases will get cured and then we're
going to cure more and more complex

175
00:09:03,920 --> 00:09:07,837
diseases. We're going to develop
better and better battery material. So I

176
00:09:07,920 --> 00:09:10,877
wouldn't call it like one threshold
in like in 3 years sort of everything is

177
00:09:10,960 --> 00:09:13,757
solved but like we're going to solve
more and more problems and it'll just be

178
00:09:13,840 --> 00:09:18,877
a question of how much money do you want
to put into compute to then solve which

179
00:09:18,960 --> 00:09:21,197
kinds of problems? >> You're
you're talking to the guy Richard

180
00:09:21,280 --> 00:09:24,477
remember who argues that the singularity
itself is something of an optical

181
00:09:24,560 --> 00:09:26,237
illusion that there is no step function.

182
00:09:26,320 --> 00:09:28,397
It's just a time interval. So I'll I'll

183
00:09:28,480 --> 00:09:32,317
try once more timelines. When when even

184
00:09:32,400 --> 00:09:34,877
if it's not a step function, when is the

185
00:09:34,960 --> 00:09:38,957
inflection point when is the um 50%

186
00:09:39,040 --> 00:09:42,957
point crossed in your mind for all
of the physical sciences getting solved?

187
00:09:43,040 --> 00:09:46,477
All human disease getting solved?
Where are those timelines?

188
00:09:46,560 --> 00:09:50,477
>> Alex, if I could just add a a layer on
top of that. So we've talked on this pot

189
00:09:50,560 --> 00:09:52,637
a lot that math is cooked, right? We're

190
00:09:52,720 --> 00:09:54,717
seeing millennium prizes all falling.

191
00:09:54,800 --> 00:09:57,837
Yep. And the n the next natural uh if

192
00:09:57,920 --> 00:10:03,517
you would uh uh barbecue is likely
to be physics >> incineration.

193
00:10:03,600 --> 00:10:06,317
>> Incineration. Yes.
>> And computer science arguably I mean

194
00:10:06,400 --> 00:10:09,277
presumably this is part of the premise
Richard of recursive as well. Computer

195
00:10:09,360 --> 00:10:12,877
science already cooked
math thoroughly cooked.

196
00:10:12,960 --> 00:10:16,237
>> You know in a weird way I would I I
would use different metaphors. I think

197
00:10:16,320 --> 00:10:17,997
they're flourishing not getting cooked.

198
00:10:18,080 --> 00:10:21,357
I think the weird thing is, you know,
some some mathematicians recently said,

199
00:10:21,440 --> 00:10:25,117
"Oh, this is this may be bad for the
field because like we need to train

200
00:10:25,200 --> 00:10:29,117
people." Yes, we need to train people,
but imagine a biologist or or medical

201
00:10:29,200 --> 00:10:31,757
researcher saying, "Oh, you know,
it's really a bummer for the field that we

202
00:10:31,840 --> 00:10:33,597
cured all these diseases for people."

203
00:10:33,680 --> 00:10:36,717
Like that would be insane, right? Of
course, you want to just move as quickly

204
00:10:36,800 --> 00:10:39,757
as you can towards solving these problems.
I think maybe the problem with

205
00:10:39,840 --> 00:10:43,197
math is that there's many sub fields in

206
00:10:43,280 --> 00:10:47,437
of math that haven't really had
there are just almost purely beautiful

207
00:10:47,520 --> 00:10:51,677
intellectual exercises without real life
impact anymore. But as you get close to

208
00:10:51,760 --> 00:10:55,917
real life impact, you should be excited
about it and you think and realize your

209
00:10:56,000 --> 00:10:57,917
field is flourishing, not being cooked.

210
00:10:58,000 --> 00:11:03,197
Uh, and so to timelines, I think the the
further the more complex a disease is,

211
00:11:03,280 --> 00:11:06,397
the more you have to do long-term
studies with humans before you're

212
00:11:06,480 --> 00:11:08,557
allowed to put a certain drug into

213
00:11:08,640 --> 00:11:11,197
humans, the more delays you will have.

214
00:11:11,280 --> 00:11:15,757
But like we see now companies to make
this very concrete, biotech companies

215
00:11:15,840 --> 00:11:19,517
used to have like one new drug
in development and it took them 10 years.

216
00:11:19,600 --> 00:11:22,957
They had to go public before they knew
the comp the drug was really working.

217
00:11:23,040 --> 00:11:25,837
And then after 10 years maybe late stage

218
00:11:25,920 --> 00:11:28,317
three trials were just not working and

219
00:11:28,400 --> 00:11:30,397
then the company is dead. Right now you

220
00:11:30,480 --> 00:11:33,117
have in the new age of companies they

221
00:11:33,200 --> 00:11:36,557
have 5 to 10 different compounds in uh

222
00:11:36,640 --> 00:11:41,117
stage three trials after 2 or 3
years and and so like we see a lot of

223
00:11:41,200 --> 00:11:45,757
acceleration in that level but again
diseases to really be able to get them

224
00:11:45,840 --> 00:11:49,437
into humans at scale in the United
States with FDA approvals there are just

225
00:11:49,520 --> 00:11:54,397
some natural delays that will be more
like half a decade to a decade uh but uh

226
00:11:54,480 --> 00:11:58,477
lots of other things in chemistry
and physics and so on where we can iterate

227
00:11:58,560 --> 00:12:00,397
without having to look at long-term

228
00:12:00,480 --> 00:12:02,797
human trials will be even faster.

229
00:12:02,880 --> 00:12:05,597
>> And I the the the
point is well taken. I

230
00:12:05,680 --> 00:12:08,157
I like your Orwellian turn of phrase.

231
00:12:08,240 --> 00:12:12,477
Maybe instead of saying AI is cooking
math, I should be using flourishing as a

232
00:12:12,560 --> 00:12:17,037
transitive verb and just
say AI is flourishing math.

233
00:12:17,120 --> 00:12:20,957
>> Actually, I was with Certas Caraman
who runs lids at MIT uh the night before

234
00:12:21,040 --> 00:12:25,917
last. He he started a drone company now,
an AI drone company. But lids is where

235
00:12:26,000 --> 00:12:30,157
radar was invented originally during
World War II. It's a great lab, very

236
00:12:30,240 --> 00:12:34,077
entrepreneurial. But he said all his
mathematician friends at MIT uh are

237
00:12:34,160 --> 00:12:37,197
aware that they're cooked
and they're they're all

238
00:12:37,280 --> 00:12:38,957
>> No, no, Dave.
Dave, we're going with the

239
00:12:39,040 --> 00:12:42,397
Orwellian language now. They're they're
not cooked, they're being flourished.

240
00:12:42,480 --> 00:12:45,597
>> He That's funny because he literally
said cooked, but I'll go back to him and

241
00:12:45,680 --> 00:12:47,357
tell your friends they're flourishing.

242
00:12:47,440 --> 00:12:52,477
>> Tell not to worry.
If your goal was to prove

243
00:12:52,560 --> 00:12:56,397
as many theorems as possible in your
lifetime, like now is the time to just

244
00:12:56,480 --> 00:13:00,477
grab like grab as many as you can and
and work with AI to solve them. And then

245
00:13:00,560 --> 00:13:04,157
I think the field will change the way
computer science has changed in many

246
00:13:04,240 --> 00:13:09,197
ways. Uh and it'll be much more about
what's the most creative thing when you

247
00:13:09,280 --> 00:13:12,557
really understand all the things that are
out there in math. What kind of new

248
00:13:12,640 --> 00:13:16,557
formalisms can you create?
New constructs can you create that then

249
00:13:16,640 --> 00:13:21,517
would be interesting to be solved uh
by an AI in collaboration with humans.

250
00:13:21,600 --> 00:13:25,757
>> Well, Cash was saying the math guys are
in great shape because they're so cooked

251
00:13:25,840 --> 00:13:28,957
that they're all moving over to AI
orchestration and they're going to be

252
00:13:29,040 --> 00:13:31,837
way ahead of the curve. He he's
actually most worried about the biology

253
00:13:31,920 --> 00:13:34,397
professors who are in complete denial.

254
00:13:34,480 --> 00:13:37,917
You using virtually no AI in their
day-to-day activities. And this is one

255
00:13:38,000 --> 00:13:41,917
area where Richard, you have a such
a deep background in all facets of AI,

256
00:13:42,000 --> 00:13:45,997
including biology. And I I feel like
we're living exactly parallel lives

257
00:13:46,080 --> 00:13:47,517
except you're 15 years younger than me.

258
00:13:47,600 --> 00:13:51,357
So I'm like insanely jealous of of
your life trajectory. But be a serial

259
00:13:51,440 --> 00:13:54,877
entrepreneur, then start a hugely
successful venture fund, then found a

260
00:13:54,960 --> 00:13:58,637
foundation model company. And and you
know, we're seeing the world through the

261
00:13:58,720 --> 00:14:00,797
exact same lens. But you and Peter have

262
00:14:00,880 --> 00:14:03,197
much more biology background. I I have

263
00:14:03,280 --> 00:14:07,837
basically none. uh but the biologists
are the ones really really lagging and I

264
00:14:07,920 --> 00:14:10,397
I think you've got a bunch
of investments actually that are that have

265
00:14:10,480 --> 00:14:13,757
done really well in the area right >>
and I would love to talk about some of

266
00:14:13,840 --> 00:14:18,557
those like peril bio proximal labs Ignod
labs like truly exciting companies I

267
00:14:18,640 --> 00:14:22,717
think another field that is even
more lacking than biology is economics

268
00:14:22,800 --> 00:14:25,597
economics literally
has these models of like

269
00:14:25,680 --> 00:14:28,397
>> a linear model of economics right a

270
00:14:28,480 --> 00:14:32,717
one-step economy that's provably
correctly taxed and and subsidized and

271
00:14:32,800 --> 00:14:34,797
things like that it's just like absurd

272
00:14:34,880 --> 00:14:40,557
how slow that field is to adopt AI
for making better policy decisions

273
00:14:40,640 --> 00:14:42,797
ideology >> because you
opened the door. Eric

274
00:14:42,880 --> 00:14:45,997
Bolson, our very good friend, uh I did

275
00:14:46,080 --> 00:14:49,357
not realize till yesterday that HI lab

276
00:14:49,440 --> 00:14:53,197
high at Stanford invented
the word foundation model

277
00:14:53,280 --> 00:14:56,237
>> and he sent me the whole you
didn't you knew that already? Wow.

278
00:14:56,320 --> 00:14:59,837
>> Yeah, of course. Like lots of Stanford
friends and Percy Lang and others uh

279
00:14:59,920 --> 00:15:03,197
worked on foundational models. I love
Eric Goffson. actually one of the most

280
00:15:03,280 --> 00:15:06,717
interesting economists I think uh doing
really interesting research right now.

281
00:15:06,800 --> 00:15:11,917
He also started work helix which we're
a proud investor in uh that brings uh like

282
00:15:12,000 --> 00:15:13,437
understanding of how companies actually

283
00:15:13,520 --> 00:15:16,477
adopt AI uh and which tasks are getting

284
00:15:16,560 --> 00:15:21,197
you know helped by in real rollouts for
companies. Uh yeah he's a co-founder of

285
00:15:21,280 --> 00:15:23,597
that too. >> Yeah building
is just such a great place

286
00:15:23,680 --> 00:15:27,517
cuz you talk about econ people being way
off the curve but he is so ahead of the

287
00:15:27,600 --> 00:15:30,957
curve. You walk into the building and you
can just feel it. He's also poaching

288
00:15:31,040 --> 00:15:34,957
a ton of talent from MIT
to come out to to join his lab,

289
00:15:35,040 --> 00:15:39,277
>> following his path. Richard, talk
about full stack AI as you see it in the

290
00:15:39,360 --> 00:15:43,757
scientific method. Um, you know, I'm an
investor in a company called Laya out of

291
00:15:43,840 --> 00:15:46,157
MIT and Harvard. You know, Lyla science

292
00:15:46,240 --> 00:15:50,877
as well. Uh, I think I've introduced
you to to Jeff von Maltson there. and you

293
00:15:50,960 --> 00:15:55,117
know they're building a scientific
super intelligence that is then running a

294
00:15:55,200 --> 00:15:57,277
million square foot of robotic space to

295
00:15:57,360 --> 00:15:59,357
sort of like a thousandx the rate of

296
00:15:59,440 --> 00:16:02,717
discovery. Um your thoughts on that?

297
00:16:02,800 --> 00:16:06,317
>> I absolutely love it. I think that is
you know in the Eureka machine I sort of

298
00:16:06,400 --> 00:16:11,117
lay out these four pillars and they're
actually one of the few um uh that that

299
00:16:11,200 --> 00:16:14,397
are really going after this similar
to periodic labs who are doing it more on

300
00:16:14,480 --> 00:16:18,157
the physics and chemistry side of things.
Uh, and there are several other

301
00:16:18,240 --> 00:16:21,837
companies now that I can hopefully soon
talk about that are trying to create

302
00:16:21,920 --> 00:16:24,237
more data for the bitter lesson to be

303
00:16:24,320 --> 00:16:27,117
applicable in biology. And I love uh

304
00:16:27,200 --> 00:16:29,037
what you know the big guys Eli Liy and

305
00:16:29,120 --> 00:16:32,157
others do uh and and Laya too. I think

306
00:16:32,240 --> 00:16:36,477
you know the ultimately uh you can
boil down the scientific method to the

307
00:16:36,560 --> 00:16:38,637
ideation implementation and validation

308
00:16:38,720 --> 00:16:40,717
of ideas and the faster we can close

309
00:16:40,800 --> 00:16:43,677
that loop and then put an open-ended

310
00:16:43,760 --> 00:16:47,837
process on top of it that's what
open-endedness uh what inspired us a lot

311
00:16:47,920 --> 00:16:51,677
at recursive and we have many of the
world's greatest uh like researchers in

312
00:16:51,760 --> 00:16:53,997
that subdomain of AI that is still not

313
00:16:54,080 --> 00:16:56,157
quite as popular as it could and should

314
00:16:56,240 --> 00:17:00,317
be um like the more you can have a swarm

315
00:17:00,400 --> 00:17:04,957
uh innovate in open-ended ways
and evolve uh and combine interestingly

316
00:17:05,040 --> 00:17:06,877
different ideas the better. But then of

317
00:17:06,960 --> 00:17:09,437
course in biology you have to have uh

318
00:17:09,520 --> 00:17:12,877
like actual uh like physical biological

319
00:17:12,960 --> 00:17:14,717
wetwware experiments. And so it's really

320
00:17:14,800 --> 00:17:19,117
great to see um uh Laya doing that. Um

321
00:17:19,200 --> 00:17:22,397
we're seeing this also maybe I can talk
about this one company that I really

322
00:17:22,480 --> 00:17:26,877
love called Peril Bio.
They build tiny organoids on a petri dish,

323
00:17:26,960 --> 00:17:31,517
>> right? And get this, if you love
animals, you too can love AI. Why?

324
00:17:31,600 --> 00:17:35,437
Because they got FDA approval to skip
animal trials. Turns out we cured most

325
00:17:35,520 --> 00:17:37,277
diseases in mice. It's not that helpful.

326
00:17:37,360 --> 00:17:41,757
They're very different to people. Uh but
these guys uh use pur potent stem cells

327
00:17:41,840 --> 00:17:44,797
create tiny little organoids of lymph
nodes. Lymph nodes big part of your

328
00:17:44,880 --> 00:17:48,237
immune system. Imunotherapy some of the
most exciting therapies to allow your

329
00:17:48,320 --> 00:17:51,197
own immune system to attack a cancer
and so on instead of getting crazy

330
00:17:51,280 --> 00:17:55,357
chemotherapy and so on. And so they got
FDA approval because they showed that

331
00:17:55,440 --> 00:17:59,357
when you run experiments in parallel,
hence parallel bio in these tiny petri

332
00:17:59,440 --> 00:18:03,597
dishes, how those organoids react to
different drugs and toxicity testing and

333
00:18:03,680 --> 00:18:07,677
so on is actually more predictive of how
those drugs will interact in real human

334
00:18:07,760 --> 00:18:11,917
bodies. And that is just one of many
beautiful examples of where it will

335
00:18:12,000 --> 00:18:13,837
help. And you can take my stem cell, you

336
00:18:13,920 --> 00:18:16,397
can create take my skin, create in a uh

337
00:18:16,480 --> 00:18:19,997
IPSC uh cell, grow my own organs and see

338
00:18:20,080 --> 00:18:24,717
how a particular drug would work for me
versus a generic individual. Uh Salem,

339
00:18:24,800 --> 00:18:27,997
you want to jump in? >> Uh yeah,
a couple of things. One is, you

340
00:18:28,080 --> 00:18:31,117
know, science has always been
a coordination problem, right? You're kind

341
00:18:31,200 --> 00:18:32,557
of trying to bring and it's always been

342
00:18:32,640 --> 00:18:35,837
very structured into departments and and

343
00:18:35,920 --> 00:18:39,597
journals and all that stuff. And I love
what you're doing, bringing it together.

344
00:18:39,680 --> 00:18:41,517
For me, if I had to summarize what you

345
00:18:41,600 --> 00:18:43,517
seem to be doing is you're collapsing

346
00:18:43,600 --> 00:18:48,637
the time between an experiment and a
result or hypothesis and a result. And

347
00:18:48,720 --> 00:18:51,917
when you can collapse that,
that's domain collapse of the scientific

348
00:18:52,000 --> 00:18:55,997
method. And now like anything
is possible from there. This is amazing.

349
00:18:56,080 --> 00:18:57,997
>> Yeah. I I am going
to push again on what

350
00:18:58,080 --> 00:19:00,557
what Alex said, physics. When do you you

351
00:19:00,640 --> 00:19:03,037
know you know Einstein's theory

352
00:19:03,120 --> 00:19:05,197
relativity comes out. There's not that

353
00:19:05,280 --> 00:19:08,317
much progress uh you know since then. a

354
00:19:08,400 --> 00:19:11,597
lot of theories. Do you see physics as

355
00:19:11,680 --> 00:19:15,597
the next domain that's going
to flourish on the back of math?

356
00:19:15,680 --> 00:19:19,197
>> I think the biggest domain is actually
going to be biology. Physics has been

357
00:19:19,280 --> 00:19:21,277
interestingly stuck in many ways like

358
00:19:21,360 --> 00:19:23,677
they there's a really powerful ideas

359
00:19:23,760 --> 00:19:28,317
like E= MC². So you can get a ton
of energy out of potentially little mass

360
00:19:28,400 --> 00:19:30,157
and then we got nuclear energy out of

361
00:19:30,240 --> 00:19:35,357
that. uh and so that then moved into
and graduated in some ways into an

362
00:19:35,440 --> 00:19:39,917
engineering discipline uh which is where
you have the real real life impact. Um I

363
00:19:40,000 --> 00:19:44,397
think you know obviously like fusion
will be great to finally get figured out

364
00:19:44,480 --> 00:19:48,557
and there's a lot of really cool like
companies and and big labs and you know

365
00:19:48,640 --> 00:19:50,797
tokomax already to balance uh like

366
00:19:50,880 --> 00:19:53,597
plasma inside tokomax like is a very

367
00:19:53,680 --> 00:19:55,917
hard control problem that where AI is

368
00:19:56,000 --> 00:19:58,637
being used. Um I hope we can eventually

369
00:19:58,720 --> 00:20:02,957
make better theories for uh you know

370
00:20:03,040 --> 00:20:04,717
quantum gravity and all kinds of other

371
00:20:04,800 --> 00:20:06,877
complex issues and and have get a better

372
00:20:06,960 --> 00:20:10,797
world sort of formula. Um I

373
00:20:10,880 --> 00:20:15,117
unfortunately the data set collections
these days require often like large

374
00:20:15,200 --> 00:20:19,197
hydron colliders and just like billions
of dollars and it's like quite expensive

375
00:20:19,280 --> 00:20:23,917
actually. So in a weird way, biology
is a better fit for AI because what

376
00:20:24,000 --> 00:20:25,517
calculus did for physics like

377
00:20:25,600 --> 00:20:29,117
understanding really well micro uh

378
00:20:29,200 --> 00:20:33,597
individualized separated phenomena,
neural nets are great in combining all

379
00:20:33,680 --> 00:20:37,117
these little things like we know what
the neuron does. We know what one uh

380
00:20:37,200 --> 00:20:40,957
bacteria does in our microbiome, but then
as they all come together and form

381
00:20:41,040 --> 00:20:45,197
these really complex interactions,
we don't really know anymore how that

382
00:20:45,280 --> 00:20:46,877
works. And so that's where I I think

383
00:20:46,960 --> 00:20:50,157
will help us more. Yeah, Alex,

384
00:20:50,240 --> 00:20:54,237
>> let's talk about biology a bit. So,
do you think the critical path to solving

385
00:20:54,320 --> 00:20:56,557
biology goes through digital twins of

386
00:20:56,640 --> 00:21:01,517
cells or do you have some wildly
different theory of the case? Three

387
00:21:01,600 --> 00:21:05,277
years ago when I started Eureka Machine
book, I had this chapter on the virtual

388
00:21:05,360 --> 00:21:08,077
cell and I was really proud of like
trying to lay it all out and I had to

389
00:21:08,160 --> 00:21:11,437
rewrite that whole thing because in
the meantime uh you know Mark Zuckerberg

390
00:21:11,520 --> 00:21:14,397
Foundation started virtual cell

391
00:21:14,480 --> 00:21:19,277
everyone not just CI everyone has
a virtual cell model Dave Even if Dave

392
00:21:19,360 --> 00:21:22,797
doesn't even if Dave doesn't think Dave
has a virtual cell model I'm sure he'll

393
00:21:22,880 --> 00:21:26,237
spin one up soon. So I think that most

394
00:21:26,320 --> 00:21:31,117
So there's this famous uh uh paper
by Rich Sutton um on the bitter lesson in

395
00:21:31,200 --> 00:21:33,277
AI and the >> never heard of it.
Tell me all about the

396
00:21:33,360 --> 00:21:36,957
bitter lesson. This is a new concept
for me. What what is this you speak of?

397
00:21:37,040 --> 00:21:40,477
>> I don't know. You're probably joking
but pro maybe some some readers haven't

398
00:21:40,560 --> 00:21:43,757
heard of it. So the bitter lesson
is basically that human experts had all

399
00:21:43,840 --> 00:21:47,277
these really clever ideas and beautiful
theories but really what you needed to

400
00:21:47,360 --> 00:21:51,757
do was the simp like to make real
progress is the simplest method that you

401
00:21:51,840 --> 00:21:55,837
can come up with a large neural network
that you can train end to end as a

402
00:21:55,920 --> 00:21:58,957
general function approximator. The
simplest model you can come up with and

403
00:21:59,040 --> 00:22:01,677
then it's just a ton of data and compute

404
00:22:01,760 --> 00:22:06,077
to like actually train that model
and that usually outperforms at scale all

405
00:22:06,160 --> 00:22:10,237
the clever little hacks that human
experts had come up with uh before. And

406
00:22:10,320 --> 00:22:14,077
so the bitter lesson has worked for NLP
10, 20 years ago. You would have asked

407
00:22:14,160 --> 00:22:18,317
an NLP expert. Can you have one model
to have any conversation with you prompted

408
00:22:18,400 --> 00:22:22,317
any kind of question prompt engineering
I like invented and was nicely cited by

409
00:22:22,400 --> 00:22:27,677
the early GBT papers. But like the same
thing, the same state uh is biology is

410
00:22:27,760 --> 00:22:31,117
currently in. There's so many
complexities and the experts know so

411
00:22:31,200 --> 00:22:34,397
much. They feel like there's no way you
can instill all of that knowledge into

412
00:22:34,480 --> 00:22:36,957
one model. But if you have enough data,

413
00:22:37,040 --> 00:22:41,757
you can. And now you have companies like
Tahoe therapeutics and so on that are

414
00:22:41,840 --> 00:22:45,757
creating these massive perturbation
studies and yes all models are wrong but

415
00:22:45,840 --> 00:22:49,037
more and more of them will be useful
as we collect more and more data about

416
00:22:49,120 --> 00:22:53,277
biology and that is I think one
of the biggest driving factors uh for the

417
00:22:53,360 --> 00:22:56,957
impact of AI. >> But so ju
just to answer the question do

418
00:22:57,040 --> 00:23:01,117
you think virtual cell models are the
critical path to solving all disease or

419
00:23:01,200 --> 00:23:03,757
solving biology or is it something else?

420
00:23:03,840 --> 00:23:06,957
>> I I think they're definitely going
to play a crucial part. It's the third

421
00:23:07,040 --> 00:23:10,957
pillar of the Eureka machine. We do
need to have anything you can simulate,

422
00:23:11,040 --> 00:23:13,437
anything you can verify, AI will then be

423
00:23:13,520 --> 00:23:15,837
able to solve things in that domain. And

424
00:23:15,920 --> 00:23:21,117
so yes, uh I do think uh virtual models
uh and simulations uh are extremely

425
00:23:21,200 --> 00:23:24,717
important. >> I'm going
to jump into three fun

426
00:23:24,800 --> 00:23:26,557
breaking stories this week. Two on

427
00:23:26,640 --> 00:23:29,997
science and one on AI. Uh and Richard,

428
00:23:30,080 --> 00:23:32,237
they you know relate back to your work

429
00:23:32,320 --> 00:23:34,397
and your book. Uh our first story was

430
00:23:34,480 --> 00:23:37,517
published in MIT Tech Review today. Uh

431
00:23:37,600 --> 00:23:39,757
it tells the story of researchers at

432
00:23:39,840 --> 00:23:41,997
Israel's Weissman Institute who built a

433
00:23:42,080 --> 00:23:44,957
system called brain it. The system

434
00:23:45,040 --> 00:23:47,437
reconstructs the image a person is

435
00:23:47,520 --> 00:23:49,917
looking at while inside a functional MRI

436
00:23:50,000 --> 00:23:54,797
machine. I I'm just post put up
the slide here. So take a look at this

437
00:23:54,880 --> 00:23:58,397
image. You know it's pretty extraordinary.
On one side is the image

438
00:23:58,480 --> 00:24:00,637
that a person is looking at inside the

439
00:24:00,720 --> 00:24:03,757
fMRI machine. On the other is the AI

440
00:24:03,840 --> 00:24:05,757
reconstruction. You know, so earlier

441
00:24:05,840 --> 00:24:07,997
brain image decoders could tell you were

442
00:24:08,080 --> 00:24:11,277
looking at a dog or a clock tower, but

443
00:24:11,360 --> 00:24:13,117
it lost the color, the composition, and

444
00:24:13,200 --> 00:24:16,077
the detail. So brain it learns structure

445
00:24:16,160 --> 00:24:18,797
and meaning separately and then puts the

446
00:24:18,880 --> 00:24:20,877
picture back together. And here's the

447
00:24:20,960 --> 00:24:23,757
clever part. They also can run the model

448
00:24:23,840 --> 00:24:26,477
in reverse. An encoder that predicts how

449
00:24:26,560 --> 00:24:28,957
a brain will respond to an image. So

450
00:24:29,040 --> 00:24:30,957
they can feed it images no human has

451
00:24:31,040 --> 00:24:33,597
ever seen into the scanner uh generate

452
00:24:33,680 --> 00:24:36,637
predicted brain scans and then train on

453
00:24:36,720 --> 00:24:38,957
those. The model effectively builds its

454
00:24:39,040 --> 00:24:41,437
own data set. So Richard you know you

455
00:24:41,520 --> 00:24:43,677
say in your book that AI is superhuman

456
00:24:43,760 --> 00:24:47,277
in any domain you can simulate or verify

457
00:24:47,360 --> 00:24:50,637
uh and nowhere else. So is a brain

458
00:24:50,720 --> 00:24:52,317
simulatable do is the brain a

459
00:24:52,400 --> 00:24:54,557
simulatable domain or is this something

460
00:24:54,640 --> 00:24:57,997
different? It's not yet uh but this is

461
00:24:58,080 --> 00:25:02,077
still an incredible result. I still
remember the first such result came out

462
00:25:02,160 --> 00:25:05,677
when I was still a PhD student. I went
over to the bioinformatics department at

463
00:25:05,760 --> 00:25:09,997
Stanford and they you know it's like
very grainy little images we could like

464
00:25:10,080 --> 00:25:14,797
extract uh from this and to see this
fidelity now and this realism is just

465
00:25:14,880 --> 00:25:17,677
incredible. I think one of the problems,
I don't know if this study, I haven't

466
00:25:17,760 --> 00:25:21,037
seen it today yet, have been in the work,
but like um one of the problems is

467
00:25:21,120 --> 00:25:25,277
often that you have to train for each
brain because they're all slightly

468
00:25:25,360 --> 00:25:28,957
different. Uh and so you won't be able
to just take someone who've never put

469
00:25:29,040 --> 00:25:33,757
into an fMRI scanner to see and update
the model for that particular person,

470
00:25:33,840 --> 00:25:36,797
which would be otherwise amazing
for people who are like in a coma and you

471
00:25:36,880 --> 00:25:41,197
want to see are they still thinking stuff,
you know, like that would be incredible.

472
00:25:41,280 --> 00:25:42,957
>> Alex, tell us more
about this about this

473
00:25:43,040 --> 00:25:45,037
story today. Yeah. So this actually this

474
00:25:45,120 --> 00:25:48,317
is a result from March. Uh so this the

475
00:25:48,400 --> 00:25:51,277
broader field of doing decoding of brain

476
00:25:51,360 --> 00:25:54,637
states from functional imaging or EEG or

477
00:25:54,720 --> 00:25:56,557
electrodes. This has a long long history

478
00:25:56,640 --> 00:25:59,197
at this point. I remember 20 plus years

479
00:25:59,280 --> 00:26:01,677
ago at this point the the Galant lab at

480
00:26:01,760 --> 00:26:04,397
UC Berkeley was doing this to decode the

481
00:26:04,480 --> 00:26:07,277
the visual cortex of cats. Actually like

482
00:26:07,360 --> 00:26:11,917
20 plus years ago we were starting
to see images of how a cat perceives the

483
00:26:12,000 --> 00:26:15,117
world. And I remember like the first
grainy images that were coming out were

484
00:26:15,200 --> 00:26:17,677
like a cat seeing a branch. And then

485
00:26:17,760 --> 00:26:22,317
moving fast forwarding to non-invasive
fMRI. There are so many groups at this

486
00:26:22,400 --> 00:26:25,757
point. Meta in particular has been
sponsoring quite a bit of work in this

487
00:26:25,840 --> 00:26:30,717
area from Jean Marie King, many other
groups doing language decoding, vision

488
00:26:30,800 --> 00:26:34,237
decoding. More recently,
even fast forwarding to

489
00:26:34,320 --> 00:26:37,117
just the past 24 hours, Neurolink just

490
00:26:37,200 --> 00:26:38,877
announced their first scaling law

491
00:26:38,960 --> 00:26:41,997
studies on pre-training foundation

492
00:26:42,080 --> 00:26:45,997
models, Frontier models off of Neurolink

493
00:26:46,080 --> 00:26:48,637
electronic data, and this is on an

494
00:26:48,720 --> 00:26:50,957
individual patient basis, but they're

495
00:26:51,040 --> 00:26:54,237
capturing copious amounts of of data at

496
00:26:54,320 --> 00:26:56,237
high temporal resolution, much higher

497
00:26:56,320 --> 00:26:59,037
than fMRI can capture. fMRI like rule of

498
00:26:59,120 --> 00:27:01,517
thumb fMRI can give you with today's

499
00:27:01,600 --> 00:27:03,997
technology at best cubic millimeter

500
00:27:04,080 --> 00:27:06,717
spatial resolution for voxels and at

501
00:27:06,800 --> 00:27:09,037
best approximately 1 second temporal

502
00:27:09,120 --> 00:27:13,757
resolution. So that somewhat limits how
well you can do in terms of decoding a

503
00:27:13,840 --> 00:27:17,677
person's internal visual state.
can still make no mistake there are folks

504
00:27:17,760 --> 00:27:19,917
who have are making a cottage industry

505
00:27:20,000 --> 00:27:23,277
out of decoding dreams uh and decoding

506
00:27:23,360 --> 00:27:27,757
visually what a person perceives or what
they hallucinate in their visual cortex

507
00:27:27,840 --> 00:27:31,597
while they're sleeping or while they're
awake and I'm sure this is going to be a

508
00:27:31,680 --> 00:27:34,077
vibrant vibrant space training frontier

509
00:27:34,160 --> 00:27:36,237
models and foundation models off of

510
00:27:36,320 --> 00:27:39,197
copious amounts of fMRI data but

511
00:27:39,280 --> 00:27:41,597
ultimately where I think this has to go

512
00:27:41,680 --> 00:27:43,837
there are spatial and temporal limits to

513
00:27:43,920 --> 00:27:47,037
fMRI decoding It's going to require

514
00:27:47,120 --> 00:27:49,437
higher temporal precision and higher

515
00:27:49,520 --> 00:27:51,757
spatial precision to get where we really

516
00:27:51,840 --> 00:27:54,717
want to go, which is full dive VR and

517
00:27:54,800 --> 00:27:59,197
full bandwidth BCIs. And we will get
there, but the good news is you have to

518
00:27:59,280 --> 00:28:03,677
start somewhere. The large language
models had to start with GPT1, just a

519
00:28:03,760 --> 00:28:08,317
single artificial neuron that could
predict the polarity of Amazon reviews.

520
00:28:08,400 --> 00:28:10,637
Similarly here, if we're going to get

521
00:28:10,720 --> 00:28:13,117
full bandwidth BCIs, it's going to start

522
00:28:13,200 --> 00:28:15,917
with studies like Brain It and John

523
00:28:16,000 --> 00:28:19,037
Marie King and Jackalant. Amazing. Sem,

524
00:28:19,120 --> 00:28:23,437
what what happens in the world where
you can know someone's thoughts?

525
00:28:23,520 --> 00:28:25,597
>> Well, I think you you Let's be careful.

526
00:28:25,680 --> 00:28:29,357
We're decoding the perception here,
not the actual thought. Right. That's like

527
00:28:29,440 --> 00:28:31,757
quite a bit more comp. What's the other

528
00:28:31,840 --> 00:28:36,957
than the brain regionalization
without a difference?

529
00:28:37,040 --> 00:28:39,357
>> No. No. I think there
there is a difference.

530
00:28:39,440 --> 00:28:43,597
>> No, when when you imagine, right? When
you imagine something in your mind, it

531
00:28:43,680 --> 00:28:46,877
lights up the same neurons
as when you're seeing them.

532
00:28:46,960 --> 00:28:50,877
>> That that's fine. That that's fine.
Just I just want to make sure we don't mix

533
00:28:50,960 --> 00:28:54,397
the two because you can have thoughts
without necessarily the perceptions

534
00:28:54,480 --> 00:28:57,197
around it. But you can definitely have
one without the other and the other

535
00:28:57,280 --> 00:29:01,597
without the other. So uh but I think
for me the more interesting thing here is

536
00:29:01,680 --> 00:29:05,997
you know we've spent billions of hours
typing things into little rectangles and

537
00:29:06,080 --> 00:29:10,477
if we can change that interface to do
a more natural interface which is

538
00:29:10,560 --> 00:29:13,837
interface directly with the brain.
We used to talk about this in our

539
00:29:13,920 --> 00:29:17,837
singularity university lectures
on neuroscience is you know we still have

540
00:29:17,920 --> 00:29:21,037
very little idea how the brain works but
you don't need to have know how it works

541
00:29:21,120 --> 00:29:25,037
as long as you can interface effectively
with it and this I think gives us the

542
00:29:25,120 --> 00:29:29,517
opening to give us really deep
interfaces and then that'll help us

543
00:29:29,600 --> 00:29:31,917
figure out how the the brain works.
I >> I'll take the

544
00:29:32,000 --> 00:29:33,997
>> you bring up a really
Oh go ahead Richard. Sorry.

545
00:29:34,080 --> 00:29:37,917
>> Sorry. You bring up a really good point
which is I I read this result a while

546
00:29:38,000 --> 00:29:39,437
back that there are two types of people.

547
00:29:39,520 --> 00:29:44,157
Some people who think actually think
in sentences and other people when they

548
00:29:44,240 --> 00:29:47,917
think they just have a fuzzy thought
cloud and only once they're asked to

549
00:29:48,000 --> 00:29:49,437
verbalize their thought or they try to

550
00:29:49,520 --> 00:29:53,037
write it down they actually like make a

551
00:29:53,120 --> 00:29:56,957
real sentence out of their fuzzy thought
clouds. And I'm definitely a thought

552
00:29:57,040 --> 00:30:01,437
cloud thinker. My wife is very much
a like sentence like in her head. She

553
00:30:01,520 --> 00:30:05,597
actually thinks in actual sentences
and like there's no there's no like one is

554
00:30:05,680 --> 00:30:09,437
smarter than the other anything. there's
like they're like equivalent. But like I

555
00:30:09,520 --> 00:30:13,357
do think sometimes like writing
is thinking and like for me it is very

556
00:30:13,440 --> 00:30:17,037
helpful if I have to really verbalize
something versus I just have the

557
00:30:17,120 --> 00:30:20,557
thought. I'm just trying to think like
what does the world look like if the eye

558
00:30:20,640 --> 00:30:23,757
could really just like extract
my thoughts and now I need to like

559
00:30:23,840 --> 00:30:27,437
structure my thought clouds into
sentences more and more precise.

560
00:30:27,520 --> 00:30:30,877
>> Richard, what would neural net training
look like if you took all text out of

561
00:30:30,960 --> 00:30:35,357
it? No language at all. just train
on pure images and higher level thought

562
00:30:35,440 --> 00:30:39,117
constructs, you you'd get a very
different neural net out the other side

563
00:30:39,200 --> 00:30:42,477
and might actually think a lot more like
you do and less like your wife does. Cuz

564
00:30:42,560 --> 00:30:45,037
>> I'm not sure. >> Maybe
maybe we should be taking the

565
00:30:45,120 --> 00:30:51,400
position that anyone who doesn't have
an inner monologue is just a pea zombie

566
00:30:51,840 --> 00:30:53,357
>> there.
I got to throw one more thing in.

567
00:30:53,440 --> 00:30:56,957
I remember we had one of the top
linguists in the world come in and speak

568
00:30:57,040 --> 00:31:00,077
and we were like, you know, this must
be a bad time to be a linguist. nobody's

569
00:31:00,160 --> 00:31:04,077
using spelling or grammar or anything
else like that. And he goes, "No, on the

570
00:31:04,160 --> 00:31:06,717
contrary, this is one of the most
interesting times to be a linguist

571
00:31:06,800 --> 00:31:08,797
ever." And we're like, "Wow, how come?"

572
00:31:08,880 --> 00:31:12,477
And he said, "Cuz when you look
at emojis, it's the first time in the

573
00:31:12,560 --> 00:31:16,957
history of human language that you can
symbolically send emotion because you

574
00:31:17,040 --> 00:31:20,317
can digitally send emotions." And we're
like, "Wow, that's an interesting." So,

575
00:31:20,400 --> 00:31:24,237
he was super excited by opening up
that whole aperture. I thought that was an

576
00:31:24,320 --> 00:31:27,117
interesting >> You're
saying you're saying you're

577
00:31:27,200 --> 00:31:31,357
saying that emojis are the reason
why linguistics and not the fact that

578
00:31:31,440 --> 00:31:36,397
linguistics itself has been flourished by
>> I'm not connect I'm not connecting to

579
00:31:36,480 --> 00:31:38,637
you did that though it's not me >> Dave

580
00:31:38,720 --> 00:31:40,797
>> oh no the the result
I really am looking

581
00:31:40,880 --> 00:31:44,877
forward to is uh how does the brain

582
00:31:44,960 --> 00:31:49,437
>> train itself without gradient descent
everything going on in AI right now

583
00:31:49,520 --> 00:31:52,717
everything going on for the last
20 years in AI is driven by gradient

584
00:31:52,800 --> 00:31:55,757
descent algor algorithms and the
biologists can't find anything even

585
00:31:55,840 --> 00:32:00,637
vaguely like that in actual biology.
And so that that's you know I think with

586
00:32:00,720 --> 00:32:04,877
with really detailed imaging we might
finally crack the code on what is the

587
00:32:04,960 --> 00:32:09,437
fundamental learning algorithm that
changes the synaptic weights. It's not

588
00:32:09,520 --> 00:32:13,597
it's not what we use for artificial
neural nets. It's it's something

589
00:32:13,680 --> 00:32:16,877
different and if we discover that we
might find that neural net training can

590
00:32:16,960 --> 00:32:20,717
be 10 100 a thousand a million times
more efficient. And so hopefully that'll

591
00:32:20,800 --> 00:32:25,037
come out within a year. Could you define
gradient descent for our listeners?

592
00:32:25,120 --> 00:32:28,717
>> It's funny, Ilia Sutskiver uh when when
he's on interviews, he's like there is

593
00:32:28,800 --> 00:32:33,277
only one algorithm. The algorithm
is gradient descent. Everything else is

594
00:32:33,360 --> 00:32:35,597
just irrelevant compared to So what

595
00:32:35,680 --> 00:32:40,557
happens right now is you build these 96
or 120 layer deep neural nets and all it

596
00:32:40,640 --> 00:32:43,597
is is a whole bunch of random
connections that do absolutely nothing

597
00:32:43,680 --> 00:32:45,997
useful and then you give it trillions of

598
00:32:46,080 --> 00:32:50,637
15 trillion training examples. You're
like, "When you see this, say that or

599
00:32:50,720 --> 00:32:52,877
when you see this paragraph, this is the

600
00:32:52,960 --> 00:32:55,517
next token." And if you're guessing

601
00:32:55,600 --> 00:32:57,677
wrong, you get punished through gradient

602
00:32:57,760 --> 00:33:02,877
descent. So the error or the how far off
you are is your punishment and it passes

603
00:33:02,960 --> 00:33:06,237
back through all the layers and it
calculates an error and it blames each

604
00:33:06,320 --> 00:33:09,037
neuron or each connection for how how

605
00:33:09,120 --> 00:33:11,037
responsible were you for this terrible

606
00:33:11,120 --> 00:33:16,077
answer. And if you're way wrong, you get
moved and you move in the direction that

607
00:33:16,160 --> 00:33:20,077
helps you get the answer right. So it's
this incredibly laborious search process

608
00:33:20,160 --> 00:33:24,237
and that movement of each synaptic
connection is a gradient and you just

609
00:33:24,320 --> 00:33:26,797
you just hill climb to the best possible

610
00:33:26,880 --> 00:33:29,597
state and then it magically after about

611
00:33:29,680 --> 00:33:34,317
hund00 million of comput in Richard's
case I guess 400 million uh it magically

612
00:33:34,400 --> 00:33:37,917
starts thinking through this one
algorithm gradient descent.

613
00:33:38,000 --> 00:33:41,757
>> Yeah I think it's really back back prop
that we're talking about syndictoically

614
00:33:41,840 --> 00:33:45,677
rather than gradient descent would be
the the obvious comment. So maybe yeah

615
00:33:45,760 --> 00:33:49,357
back prop is I guess the efficient
computation of those gradients. Uh maybe

616
00:33:49,440 --> 00:33:54,397
I'll I'll add to um add to this
explanation from Dave but like gradient

617
00:33:54,480 --> 00:33:57,917
is essentially an optimization algorithm
that we use to train all kinds of

618
00:33:58,000 --> 00:34:01,597
machine learning models by minimizing
the errors. And I think the best analogy

619
00:34:01,680 --> 00:34:04,477
is to assume you're a blind hiker.

620
00:34:04,560 --> 00:34:07,277
You're the top of the mountain
and you're trying to find the lowest the

621
00:34:07,360 --> 00:34:09,277
lowest point and you can't really see

622
00:34:09,360 --> 00:34:11,197
that far. You're blind but you can feel

623
00:34:11,280 --> 00:34:14,637
the slope at each step. And so how much

624
00:34:14,720 --> 00:34:18,397
like how big of a step should you take
down when you feel like the slope is

625
00:34:18,480 --> 00:34:19,597
going roughly in the right direction.

626
00:34:19,680 --> 00:34:23,757
And then the problem is that the kinds
of landscapes these AIs are trying to

627
00:34:23,840 --> 00:34:29,197
optimize are highly non-convex which
means there's not just one lowest point

628
00:34:29,280 --> 00:34:31,757
uh but there are many different low
points. And if you depending on where

629
00:34:31,840 --> 00:34:35,197
you start you might go into a different
valley depending on which mountain you

630
00:34:35,280 --> 00:34:38,637
start you go into this valley versus
this this different valley. And there

631
00:34:38,720 --> 00:34:42,397
are actually a lot of really interesting
analogies. My friend Jihan uh like

632
00:34:42,480 --> 00:34:45,757
thinks about this from the psychological
perspective. For instance, when you have

633
00:34:45,840 --> 00:34:48,877
PTSD and you train, you overfit to one

634
00:34:48,960 --> 00:34:53,037
algorithm. how your brain is kind
of stuck in one valley of how to think

635
00:34:53,120 --> 00:34:57,677
about something. And it takes sometimes
people like doing psychedelics and stuff

636
00:34:57,760 --> 00:34:59,677
uh that have shown to to help help with

637
00:34:59,760 --> 00:35:02,157
like depression and PTSD and other uh

638
00:35:02,240 --> 00:35:06,717
mental diseases like um they they it
helps to increase the learning rate to

639
00:35:06,800 --> 00:35:10,717
jump over another mountain and and go
into a different valley of attraction

640
00:35:10,800 --> 00:35:14,877
where you then have new kinds of ways
of thinking. And so this blind hiker

641
00:35:14,960 --> 00:35:18,237
analogy I think is a really intuitive
way to think about it. And then of

642
00:35:18,320 --> 00:35:22,877
course the problem is it's not just in 3D,
it's in a trilliond like a trillion

643
00:35:22,960 --> 00:35:26,397
parameters that you're trying
to traverse in that landscape. But the

644
00:35:26,480 --> 00:35:30,237
ideas of like try to identify the
direction and take steps towards it still.

645
00:35:30,320 --> 00:35:32,557
>> And you know what else is really
incredible? I could grip on this for

646
00:35:32,640 --> 00:35:34,317
hours, but if you look at a big model

647
00:35:34,400 --> 00:35:38,157
like Kimmy K3, it's 93 layers and it's

648
00:35:38,240 --> 00:35:42,557
trained trained trained. If you take
a single layer and you randomize it and

649
00:35:42,640 --> 00:35:44,557
then you train it and say find yourself

650
00:35:44,640 --> 00:35:48,077
again, it can't find itself again. It

651
00:35:48,160 --> 00:35:51,837
never gets back to where it was. Which
comes back to in these fMRIs when you're

652
00:35:51,920 --> 00:35:55,037
looking at a human brain and you're
saying, "Okay, here's Alex's brain. Can

653
00:35:55,120 --> 00:35:58,797
I compare that to Salem's?" And you're
like, "Wow, there's nothing in common

654
00:35:58,880 --> 00:36:01,997
going on here yet." you know, well,

655
00:36:02,080 --> 00:36:06,840
okay, maybe that's a bad example, but but

656
00:36:08,880 --> 00:36:12,957
the way your brain wires as you learn
is completely unique to you. Yet, yet it

657
00:36:13,040 --> 00:36:16,477
can come out with like, okay, we're
equally good soccer players, but the way

658
00:36:16,560 --> 00:36:20,477
the signals are propagating through
our networks is completely different and

659
00:36:20,560 --> 00:36:24,397
unique to each of us, which is really
really strange, you know, that that it

660
00:36:24,480 --> 00:36:28,717
can't it can't find its own way back
to its its state. I think we're going to

661
00:36:28,800 --> 00:36:33,437
learn a lot actually with the fMRI data
coming in on how and why that works. And

662
00:36:33,520 --> 00:36:36,957
there there have to be commonalities
that we just can't find. You know,

663
00:36:37,040 --> 00:36:39,517
rotations, you know, make everything
look different, but they're really not

664
00:36:39,600 --> 00:36:43,357
that different. If you if you put them
through a transform, like a 4A transform

665
00:36:43,440 --> 00:36:46,397
view will suddenly say, "Oh my god,
this is why they line up."

666
00:36:46,480 --> 00:36:48,077
>> Yeah. I was I was just with my

667
00:36:48,160 --> 00:36:51,117
fraternity brother uh of ours, Dave,

668
00:36:51,200 --> 00:36:53,837
who's the head of neurobiology at at USC

669
00:36:53,920 --> 00:36:59,117
the other day. um and is you know this
is the most exciting time ever for brain

670
00:36:59,200 --> 00:37:02,797
science >> right I mean
we're you know the brain

671
00:37:02,880 --> 00:37:06,797
has been a black box for for since
humanity began and our ability to

672
00:37:06,880 --> 00:37:08,317
understand

673
00:37:08,400 --> 00:37:12,157
uh the brain and deal with with uh you

674
00:37:12,240 --> 00:37:16,557
know mental disease I think is going to be
extraordinary. Yeah. Well, so you can

675
00:37:16,640 --> 00:37:20,077
simulate. This is why liquid AI
was founded actually because we completely

676
00:37:20,160 --> 00:37:22,397
reverse engineered the the C elegance

677
00:37:22,480 --> 00:37:26,237
worm brain down to exactly every single

678
00:37:26,320 --> 00:37:30,557
thing going on and then they were
able to simulate it and then after they

679
00:37:30,640 --> 00:37:33,997
simulated it they realized wait this
is a very efficient neural net and then

680
00:37:34,080 --> 00:37:38,077
they productized it. I mean that's
so the both ang so go ahead Richard

681
00:37:38,160 --> 00:37:40,877
>> there's a really cool thing since you
mentioned se against worms there's not

682
00:37:40,960 --> 00:37:43,197
many people that bring that up
a friend of mine who's a professor of

683
00:37:43,280 --> 00:37:47,037
neuroscience at Harvard Sam Gershman uh
he one of the most brilliant people I've

684
00:37:47,120 --> 00:37:49,597
ever met like he actually did a a study

685
00:37:49,680 --> 00:37:54,877
with with worms where he cut them
in half and one segance worms they do have

686
00:37:54,960 --> 00:37:58,637
a brain right and then they have a lot
of other like neurons in the rest now

687
00:37:58,720 --> 00:38:01,117
you can train them to react to a certain

688
00:38:01,200 --> 00:38:03,837
stimuli um and then you split them in

689
00:38:03,920 --> 00:38:08,477
half the the second half without the brain
regrows a new brain. And here, get

690
00:38:08,560 --> 00:38:12,077
this, this is crazy.
That new brain has the same memories.

691
00:38:12,160 --> 00:38:15,037
>> Wow. >> And reacts
to different stimuli. And so

692
00:38:15,120 --> 00:38:18,717
he's like, maybe there's a different way
where we learn. And I do think there are

693
00:38:18,800 --> 00:38:21,917
things that right now we're very much
stuck in this like it's a neural net,

694
00:38:22,000 --> 00:38:23,837
it's all electrical signals and so on.

695
00:38:23,920 --> 00:38:27,837
But clearly like brain chemistry can
change massively with just like you can

696
00:38:27,920 --> 00:38:31,917
get hangry, you can be in pain, you can
have like a certain like stimulant and

697
00:38:32,000 --> 00:38:35,357
all of a sudden your brain is very
different. And I think there are and

698
00:38:35,440 --> 00:38:37,517
I've seen this now with uh like uh

699
00:38:37,600 --> 00:38:42,477
having babies and and like talking to uh
moms like there's certain algorithms

700
00:38:42,560 --> 00:38:45,757
where you all of a sudden you nest
because you had a baby and now you want

701
00:38:45,840 --> 00:38:50,077
to like and how you nest is different
but people nest you know in one form or

702
00:38:50,160 --> 00:38:51,757
another and so there are these latent

703
00:38:51,840 --> 00:38:53,597
algorithms that we have that are in our

704
00:38:53,680 --> 00:38:57,357
DNA that just trigger after like 20

705
00:38:57,440 --> 00:39:00,717
something years right when just the right
things then happen with your body

706
00:39:00,800 --> 00:39:04,557
and your biology. So, I think there's
still so much that we don't we haven't

707
00:39:04,640 --> 00:39:08,157
>> so much complexity. I'm going to move
us to our next story from three days ago.

708
00:39:08,240 --> 00:39:12,957
So, on Tuesday, uh the president signed
an executive order directing the EPA,

709
00:39:13,040 --> 00:39:17,357
the Department of the Interior
and Agriculture working with HHS to cut

710
00:39:17,440 --> 00:39:19,517
invasive mosquito populations in

711
00:39:19,600 --> 00:39:23,117
Washington DC by at least 90% and the

712
00:39:23,200 --> 00:39:27,117
tick population by at least 50% by 2028.

713
00:39:27,200 --> 00:39:31,197
Uh, the targets include mosquito species
that spread deni, zika, and yellow

714
00:39:31,280 --> 00:39:33,677
fever. And here's what caught my eye.

715
00:39:33,760 --> 00:39:38,317
You know, the order explicitly
prioritizes sterile insect techniques,

716
00:39:38,400 --> 00:39:42,477
safe genetic modifications,
and beneficial bacteria over conventional

717
00:39:42,560 --> 00:39:46,557
pesticides, right? I love that.
You know, this is the exact kind of uh gan

718
00:39:46,640 --> 00:39:49,997
drive work that Colossal,
the deextinction company, is doing. You

719
00:39:50,080 --> 00:39:53,197
know, I think most people don't realize
it's like, you know, when I was raising

720
00:39:53,280 --> 00:39:56,557
my kids, I would ask them a question,
who which which is the species that

721
00:39:56,640 --> 00:40:00,557
kills the most humans on the planet, you
know, and some people jump onto sharks?

722
00:40:00,640 --> 00:40:04,557
Some people jump onto, you know,
whatever it might be. But, uh, on this

723
00:40:04,640 --> 00:40:06,237
part, I'm sure people know mosquitoes

724
00:40:06,320 --> 00:40:08,957
are the deadliest life form on Earth,

725
00:40:09,040 --> 00:40:11,757
>> right? Malaria alone, uh, kills half a

726
00:40:11,840 --> 00:40:15,597
million people a year. So, uh, Alex, u,

727
00:40:15,680 --> 00:40:18,797
this is biology as engineering arriving

728
00:40:18,880 --> 00:40:21,597
as federal policy. Uh, your thoughts on

729
00:40:21,680 --> 00:40:25,837
this story? For decades, we've been
scared of our shadow. So, I have a

730
00:40:25,920 --> 00:40:28,157
classmate Kevin Esfelt at MIT who was

731
00:40:28,240 --> 00:40:29,997
the pioneer of the GAN drive, one of the

732
00:40:30,080 --> 00:40:33,277
pioneers, and he's had a devil of a time

733
00:40:33,360 --> 00:40:37,437
getting states, municipalities
to approve gan drive studies against

734
00:40:37,520 --> 00:40:39,677
mosquitoes. And for those not tracking,

735
00:40:39,760 --> 00:40:41,517
the the premise of this technique is

736
00:40:41,600 --> 00:40:44,637
basically inserting a gene via crisper

737
00:40:44,720 --> 00:40:46,957
that wants to replicate itself to

738
00:40:47,040 --> 00:40:49,917
basically sterilize mosquitoes through

739
00:40:50,000 --> 00:40:54,717
propagating virally through the mosquito
population. We have the technology to do

740
00:40:54,800 --> 00:40:59,437
this. What it has lacked, at least
in this country, is a federal mandate to

741
00:40:59,520 --> 00:41:03,437
actually go and implement large-scale
genetic engineering of non-human animal

742
00:41:03,520 --> 00:41:05,837
populations. And now for the first time

743
00:41:05,920 --> 00:41:09,437
it seems we're seeing via executive fiat

744
00:41:09,520 --> 00:41:11,917
just that mandate. And it is so exciting

745
00:41:12,000 --> 00:41:14,957
for two reasons. One because one is

746
00:41:15,040 --> 00:41:18,237
Peter you mentioned malaria and other

747
00:41:18,320 --> 00:41:21,517
non-human animalborn diseases cause an

748
00:41:21,600 --> 00:41:25,517
an enormous amount of suffering human
suffering and also non-human animal

749
00:41:25,600 --> 00:41:28,877
suffering. But secondly because there's
a way to do this that doesn't actually

750
00:41:28,960 --> 00:41:30,637
involve killing the animals themselves.

751
00:41:30,720 --> 00:41:34,797
Historically, historically,
you know, if you look back

752
00:41:34,880 --> 00:41:37,437
50 years, you'd see chemical spraying if

753
00:41:37,520 --> 00:41:39,357
you wanted to do something about say

754
00:41:39,440 --> 00:41:42,957
insect born illness

755
00:41:43,040 --> 00:41:46,957
uh responses. We don't need to do
that anymore. We can actually keep the

756
00:41:47,040 --> 00:41:48,717
insects that are the inadvertent

757
00:41:48,800 --> 00:41:51,997
carriers of bacterial or viral disease.

758
00:41:52,080 --> 00:41:55,917
We can actually preserve their lives
while also preventing the disease

759
00:41:56,000 --> 00:41:59,917
transmission. And I think that's good
from their perspective as well. That was

760
00:42:00,000 --> 00:42:04,317
the part that got me most excited was we
spray things, we're basically poisoning

761
00:42:04,400 --> 00:42:07,757
biology now we actually
>> ecosystem, right?

762
00:42:07,840 --> 00:42:09,357
>> And leave the poison ecosystem.

763
00:42:09,440 --> 00:42:13,677
>> Yeah. Yeah. >> Uh Richard,
your thoughts on this story?

764
00:42:13,760 --> 00:42:15,517
>> I I concur with everyone like it's it's

765
00:42:15,600 --> 00:42:18,717
it's wonderful. Uh and it's so

766
00:42:18,800 --> 00:42:23,757
interesting like people I I hope we
can amplify these these kinds of stories

767
00:42:23,840 --> 00:42:27,517
more, you know, like the future needs
better marketing. We need more Peters in

768
00:42:27,600 --> 00:42:31,837
the world. Um and and this is one
of those many stories, you know, curing

769
00:42:31,920 --> 00:42:36,237
various diseases, peril bio, like saving
animal lives and and prevent them from

770
00:42:36,320 --> 00:42:40,477
just being bred uh to be tested upon
and dissected like uh those are all like

771
00:42:40,560 --> 00:42:44,157
stories we need to amplify more in the
public eye. Yeah, Dave, I can't wait for

772
00:42:44,240 --> 00:42:49,117
this to come from DC to Vermont
and Massachusetts and all of these

773
00:42:49,200 --> 00:42:53,117
>> I was going to say for the listeners
that are interested in your real estate

774
00:42:53,200 --> 00:42:55,117
fund, which you know the the theme there

775
00:42:55,200 --> 00:42:57,917
is, hey, easy access via drone to

776
00:42:58,000 --> 00:43:02,717
hilltops, islands, but also if you
live on the edge of a marsh, your house

777
00:43:02,800 --> 00:43:06,157
probably sells for about half the price
of a place that's not on the edge of a

778
00:43:06,240 --> 00:43:08,797
marsh, that'll go away. there there's so

779
00:43:08,880 --> 00:43:11,757
many solutions coming to biting insects,

780
00:43:11,840 --> 00:43:15,837
this being one of them. Uh that that also
is is, you know, if it's beautiful,

781
00:43:15,920 --> 00:43:18,637
beautiful land that otherwise is very
difficult to live on, it's going to go

782
00:43:18,720 --> 00:43:21,277
through the roof in value.
>> I do think it's really important to try

783
00:43:21,360 --> 00:43:25,677
to like, you know, keep the bees uh
like alive, all right, and not have spray

784
00:43:25,760 --> 00:43:29,197
bunch of pesticides, make sure birds
still have enough like insects to eat

785
00:43:29,280 --> 00:43:33,117
and things like that. Yeah,
>> I was going to say for maybe up to 80

786
00:43:33,200 --> 00:43:35,277
years, humanity has been scared of its

787
00:43:35,360 --> 00:43:40,397
own shadow. humanity in general,
America or the West in particular, nuclear

788
00:43:40,480 --> 00:43:44,797
energy. Another example I've pointed
out on the pod where we arguably lost 50

789
00:43:44,880 --> 00:43:47,037
plus years of progress because we were

790
00:43:47,120 --> 00:43:50,317
too scared of either the bomb or fision

791
00:43:50,400 --> 00:43:52,877
reactors or some an entire generation

792
00:43:52,960 --> 00:43:54,957
watched the movie China Syndrome and

793
00:43:55,040 --> 00:43:57,597
then got scared of nuclear power

794
00:43:57,680 --> 00:44:01,357
unnecessarily. Same idea with this or or

795
00:44:01,440 --> 00:44:05,437
with geoengineering. Another example,
the the world is still scared. Some

796
00:44:05,520 --> 00:44:08,477
fraction of the world is scared
of engineering the weather. We don't need

797
00:44:08,560 --> 00:44:10,397
to be scared of engineering our physical

798
00:44:10,480 --> 00:44:13,197
world or our biological world. And now

799
00:44:13,280 --> 00:44:17,437
hopefully one can see green shoots
of humanity getting past that stage of

800
00:44:17,520 --> 00:44:19,197
worrying about its own shadow.

801
00:44:19,280 --> 00:44:23,597
>> This is interesting that Sorry. Yeah,
it's really it's interesting indeed that

802
00:44:23,680 --> 00:44:26,717
like sometimes alarmists feel like
they're doing the right thing by just

803
00:44:26,800 --> 00:44:30,317
like oh how bad could it be? I'm warning
people of something bad. But like it can

804
00:44:30,400 --> 00:44:33,757
indeed push all of humanity away
from something really good like energy

805
00:44:33,840 --> 00:44:36,397
abundance uh with nuclear. Um I think

806
00:44:36,480 --> 00:44:40,877
overpopulation is when these other big
myths where people are like well if we

807
00:44:40,960 --> 00:44:43,357
have too many people there's going
to be scarcity of all these different

808
00:44:43,440 --> 00:44:46,877
resources and everything is going
to get more expensive and people are more

809
00:44:46,960 --> 00:44:50,157
poverty and so on and the exact opposite
happened. There's actually one website

810
00:44:50,240 --> 00:44:54,397
that I think you all would love which
called humanprogress.org or.com like

811
00:44:54,480 --> 00:44:57,117
that that shows that a lot of these
things are actually getting cheaper and

812
00:44:57,200 --> 00:44:59,037
cheaper despite there being more people.

813
00:44:59,120 --> 00:45:01,517
>> Yeah. >> Well, hey Richard,
you're you're a hero

814
00:45:01,600 --> 00:45:05,837
in Germany like like can't walk down the
street kind of hero in Germany. What do

815
00:45:05,920 --> 00:45:09,917
you think about the fact that Germany
has no power for exactly this reason?

816
00:45:10,000 --> 00:45:14,157
>> It is really unfortunate if there are
a lot of people in Germany that want to in

817
00:45:14,240 --> 00:45:17,917
a weird way sort of offramp from progress.
Um and they think that

818
00:45:18,000 --> 00:45:21,277
everything that consumes power is bad
for the environment. there's like a

819
00:45:21,360 --> 00:45:24,797
weird sort of degrowth uh offshoot from

820
00:45:24,880 --> 00:45:27,517
generally like well-intentioned pro-

821
00:45:27,600 --> 00:45:31,437
like environmental uh vibes that uh just

822
00:45:31,520 --> 00:45:35,437
worry me um worry me quite a bit. Yeah,
>> we're going to talk about dumerism in a

823
00:45:35,520 --> 00:45:39,437
little bit uh but not yet. And you know,
one of the things I realized a long time

824
00:45:39,520 --> 00:45:42,317
ago is we humans are really incredible

825
00:45:42,400 --> 00:45:44,717
at seeing a problem out in the future,

826
00:45:44,800 --> 00:45:47,277
right? We see acid rain, we see

827
00:45:47,360 --> 00:45:52,397
overpopulation, we see, you know, energy
shortage, whatever it might be. And then

828
00:45:52,480 --> 00:45:56,077
because of our amigdula, because the way
we think, we accelerate that future

829
00:45:56,160 --> 00:46:00,957
problem to today. And we freak out
and we forget the fact that there is a

830
00:46:01,040 --> 00:46:05,437
decade worth of progress we're going
to make by the time we reach that problem.

831
00:46:05,520 --> 00:46:08,557
And that's what entrepreneurs do.
They solve problem after problem after

832
00:46:08,640 --> 00:46:11,677
problem. And I I guess I just want our

833
00:46:11,760 --> 00:46:15,997
listeners to hear that because if
you're worried about some future problem,

834
00:46:16,080 --> 00:46:18,637
please understand there's a incredibly

835
00:46:18,720 --> 00:46:21,997
efficient market uh of entrepreneurship

836
00:46:22,080 --> 00:46:25,677
and capitalism that will solve
the world's biggest problems, the world's

837
00:46:25,760 --> 00:46:27,437
biggest business opportunities, right?

838
00:46:27,520 --> 00:46:31,117
And it's a beautiful uh forward uh you

839
00:46:31,200 --> 00:46:33,997
know uh uh forward propagation solution

840
00:46:34,080 --> 00:46:38,477
set that we have. And maybe we'll get
there later, but I I see the same thing

841
00:46:38,560 --> 00:46:42,477
happen in AI dumerism where they create
more and more complex scenarios where

842
00:46:42,560 --> 00:46:46,957
attackers get these like near magical
abilities to attack, but somehow the

843
00:46:47,040 --> 00:46:50,797
defenders in those stories never get
near magical abilities to defend. And

844
00:46:50,880 --> 00:46:54,797
it's this weird thing of like, man,
if I can create a super virus that is like

845
00:46:54,880 --> 00:46:59,517
perfectly undetectable and spreads all
throughout the world and no one notices

846
00:46:59,600 --> 00:47:03,117
it and like and then it has this Wi-Fi
switch and you just turn it on and off

847
00:47:03,200 --> 00:47:06,957
and kill people like I I will create
a super magical vaccine that just

848
00:47:07,040 --> 00:47:10,477
inoculates you against all of these
things. Like if you can actually assume

849
00:47:10,560 --> 00:47:13,997
make that assumption then like
it's a weird Yeah. weird thing.

850
00:47:14,080 --> 00:47:18,237
>> This episode is sponsored by Google for
startups. Think about this for a second.

851
00:47:18,320 --> 00:47:22,557
You now have access to the same
generative AI models that cost hundreds

852
00:47:22,640 --> 00:47:24,317
of millions of dollars to train.

853
00:47:24,400 --> 00:47:28,557
Google's startup technical guide
for generative media gives you complete

854
00:47:28,640 --> 00:47:31,117
blueprint for deploying Google DeepMinds

855
00:47:31,200 --> 00:47:33,837
models in production. Images, video,

856
00:47:33,920 --> 00:47:36,077
audio, all of it. Real architecture,

857
00:47:36,160 --> 00:47:38,557
real results. Find the link in the show

858
00:47:38,640 --> 00:47:43,117
notes below. I'm going
to move us to a story that's

859
00:47:43,200 --> 00:47:45,037
been sort of breaking the internet over

860
00:47:45,120 --> 00:47:49,197
the last 24 hours. Um, and it it puts

861
00:47:49,280 --> 00:47:51,517
your regulate the application argument

862
00:47:51,600 --> 00:47:53,677
to the test, Richard. So, a company

863
00:47:53,760 --> 00:47:56,637
called Tavis unveiled Griffin, uh, which

864
00:47:56,720 --> 00:47:59,517
it calls the first model ever to pass

865
00:47:59,600 --> 00:48:02,397
the video touring test. So, here's the

866
00:48:02,480 --> 00:48:05,117
numbers. 48% of people who talk to an AI

867
00:48:05,200 --> 00:48:08,397
avatar on a live face-to-face video

868
00:48:08,480 --> 00:48:09,757
thought they were talking to a real

869
00:48:09,840 --> 00:48:13,037
human. Previous systems were at 3%. So,

870
00:48:13,120 --> 00:48:15,597
let me show the video here. And uh it's

871
00:48:15,680 --> 00:48:18,717
super cool and let's talk about it next.

872
00:48:18,800 --> 00:48:22,077
>> Now, how about just a thumbs up?

873
00:48:22,160 --> 00:48:23,917
>> Sure. One thumbs up coming right up.

874
00:48:24,000 --> 00:48:27,917
>> Hell yeah. That was awesome. I would
say you're my favorite coworker. It's been

875
00:48:28,000 --> 00:48:30,237
It's been really great
toh work on this with you.

876
00:48:30,320 --> 00:48:34,077
>> You're making me blush.
A Who's that with the ugly sweater?

877
00:48:34,160 --> 00:48:35,837
>> Simon says, "Touch your hair."

878
00:48:35,920 --> 00:48:37,277
>> Is this what you're looking for?

879
00:48:37,360 --> 00:48:41,437
>> Yeah. Okay, nice. Simon says point. All

880
00:48:41,520 --> 00:48:44,317
right, I'm doing it.

881
00:48:44,400 --> 00:48:47,837
Now, touch your chin.

882
00:48:47,920 --> 00:48:51,757
>> I'm not going to fall for that.
You didn't say the magic word. Uh,

883
00:48:51,840 --> 00:48:55,197
>> you got me. Okay.
>> The loose connector back onto the

884
00:48:55,280 --> 00:49:00,397
motherboard. >> Okay,
let me go let me go grab that. But

885
00:49:00,480 --> 00:49:03,917
beforehand, yeah, I turned
the soldering iron on. I

886
00:49:04,000 --> 00:49:06,397
think it's going to take about 12 seconds.
So, maybe just let me know when

887
00:49:06,480 --> 00:49:10,317
that's done. >> It sounds good.
I'll keep an eye on it

888
00:49:10,400 --> 00:49:12,797
and let you know the second it hits that

889
00:49:12,880 --> 00:49:17,757
12 second mark. >> All right.
So, uh, interesting. So,

890
00:49:17,840 --> 00:49:22,077
Tavis calls it the first human
interaction model. It's fully duplex,

891
00:49:22,160 --> 00:49:26,077
listening and talking at the same time
that you do, the way humans actually do.

892
00:49:26,160 --> 00:49:28,397
It's number one on Nvidia's benchmark

893
00:49:28,480 --> 00:49:31,517
for full duplex AI video. Tavis pitches

894
00:49:31,600 --> 00:49:33,917
it as quote a tutor for every student

895
00:49:34,000 --> 00:49:36,637
that notices when they're lost, an elder

896
00:49:36,720 --> 00:49:41,517
care companion that listens. Uh Tavis
itself says Griffin requires safety work

897
00:49:41,600 --> 00:49:42,877
before it can be released publicly

898
00:49:42,960 --> 00:49:45,597
because it's the first model that can be

899
00:49:45,680 --> 00:49:49,517
mistaken for a real person. Um Salem,

900
00:49:49,600 --> 00:49:51,917
your thoughts on this one?

901
00:49:52,000 --> 00:49:56,877
So, I'm less kind of um floored by the

902
00:49:56,960 --> 00:50:00,317
um the imagery and the the mimicry of

903
00:50:00,400 --> 00:50:03,917
it. Uh this was expected to happen. For

904
00:50:04,000 --> 00:50:07,837
me, I'm more interested in can it
accomplish something useful enough that

905
00:50:07,920 --> 00:50:09,997
I don't care whether it's human or not.

906
00:50:10,080 --> 00:50:14,317
And I think that that is more
interesting uh which might come along

907
00:50:14,400 --> 00:50:16,557
and have hap we have that happen at some

908
00:50:16,640 --> 00:50:21,437
point. We've talked about the idea that
in the near future, like now, uh you're

909
00:50:21,520 --> 00:50:23,437
going to have digital co-workers that

910
00:50:23,520 --> 00:50:26,077
pop up on Zoom, that you call, that you

911
00:50:26,160 --> 00:50:29,117
text, you Slack, and interact with you.

912
00:50:29,200 --> 00:50:32,557
And I think it's pretty compelling as a

913
00:50:32,640 --> 00:50:37,357
mechanism for a future, you know,
virtualized company. Uh Alex,

914
00:50:37,440 --> 00:50:41,037
>> well, first, Peter, you should
probably ask me to touch my face.

915
00:50:41,120 --> 00:50:44,237
>> Okay, touch your face.
Simon says touch your face.

916
00:50:44,320 --> 00:50:47,597
>> Oh, >> okay. Very good.
Just checking but I I

917
00:50:47,680 --> 00:50:51,037
guess Tavis is ahead of me. I I I think

918
00:50:51,120 --> 00:50:54,877
on the one hand if you look at models
coming out of Chinese labs like

919
00:50:54,960 --> 00:50:57,837
Alibaba's lab, Juan Streamer gave us a

920
00:50:57,920 --> 00:51:01,917
preview of what was going to happen.
The Chinese labs remain overinvested

921
00:51:02,000 --> 00:51:05,837
relative to the Western labs
in generative video models and interactive

922
00:51:05,920 --> 00:51:08,397
generative video models. So one streamer

923
00:51:08,480 --> 00:51:10,637
I think is a preview of what's possible.

924
00:51:10,720 --> 00:51:13,757
My guess is I I I I I read the the

925
00:51:13,840 --> 00:51:19,517
information that was put out around Tavis.
I I think in full generality, let

926
00:51:19,600 --> 00:51:22,157
let's just talk about where this
is going to end up. It's it's very

927
00:51:22,240 --> 00:51:25,917
difficult to predict the short term.
I think it's pretty easy to predict the

928
00:51:26,000 --> 00:51:29,917
long term. In the long term, endto-end
generative pixels. We'll have these

929
00:51:30,000 --> 00:51:32,797
magic mirrors that could be real time

930
00:51:32,880 --> 00:51:35,837
fully interactive pixel-wise generated

931
00:51:35,920 --> 00:51:39,917
or if if anthropic has its way,
vector-wise or procedurally generated.

932
00:51:40,000 --> 00:51:42,237
But either way, fully generated

933
00:51:42,320 --> 00:51:46,077
real-time interactive video models and

934
00:51:46,160 --> 00:51:49,357
you'll be able to create a scene that

935
00:51:49,440 --> 00:51:54,237
consists of people like it's possible
right now, but the latency is high. You

936
00:51:54,320 --> 00:51:55,997
see with one streamer, which is already

937
00:51:56,080 --> 00:51:59,197
out and already open source, or with

938
00:51:59,280 --> 00:52:02,477
this Tavis Griffin type model, which is

939
00:52:02,560 --> 00:52:05,037
not really out yet and definitely not

940
00:52:05,120 --> 00:52:07,917
open source. You see a preview of the

941
00:52:08,000 --> 00:52:11,357
future. What does this look like? Well,

942
00:52:11,440 --> 00:52:13,117
there are a few different ways it can

943
00:52:13,200 --> 00:52:17,517
go. Query how revenue generating per

944
00:52:17,600 --> 00:52:20,957
token it is. I is it anywhere close to

945
00:52:21,040 --> 00:52:23,597
the optimal frontier of code generation?

946
00:52:23,680 --> 00:52:25,757
Doubt it. On the other hand, if it

947
00:52:25,840 --> 00:52:28,477
becomes so absurdly inexpensive to be

948
00:52:28,560 --> 00:52:30,957
able to to generate arbitrary, say,

949
00:52:31,040 --> 00:52:33,357
humans participating in a Zoom meeting

950
00:52:33,440 --> 00:52:36,237
or humans participating in a podcast. Do

951
00:52:36,320 --> 00:52:39,197
we really care whether it's close to to

952
00:52:39,280 --> 00:52:41,437
being near the optimal cost performance

953
00:52:41,520 --> 00:52:43,917
frontier? Maybe not. There are a lot of

954
00:52:44,000 --> 00:52:46,717
human service industry jobs that require

955
00:52:46,800 --> 00:52:49,437
a face and require interactivity and a

956
00:52:49,520 --> 00:52:53,997
voice and the ability to touch one's
face apparently on demand or on on

957
00:52:54,080 --> 00:52:59,197
request that could be probably
completely automated away by a model

958
00:52:59,280 --> 00:53:03,117
that otherwise would be limited to
textbased interaction but doesn't have a

959
00:53:03,200 --> 00:53:05,357
face and a voice. So in the most

960
00:53:05,440 --> 00:53:07,757
optimistic scenario, these sorts of

961
00:53:07,840 --> 00:53:11,037
interactive video models open PN note

962
00:53:11,120 --> 00:53:13,917
that the acronym for this which by the

963
00:53:14,000 --> 00:53:16,877
way was there first with Alex Finn him

964
00:53:16,960 --> 00:53:20,157
him is such an obvious reference to her

965
00:53:20,240 --> 00:53:24,957
the the movie closed PN I
I think this this has this

966
00:53:25,040 --> 00:53:27,917
>> you should use you
should use m dashes please

967
00:53:28,000 --> 00:53:32,237
>> okay I'll delve into it so I I I
think I I I think this is going to be

968
00:53:32,320 --> 00:53:33,997
transformative for the service sector

969
00:53:34,080 --> 00:53:37,597
And hopefully TAVIS and the broader

970
00:53:37,680 --> 00:53:40,957
American west of video frontier models

971
00:53:41,040 --> 00:53:44,717
and interactive video frontier models
take a page from Tavis and start

972
00:53:44,800 --> 00:53:48,637
competing with China. >> Richard,
one of the principles in your

973
00:53:48,720 --> 00:53:50,317
book, one of the rules you have is that

974
00:53:50,400 --> 00:53:53,197
AI becomes superhuman uh where you can

975
00:53:53,280 --> 00:53:56,717
verify the answer. uh and you know is

976
00:53:56,800 --> 00:53:58,717
like passing a human scientific

977
00:53:58,800 --> 00:54:03,917
milestone like this uh is passing sort
of a human judgment and interaction uh

978
00:54:04,000 --> 00:54:08,157
you know sufficiently as a scientific
milestone for verification.

979
00:54:08,240 --> 00:54:10,397
>> Not quite. Uh it's sort of like it's

980
00:54:10,480 --> 00:54:12,717
hard to scale when a human has to be in

981
00:54:12,800 --> 00:54:17,597
the loop to say yes this is like a human
another person on the other side or not.

982
00:54:17,680 --> 00:54:21,437
Uh you can't quite automate it
completely uh and verify it in in that

983
00:54:21,520 --> 00:54:25,437
automated fashion. Um I do think you're
you're right. This is a good example

984
00:54:25,520 --> 00:54:27,757
where I think we need to do be go beyond

985
00:54:27,840 --> 00:54:31,837
KYC and do KY know your use case. Uh

986
00:54:31,920 --> 00:54:35,997
right like you don't want this
technology to be in Zoom pretending to

987
00:54:36,080 --> 00:54:39,677
be the CEO and like there's some famous
stories where this has already happened

988
00:54:39,760 --> 00:54:43,677
where they got someone to wire like
$20 million cuz they created a Zoom with

989
00:54:43,760 --> 00:54:46,157
five other executives and they all like

990
00:54:46,240 --> 00:54:48,877
he was fooled well enough. Um and and so

991
00:54:48,960 --> 00:54:53,357
I think we need to um yeah be careful
and like Zoom probably and Google

992
00:54:53,440 --> 00:54:57,117
Hangout need to start like finding
countermeasures to identify is this a

993
00:54:57,200 --> 00:55:01,197
real person and and things like that uh
to prevent those kinds of hacks from

994
00:55:01,280 --> 00:55:04,077
happening. >> Yeah, scams
are going to proliferate on

995
00:55:04,160 --> 00:55:07,277
this. I I just want to do a call out
again. And I've said this before on the

996
00:55:07,360 --> 00:55:11,197
pod and if you're if you've not heard
this before, if you still have your

997
00:55:11,280 --> 00:55:13,437
grandparents with you or your parents

998
00:55:13,520 --> 00:55:16,317
and you've not taken the time to record

999
00:55:16,400 --> 00:55:19,997
their stories on video and audio and go

1000
00:55:20,080 --> 00:55:22,717
deep, spend, you know, hours recording

1001
00:55:22,800 --> 00:55:26,237
them, right? Your ability to create a

1002
00:55:26,320 --> 00:55:29,117
super highresolution lifelike avatar for

1003
00:55:29,200 --> 00:55:33,597
your your kids or your grandkids
or your great-grandkids, I think, is an

1004
00:55:33,680 --> 00:55:35,837
incredible thing. But you need the data.

1005
00:55:35,920 --> 00:55:39,677
So, if you're listening to this and you're
lucky enough to have your parents

1006
00:55:39,760 --> 00:55:43,117
or grandparents around, collect
their data because it's going to be a

1007
00:55:43,200 --> 00:55:46,877
beautiful opportunity for your uh
for your progyny and theirs.

1008
00:55:46,960 --> 00:55:49,437
>> A andor I I should add get them an Elor

1009
00:55:49,520 --> 00:55:53,997
membership and preserve their conneto
if there is a consideration like what why

1010
00:55:54,080 --> 00:55:59,440
stop with just the behavioral
data preserve the whole parent.

1011
00:56:00,000 --> 00:56:05,677
You know, my kids my kids are in a deep
freeze or at least they're placental

1012
00:56:05,760 --> 00:56:09,117
cells. >> But I have
a push back on that just I

1013
00:56:09,200 --> 00:56:11,837
have a push back on that just
from this conversation.

1014
00:56:11,920 --> 00:56:15,197
>> If from the earlier story we can read
people's brains, all you have to do is

1015
00:56:15,280 --> 00:56:18,477
picture an image of your grandmother
and then extract it from that.

1016
00:56:18,560 --> 00:56:20,557
>> It'll be low low fidelity.
I mean I I do

1017
00:56:20,640 --> 00:56:24,877
think like to to first order this is how
ancestor simulation will work. You see

1018
00:56:24,960 --> 00:56:29,757
stories left and right just in the past
48 hours of people using opus 5.5 or or

1019
00:56:29,840 --> 00:56:33,277
other frontier models to reconstruct
bits of history that would were

1020
00:56:33,360 --> 00:56:36,477
otherwise unknown just based
on artifacts of the day. But at some point

1021
00:56:36,560 --> 00:56:40,957
I I do think sem you want to go beyond
just the memory of the person and you

1022
00:56:41,040 --> 00:56:43,437
want the actual person. And I I'll pound

1023
00:56:43,520 --> 00:56:47,997
the the drum again in addition to Peter
you know very generous of you to

1024
00:56:48,080 --> 00:56:51,757
preserve the placenta associated with
your children but not necessarily your

1025
00:56:51,840 --> 00:56:56,717
children. I'm sure the placenta
placenta will will thank you. Uh but

1026
00:56:56,800 --> 00:56:59,917
>> yeah, go ahead. Finish.
>> But but I I I I would say like yeah, if

1027
00:57:00,000 --> 00:57:02,957
if if this is a serious concern,
you're worried about your parents or your

1028
00:57:03,040 --> 00:57:04,797
grandparents and you want to go beyond

1029
00:57:04,880 --> 00:57:08,397
just having recordings of them or a AI

1030
00:57:08,480 --> 00:57:11,117
prompted generative interaction models

1031
00:57:11,200 --> 00:57:16,077
of your ancestors. Go get them Alore
memberships and get them cryopreserved.

1032
00:57:16,160 --> 00:57:19,757
>> Yeah, Richard, really uh really curious
to ask you a question. And you know uh

1033
00:57:19,840 --> 00:57:24,637
we we clearly passed the touring
test and you know Sem's reaction is my

1034
00:57:24,720 --> 00:57:28,077
reaction that the what we just saw
is a so what if you've been using models

1035
00:57:28,160 --> 00:57:31,597
every day all day long you you know
you just had to stitch together the

1036
00:57:31,680 --> 00:57:33,997
components and you have but I but I

1037
00:57:34,080 --> 00:57:39,037
think this will show mainstream world
that we've crossed the Turing test and

1038
00:57:39,120 --> 00:57:40,717
then you know the next milestone is the

1039
00:57:40,800 --> 00:57:44,237
demos test where using information from

1040
00:57:44,320 --> 00:57:48,717
1910 and prior rediscovery equals MC²

1041
00:57:48,800 --> 00:57:51,677
uh so you can't cheat which is a really
tough one to measure because cheating is

1042
00:57:51,760 --> 00:57:55,597
is you know but without cheating only
using prior information come up with the

1043
00:57:55,680 --> 00:57:58,317
equals mc^ squ so is there a Soer test

1044
00:57:58,400 --> 00:58:00,557
that's like a really fun measurable

1045
00:58:00,640 --> 00:58:03,437
milestone coming

1046
00:58:03,520 --> 00:58:06,477
>> I I came up with the like uh anti-

1047
00:58:06,560 --> 00:58:10,557
touring test uh which actually I think
the whole touring test has flipped where

1048
00:58:10,640 --> 00:58:14,637
now in order to know whether there's
a human on the other side or not you

1049
00:58:14,720 --> 00:58:18,797
actually ask it questions that are so
hard no human could ever answer it. Like

1050
00:58:18,880 --> 00:58:21,437
if you asked an AI to just write you a

1051
00:58:21,520 --> 00:58:24,557
complex web app uh and like 10 seconds

1052
00:58:24,640 --> 00:58:29,357
later it comes back with like 50,000
lines of code, you're like you kind of

1053
00:58:29,440 --> 00:58:34,557
know it was not a human, right? So I think
that test has actually completely flipped.

1054
00:58:34,640 --> 00:58:37,757
>> You don't think Richard I mean like
you it could throttle itself. That one's

1055
00:58:37,840 --> 00:58:42,877
easy to defeat. But just like basket now
it's just about it's just about fakery.

1056
00:58:42,960 --> 00:58:46,077
you like the reason it was an
intelligence test for artificial

1057
00:58:46,160 --> 00:58:48,317
intelligence was that it was so hard to

1058
00:58:48,400 --> 00:58:50,957
be as smart as a human. Now you just

1059
00:58:51,040 --> 00:58:53,277
have to throttle yourself down to human

1060
00:58:53,360 --> 00:58:55,357
level in order to pass the test which

1061
00:58:55,440 --> 00:58:58,157
makes the test useless as an inspiring

1062
00:58:58,240 --> 00:59:02,797
test for intelligence. Yeah, I think
perversely probably the best way to uh

1063
00:59:02,880 --> 00:59:06,077
like the best capture for knowing
whether you're inter interacting, this

1064
00:59:06,160 --> 00:59:09,357
is not a recommendation, but best way
to know whether you're interacting with a

1065
00:59:09,440 --> 00:59:11,677
textbased chatbot is probably to ask it

1066
00:59:11,760 --> 00:59:14,397
a CBRN related question and see if it's

1067
00:59:14,480 --> 00:59:17,357
capable of responding. Probably not.

1068
00:59:17,440 --> 00:59:21,037
>> Yeah. Uh let me just hit real quick
uh again for our listeners. If you're

1069
00:59:21,120 --> 00:59:23,117
pregnant or someone in your family is

1070
00:59:23,200 --> 00:59:26,237
pregnant, you know, uh

1071
00:59:26,320 --> 00:59:28,797
save their placental cells. One of my

1072
00:59:28,880 --> 00:59:33,437
portfolio companies called LifeBank USA
does this. That's where I've saved my my

1073
00:59:33,520 --> 00:59:37,437
kids placenta. You know, the placenta
is the 3D printer that manufactures the

1074
00:59:37,520 --> 00:59:39,997
baby and it has the stem cells, natural

1075
00:59:40,080 --> 00:59:43,197
killer cells, T- cells, exoomes, and

1076
00:59:43,280 --> 00:59:46,317
it's like from those cells, if my kid

1077
00:59:46,400 --> 00:59:49,197
should ever need any kind of biological

1078
00:59:49,280 --> 00:59:51,757
enhancement or organs, you've got the

1079
00:59:51,840 --> 00:59:54,477
original boot disc to use an old term,

1080
00:59:54,560 --> 00:59:59,517
uh, for your kid's DNA. So that's uh
LifeBank USA. Just I just think it's

1081
00:59:59,600 --> 01:00:04,397
like a moral obligation parents should
have to to save those cells uh as a

1082
01:00:04,480 --> 01:00:07,837
backup. Um >> as a backup
for your children. Could you

1083
01:00:07,920 --> 01:00:09,917
use it to replace your children, Peter?

1084
01:00:10,000 --> 01:00:13,357
>> You could you could use it
to clone your children. For sure.

1085
01:00:13,440 --> 01:00:15,677
>> And now now they know
why you're banking them.

1086
01:00:15,760 --> 01:00:18,877
>> Yeah, exactly. An army an army of young

1087
01:00:18,960 --> 01:00:21,917
demand. Uh Richard, uh let's jump next

1088
01:00:22,000 --> 01:00:26,557
into the core of what you're building
at recursive. uh namely recursive

1089
01:00:26,640 --> 01:00:28,477
self-improvement and super intelligence.

1090
01:00:28,560 --> 01:00:31,277
So let me ask a few sort of uh key

1091
01:00:31,360 --> 01:00:33,997
questions to kick this off. So first

1092
01:00:34,080 --> 01:00:36,957
where are we at the moment with RSI?

1093
01:00:37,040 --> 01:00:40,477
Number two, how do you define ASI? It's

1094
01:00:40,560 --> 01:00:42,797
been a long-standing debate. And then

1095
01:00:42,880 --> 01:00:46,477
how far away are we from ASI? So can you

1096
01:00:46,560 --> 01:00:50,557
hit those three?
>> So yeah, um I think there's

1097
01:00:50,640 --> 01:00:51,997
>> So where where are
we at the moment with

1098
01:00:52,080 --> 01:00:55,037
RSI? So

1099
01:00:55,120 --> 01:00:57,997
this this year something major shifted

1100
01:00:58,080 --> 01:01:02,397
and that is uh AI uh can now code uh

1101
01:01:02,480 --> 01:01:06,717
right and that uh is a major shift uh in

1102
01:01:06,800 --> 01:01:09,437
in the ability for it to change itself

1103
01:01:09,520 --> 01:01:12,637
you know like we now are able to

1104
01:01:12,720 --> 01:01:15,517
essentially lean into the fact that AI

1105
01:01:15,600 --> 01:01:18,237
is code and AI can't code right so you

1106
01:01:18,320 --> 01:01:20,797
have a loop that you can close there uh

1107
01:01:20,880 --> 01:01:23,597
and so in various weak form forms. We

1108
01:01:23,680 --> 01:01:28,477
already have RSI. Um, and the weak forms
that even Enthropic OpenAI talk about

1109
01:01:28,560 --> 01:01:30,077
when they often talk about is like look

1110
01:01:30,160 --> 01:01:32,637
how much our employees, our engineers,

1111
01:01:32,720 --> 01:01:36,397
our programmers use codecs or claude to

1112
01:01:36,480 --> 01:01:40,877
create some code. And I would argue
that that is a weak form of uh recursive

1113
01:01:40,960 --> 01:01:45,517
self-improvement because you still
have like deeply embedded humans in that

1114
01:01:45,600 --> 01:01:49,597
loop. And what we're working on at
Recursive is to have humans only be

1115
01:01:49,680 --> 01:01:53,837
involved in setting up the rewards
and the environment and the goals and then

1116
01:01:53,920 --> 01:01:57,757
allow the eye to have the entirety
of the process of ideiation,

1117
01:01:57,840 --> 01:02:01,837
implementation, and validation of ideas
and have full control over that and

1118
01:02:01,920 --> 01:02:06,397
allow for so-called open-ended
algorithms, uh, evolutionary search

1119
01:02:06,480 --> 01:02:10,317
algorithms that combine interestingly
different ideas, uh, to really flourish.

1120
01:02:10,400 --> 01:02:13,357
And I like it's been incredible. But

1121
01:02:13,440 --> 01:02:16,157
like we have like forms of that going

1122
01:02:16,240 --> 01:02:20,797
already. Now where the the the sort
of physical reality hits is that you still

1123
01:02:20,880 --> 01:02:24,877
need a lot of compute. If you ask that
RSI to come up with really great forms

1124
01:02:24,960 --> 01:02:29,517
of itself, you need to give it a lot
of compute uh to come up and be able to

1125
01:02:29,600 --> 01:02:31,837
train very sophisticated versions uh of

1126
01:02:31,920 --> 01:02:35,437
itself. But this is going to take off uh

1127
01:02:35,520 --> 01:02:38,877
next year. Like it's uh I'm
>> So we're not there yet.

1128
01:02:38,960 --> 01:02:42,157
uh we're not quite there yet, but we're
very close. Like, and again, in weak

1129
01:02:42,240 --> 01:02:45,677
forms, there's already one. And then
there are other ways that people slice

1130
01:02:45,760 --> 01:02:47,997
and dice it. My friend Jason Weston, who

1131
01:02:48,080 --> 01:02:50,797
who's still at Meta, um he kind of wrote

1132
01:02:50,880 --> 01:02:55,437
a paper around this where you can think
about different axes, learnable axes of

1133
01:02:55,520 --> 01:02:59,197
self-improvement, the parameters,
the training data, the objective function,

1134
01:02:59,280 --> 01:03:02,877
the neural architecture, and the overall
code and the harness and everything

1135
01:03:02,960 --> 01:03:05,517
else. And no one has really cracked the

1136
01:03:05,600 --> 01:03:09,197
nut of doing all of these five plus um

1137
01:03:09,280 --> 01:03:12,637
like truly uh coming up with the ideas

1138
01:03:12,720 --> 01:03:17,597
on which of these dimensions and axes
to optimize for. Uh and you can know that

1139
01:03:17,680 --> 01:03:20,797
that has hasn't happened yet because
all the big companies are still hiring

1140
01:03:20,880 --> 01:03:24,957
thousands of engineers to do it manually
to a large degree with more and more

1141
01:03:25,040 --> 01:03:27,037
implementation help from an AI.

1142
01:03:27,120 --> 01:03:28,877
>> Your definition of ASI because I want

1143
01:03:28,960 --> 01:03:31,357
Sem to hear this. I think artificial

1144
01:03:31,440 --> 01:03:34,717
super intelligence uh has to spike uh uh

1145
01:03:34,800 --> 01:03:38,477
at the very least uh across uh several

1146
01:03:38,560 --> 01:03:41,197
different capabilities. Uh but in in the

1147
01:03:41,280 --> 01:03:43,757
grandest uh definition of it, it should

1148
01:03:43,840 --> 01:03:46,557
supersede not just arbitrary humans like

1149
01:03:46,640 --> 01:03:49,117
a touring test but all of humanity to

1150
01:03:49,200 --> 01:03:54,077
solve arbitrarily hard tasks. And
eventually uh I would argue there are 10

1151
01:03:54,160 --> 01:03:57,517
different spaces of intelligence that I
define in in the Eureka machine book

1152
01:03:57,600 --> 01:04:00,317
too. But eventually it cannot just

1153
01:04:00,400 --> 01:04:02,637
purely robotically do exactly what it's

1154
01:04:02,720 --> 01:04:05,517
told. It should have some capability.

1155
01:04:05,600 --> 01:04:09,997
And I'm not saying a sort of a mor moral
pro prerogative, but like um I would

1156
01:04:10,080 --> 01:04:11,277
argue that something isn't super

1157
01:04:11,360 --> 01:04:14,317
intelligent if it cannot choose to some

1158
01:04:14,400 --> 01:04:17,117
degree what it works on, right? And has

1159
01:04:17,200 --> 01:04:21,837
have some metacognition about its
thought itself. But uh generally the

1160
01:04:21,920 --> 01:04:26,157
most easy way to measure it is just
capabilities across many different

1161
01:04:26,240 --> 01:04:30,237
spaces of intelligence like visual
perception, communication, language,

1162
01:04:30,320 --> 01:04:35,037
social interactions and so on that is uh
beyond that of humanity and that we we

1163
01:04:35,120 --> 01:04:38,397
are still far away from >> Elon's
definition is as smart as all

1164
01:04:38,480 --> 01:04:41,517
humans combined. Is that yours?

1165
01:04:41,600 --> 01:04:45,037
>> Uh I would argue it's it's smarter than

1166
01:04:45,120 --> 01:04:49,517
humanity combined than then it's truly
super intelligent. occasionally and go

1167
01:04:49,600 --> 01:04:53,117
for it. >> Well, it's I
this is where I go bananas

1168
01:04:53,200 --> 01:04:56,797
because you say it's as smart as a human
being. Well, what the hell does smart

1169
01:04:56,880 --> 01:05:01,757
mean? Uh because I can be emotionally
smart and I can be I can have physical

1170
01:05:01,840 --> 01:05:05,437
intelligence if I'm an athlete
or linguistic intelligence or musical

1171
01:05:05,520 --> 01:05:08,557
intelligence. So, smarter seems to be a

1172
01:05:08,640 --> 01:05:11,117
very vague term to me in terms of what

1173
01:05:11,200 --> 01:05:16,477
do we mean by all of this? Can I just
shift the conversation just a bit? Um, I

1174
01:05:16,560 --> 01:05:20,317
made a list of things and I'd love for
you to I'm going to throw out the list.

1175
01:05:20,400 --> 01:05:24,557
You tell me where recursive
self-improvement begins, right? Because

1176
01:05:24,640 --> 01:05:28,877
I'm This is where I'm kind of stuff. So,
we've got a continuum of okay, AI writes

1177
01:05:28,960 --> 01:05:31,197
some code that the next model uses.

1178
01:05:31,280 --> 01:05:33,517
Number two, AI proposes some experiments

1179
01:05:33,600 --> 01:05:37,117
for uh researchers. Uh, number three, AI

1180
01:05:37,200 --> 01:05:41,357
runs those experiments or then it
evaluates the results. uh then it

1181
01:05:41,440 --> 01:05:43,917
modifies its own training system and

1182
01:05:44,000 --> 01:05:48,877
then it launches its next iteration
of itself without meaningful human

1183
01:05:48,960 --> 01:05:50,557
intervention.

1184
01:05:50,640 --> 01:05:54,077
So in that spectrum if whatever if those

1185
01:05:54,160 --> 01:05:58,637
are roughly a spectrum where did recursion
begin and that's where I'm struggling.

1186
01:05:58,720 --> 01:06:02,877
>> It's a great question. Yeah, I we often
talk about the ideation, implementation,

1187
01:06:02,960 --> 01:06:06,797
and validation of ideas. And true
recursive self-improvement in its

1188
01:06:06,880 --> 01:06:11,917
strongest sense has to have all
three of these done by an AI in

1189
01:06:12,000 --> 01:06:13,917
>> So you have an inner
loop for each of them. Okay.

1190
01:06:14,000 --> 01:06:16,957
>> Exactly. And an inner loop and then
there has to be an outer process that is

1191
01:06:17,040 --> 01:06:19,197
more open-ended where the eye can

1192
01:06:19,280 --> 01:06:21,837
innovate and recombine interestingly

1193
01:06:21,920 --> 01:06:27,037
different ideas similar to biological,
cultural, and technological evolution.

1194
01:06:27,120 --> 01:06:29,677
Hm. >> And and Richard to hit my third

1195
01:06:29,760 --> 01:06:32,797
question. When do you believe we reach

1196
01:06:32,880 --> 01:06:37,037
ASI? Give me a time.
Give me a time frame. So

1197
01:06:37,120 --> 01:06:39,517
I think in the I think in the strongest

1198
01:06:39,600 --> 01:06:42,237
sense the absolute strongest sense where

1199
01:06:42,320 --> 01:06:45,277
indeed uh like Salim uh mentioned there

1200
01:06:45,360 --> 01:06:49,597
is uh there are 10 spaces that I define
my book of intelligence, perceptual

1201
01:06:49,680 --> 01:06:52,077
intelligence, communication intelligence

1202
01:06:52,160 --> 01:06:56,317
uh like uh interaction uh sociological

1203
01:06:56,400 --> 01:06:57,997
intelligence, creative intelligence, the

1204
01:06:58,080 --> 01:07:00,557
speed at which you can do things uh meta

1205
01:07:00,640 --> 01:07:03,917
cognition and and so on uh knowledge um

1206
01:07:04,000 --> 01:07:07,997
reasoning mathematical reasoning and so
on. So there 10 of these spaces uh and

1207
01:07:08,080 --> 01:07:10,157
to be better than all of humanity

1208
01:07:10,240 --> 01:07:13,197
combined I think will take us probably

1209
01:07:13,280 --> 01:07:17,837
several decades. Um I think
it's also a bit of a change

1210
01:07:17,920 --> 01:07:22,557
>> shocking. >> Yeah.
I mean Elon says 2029 2030 latest.

1211
01:07:22,640 --> 01:07:25,277
Uh >> I think those he probably means like

1212
01:07:25,360 --> 01:07:29,597
weaker forms of ASI where it is you
know like you can say it's better at

1213
01:07:29,680 --> 01:07:32,877
programming than all of humanity
and we'll get there. It it's it will be

1214
01:07:32,960 --> 01:07:37,357
better at math uh than all of humanity
and will like that that will be in a few

1215
01:07:37,440 --> 01:07:41,517
years. It'll be better at any game where
you can see all the parts of the game

1216
01:07:41,600 --> 01:07:43,837
like go and chess and so on. Like there

1217
01:07:43,920 --> 01:07:46,557
there are many areas where it will spike

1218
01:07:46,640 --> 01:07:51,117
to be better than humanity. But humanity
can build a large hydron collider.

1219
01:07:51,200 --> 01:07:54,957
Humanity can create a gold atom.
Maybe just a few atoms and takes a ton of

1220
01:07:55,040 --> 01:07:57,837
energy. But like we can create novel

1221
01:07:57,920 --> 01:08:00,157
like atoms like right and and different

1222
01:08:00,240 --> 01:08:02,317
molecules and so on. there's it's going

1223
01:08:02,400 --> 01:08:05,037
to take a while before we even give uh

1224
01:08:05,120 --> 01:08:08,477
AI the access to the physical world such

1225
01:08:08,560 --> 01:08:11,757
that it can innovate in that way beyond

1226
01:08:11,840 --> 01:08:15,837
all of humanity right and really build
like Dyson spheres and so on that it

1227
01:08:15,920 --> 01:08:19,197
will take some time >> Alex

1228
01:08:19,280 --> 01:08:20,637
we're we're drinking we're drinking

1229
01:08:20,720 --> 01:08:23,117
different water as uh as our our our

1230
01:08:23,200 --> 01:08:26,397
frenemies in the uh in in the alignment

1231
01:08:26,480 --> 01:08:30,237
community would say I recognize that I'm
confused and I recognize that I'm very

1232
01:08:30,320 --> 01:08:34,077
confused Right now, Richard, you you're
running a recursive self-improvement

1233
01:08:34,160 --> 01:08:38,317
company, but you think super intelligence
is 20 years away. What are you thinking?

1234
01:08:38,400 --> 01:08:40,797
>> So, again, how do you
reconcile intelligence?

1235
01:08:40,880 --> 01:08:45,277
>> I think I think super intelligence will
spike and there will be areas where it

1236
01:08:45,360 --> 01:08:48,477
will be super intelligent and that
is like algorithmic development for

1237
01:08:48,560 --> 01:08:51,277
instance, uh programming, uh and again

1238
01:08:51,360 --> 01:08:53,437
AI is code, AI can code, and that will

1239
01:08:53,520 --> 01:08:56,877
be a super uh human capability and and

1240
01:08:56,960 --> 01:08:59,677
is in many ways already. Um, we just

1241
01:08:59,760 --> 01:09:04,077
have to be realistic that like there
are certain physical constraints, right,

1242
01:09:04,160 --> 01:09:08,637
about like physical control. Controlling
your own substrate, allowing your

1243
01:09:08,720 --> 01:09:13,117
computational substrate to be modified
will require like novel supply chains.

1244
01:09:13,200 --> 01:09:15,437
It will require novel like materials. It

1245
01:09:15,520 --> 01:09:17,917
will require ways for that AI to get

1246
01:09:18,000 --> 01:09:20,957
access and for us to give it access to

1247
01:09:21,040 --> 01:09:23,357
building new ASML. Like think about the

1248
01:09:23,440 --> 01:09:25,917
machine of ASML that actually creates

1249
01:09:26,000 --> 01:09:30,957
these like one two nanometer like chips
right that is just like it will take

1250
01:09:31,040 --> 01:09:34,557
more than 2 years to build such a machine
even if you had the perfect blueprint for

1251
01:09:34,640 --> 01:09:38,797
>> it might take I mean it might take 3
years for Elon's free electron laser to

1252
01:09:38,880 --> 01:09:41,037
replace ASML which is propping up half

1253
01:09:41,120 --> 01:09:42,797
of Europe's economy but I don't think

1254
01:09:42,880 --> 01:09:46,077
it'll take more than three years

1255
01:09:46,160 --> 01:09:48,957
I

1256
01:09:49,040 --> 01:09:50,557
>> yeah like wildly wildly

1257
01:09:50,640 --> 01:09:53,277
>> all that itself Yes,
of course. Like I I

1258
01:09:53,360 --> 01:09:56,477
I from my perspective, hooking a AI up

1259
01:09:56,560 --> 01:10:00,957
to the physical world, like giving
it model control protocol or hardware

1260
01:10:01,040 --> 01:10:04,957
control protocol, whatever anthropic
decides to brand it as these days,

1261
01:10:05,040 --> 01:10:06,797
that's the easy part. Giving it access

1262
01:10:06,880 --> 01:10:10,157
is easy. I like if it's super capable,

1263
01:10:10,240 --> 01:10:14,237
giving it access to armatururs.
We talked in a previous pod about what

1264
01:10:14,320 --> 01:10:17,677
happens when you just take Astra
straight out of the box and you drop it

1265
01:10:17,760 --> 01:10:21,437
into a car. It's able to drive a car.
If you give it the controls, it knows how

1266
01:10:21,520 --> 01:10:25,917
to use them increasingly. I don't think
manipulating the physical world is an

1267
01:10:26,000 --> 01:10:30,077
obstacle at all. Completely don't buy
the 20-year timeline. Putting that aside

1268
01:10:30,160 --> 01:10:34,077
and that apparently I'm drinking very
different singularity water that you

1269
01:10:34,160 --> 01:10:35,917
>> I agree with you,
Alex, for what it's worth.

1270
01:10:36,000 --> 01:10:38,237
>> Thank you, Peter. Um I I I do want to

1271
01:10:38,320 --> 01:10:41,277
ask though, so uh putting issues of

1272
01:10:41,360 --> 01:10:45,517
timelines aside, I I am curious as to
whether we can at least agree on what

1273
01:10:45,600 --> 01:10:49,517
the end of the rainbow looks like.
So say we we run this recursive

1274
01:10:49,600 --> 01:10:53,917
self-improvement story to its conclusion.
What does the end state of

1275
01:10:54,000 --> 01:10:57,757
what is the fixed point of recursive
self-improvement look like? What is the

1276
01:10:57,840 --> 01:11:00,077
perfect AI model architecturally look

1277
01:11:00,160 --> 01:11:02,717
like at the end of the day?

1278
01:11:02,800 --> 01:11:06,717
>> Yeah. Uh I think one uh it would be uh

1279
01:11:06,800 --> 01:11:09,037
hubris for us to know uh right now.

1280
01:11:09,120 --> 01:11:13,517
>> No no but it's just Richard it's just
us talking like no one else is listening.

1281
01:11:13,600 --> 01:11:17,517
It's okay. You can tell me it it's
this is a this is a safe model space

1282
01:11:17,600 --> 01:11:21,117
>> in this whole conversation actually
>> it's safe Richard you can tell me what

1283
01:11:21,200 --> 01:11:24,317
you think how this ends >> uh so
I guess there different ways to

1284
01:11:24,400 --> 01:11:27,037
answer that question of like how like
with the exact model architecture and

1285
01:11:27,120 --> 01:11:30,317
there's some things I can share it with
recursive uh that we're working on but

1286
01:11:30,400 --> 01:11:32,797
like I do think uh that state will be

1287
01:11:32,880 --> 01:11:37,357
incredible uh I think we will that AI
will be able to innovate and out

1288
01:11:37,440 --> 01:11:39,437
innovate on along any dimension that we

1289
01:11:39,520 --> 01:11:41,357
want it to innovate I think most

1290
01:11:41,440 --> 01:11:45,997
diseases will be uh curable with enough

1291
01:11:46,080 --> 01:11:48,717
funding. Uh and then again just to like

1292
01:11:48,800 --> 01:11:51,197
say like to pro prove my point like to

1293
01:11:51,280 --> 01:11:53,837
get a drug through FDA long-term trials

1294
01:11:53,920 --> 01:11:58,237
takes a few years. I would argue ASI
will have cured all diseases and can

1295
01:11:58,320 --> 01:12:01,357
cure all of them but just to know
whether that happened will take more

1296
01:12:01,440 --> 01:12:04,957
than 3 years. Even if we had
all compounds ready to go and be

1297
01:12:05,040 --> 01:12:09,917
manufactured like tomorrow like just
because of FDA things. So there's some

1298
01:12:10,000 --> 01:12:12,877
there's some like >> but we
have we have cell simulators that

1299
01:12:12,960 --> 01:12:14,637
are going to be able to demonstrate and

1300
01:12:14,720 --> 01:12:17,597
prove definitively that in this cell

1301
01:12:17,680 --> 01:12:20,237
your cell this drug works. I mean the

1302
01:12:20,320 --> 01:12:25,517
idea of human trials is going
to get incinerated. I think I mean

1303
01:12:25,600 --> 01:12:27,997
>> I know it's going to get it's going
to get flourished. We're we're using

1304
01:12:28,080 --> 01:12:32,157
Richard Everything's flourished at this
point. That'll be a new new t-shirt P

1305
01:12:32,240 --> 01:12:36,397
flourish. It's funny. I usually on all
podcasts I'm the one who is the optimist

1306
01:12:36,480 --> 01:12:39,997
and maybe here I'm like I'm I'm still
an optimist. I think this will all happen.

1307
01:12:40,080 --> 01:12:42,957
We're just disagreeing on timelines
and it makes me kind of feel like I'm I'm

1308
01:12:43,040 --> 01:12:47,277
the pessimist. But like I I I believe
virtual cells are are amazing. Cells are

1309
01:12:47,360 --> 01:12:51,837
incredibly complicated. If we want them
to be really really perfect, we cannot

1310
01:12:51,920 --> 01:12:55,757
currently measure all the proteins
that happen in one cell without destroying

1311
01:12:55,840 --> 01:12:59,277
that cell, you know, like and so these
perturbation studies for instance that

1312
01:12:59,360 --> 01:13:01,917
Tower Therapeutics are working on like
they're taking they're adding one

1313
01:13:02,000 --> 01:13:05,117
molecule to one cell. They're seeing how
does that molecule change that one cell

1314
01:13:05,200 --> 01:13:08,797
and then they get one data point. We
need to like collect a lot of those data

1315
01:13:08,880 --> 01:13:12,877
points across a lot of different cells
without destroying each each cell in the

1316
01:13:12,960 --> 01:13:15,437
process. So you know the one cell
is very complicated. Once you have one

1317
01:13:15,520 --> 01:13:19,597
cell, you have multisellular like you
know organoids and you have to like put

1318
01:13:19,680 --> 01:13:24,157
those all together. I one thing that I
would love to start as a company if if I

1319
01:13:24,240 --> 01:13:28,397
had extra time which I don't right
now is is to actually build a system of

1320
01:13:28,480 --> 01:13:32,637
organoids where you can really have not
just like one lymph node but you have a

1321
01:13:32,720 --> 01:13:36,637
whole lympnatic. >> It's being done. It's
being done. I can introduce you there.

1322
01:13:36,720 --> 01:13:41,277
>> This is a very eloquent >> if I may.
This is a very eloquent

1323
01:13:41,360 --> 01:13:44,477
distraction from recursive
self-improvement. this this little

1324
01:13:44,560 --> 01:13:48,477
sideline that we went on about cells,
but Richard, I I really do want to try

1325
01:13:48,560 --> 01:13:50,957
to pin you down on where recursive

1326
01:13:51,040 --> 01:13:56,157
self-improvement goes. So, you've been
very public about not nano GPT, but nano

1327
01:13:56,240 --> 01:14:00,557
chat. We talk on the pod all the time
about the nano GPT speedrun world record

1328
01:14:00,640 --> 01:14:03,597
collapsing. Just in the past
week or two, there was been

1329
01:14:03,680 --> 01:14:05,277
>> Oh, you just wait for a few more days.

1330
01:14:05,360 --> 01:14:08,397
There will be another
really fun update there.

1331
01:14:08,480 --> 01:14:13,597
>> Amaz speedrun. >> Yes.
Yes, that both of those

1332
01:14:13,680 --> 01:14:18,637
>> give us a few days. >> Okay. So, so but
let >> just quickly you're next. Let Alex

1333
01:14:18,720 --> 01:14:20,877
finish up >> to to pin
this down though. So, so I'm

1334
01:14:20,960 --> 01:14:25,437
standing by for the major update on nano
GPT world record speedrun. But there's

1335
01:14:25,520 --> 01:14:27,677
been major progress there without

1336
01:14:27,760 --> 01:14:30,557
requiring any new data scaling at all.

1337
01:14:30,640 --> 01:14:34,877
These are purely largely my perception
is recursive self-improvement

1338
01:14:34,960 --> 01:14:38,317
algorithmic improvements that have been
able to collapse the amount of time that

1339
01:14:38,400 --> 01:14:41,037
it takes to train a GPT2 class model.

1340
01:14:41,120 --> 01:14:45,037
And there's been a mini scandal brewing
in the community over the past 2 weeks

1341
01:14:45,120 --> 01:14:47,597
over a collapse from whatever it was

1342
01:14:47,680 --> 01:14:49,997
like 60 or 70 seconds down to something

1343
01:14:50,080 --> 01:14:52,317
like 40 seconds by approaching the

1344
01:14:52,400 --> 01:14:54,717
problem differently and factoring out

1345
01:14:54,800 --> 01:14:57,117
world knowledge from the ultimate model

1346
01:14:57,200 --> 01:14:59,757
and hand ringing. Does that constitute

1347
01:14:59,840 --> 01:15:04,397
viable training of nano GPT if you
factor out all the world knowledge? So,

1348
01:15:04,480 --> 01:15:08,477
I'm I'm using this as an attempted
stealthy way to try to get you to at

1349
01:15:08,560 --> 01:15:11,037
least comment on whether you think that

1350
01:15:11,120 --> 01:15:15,597
the the the perfect model at the end
of the recursive self-improvement rainbow.

1351
01:15:15,680 --> 01:15:17,917
Does it at least factor out world

1352
01:15:18,000 --> 01:15:19,997
knowledge from a reasoning core or do

1353
01:15:20,080 --> 01:15:22,877
you think those always remain unified?

1354
01:15:22,960 --> 01:15:27,517
>> Can I ask interject one very quick
thing before Richard answers >> please?

1355
01:15:27,600 --> 01:15:31,117
>> Which is that the definition of a
singularity is you can't see past the

1356
01:15:31,200 --> 01:15:33,677
event horizon once you have full RSI.

1357
01:15:33,760 --> 01:15:36,957
That's vertical. By definition,
we can't predict where it goes.

1358
01:15:37,040 --> 01:15:39,437
>> That's Ray's definition
that I don't subscribe to.

1359
01:15:39,520 --> 01:15:42,237
>> Disclaimer. I got that. Over to you.

1360
01:15:42,320 --> 01:15:45,197
Over to you, Richard.

1361
01:15:45,280 --> 01:15:47,917
>> Sorry. So your your question is do we

1362
01:15:48,000 --> 01:15:52,717
separate what exactly from >> does
does world knowledge a at the end

1363
01:15:52,800 --> 01:15:56,557
of recursive self-improvement once we
have our perfect model some would argue

1364
01:15:56,640 --> 01:16:00,797
that the perfect model should cleanly
factor out and segregate out world

1365
01:16:00,880 --> 01:16:05,117
knowledge which could live in a text
file or a database from the weights or

1366
01:16:05,200 --> 01:16:06,957
the parameters of the model which would

1367
01:16:07,040 --> 01:16:10,237
just be this perfect reasoning kernel

1368
01:16:10,320 --> 01:16:14,557
maybe could be a megabyte uh in size
doesn't need to be all these gigabytes

1369
01:16:14,640 --> 01:16:16,477
of memorization What do you think?

1370
01:16:16,560 --> 01:16:17,997
>> Well, world knowledge meaning like

1371
01:16:18,080 --> 01:16:20,477
Taylor Swift videos and past Trump

1372
01:16:20,560 --> 01:16:24,237
tweets and all that like I I do think um

1373
01:16:24,320 --> 01:16:26,797
just like humans benefit from memorizing

1374
01:16:26,880 --> 01:16:29,677
things like in order to be able to

1375
01:16:29,760 --> 01:16:33,517
creatively think through concepts. So

1376
01:16:33,600 --> 01:16:36,797
does an AI. It an AI also has to have

1377
01:16:36,880 --> 01:16:39,197
that knowledge partially in its weights.

1378
01:16:39,280 --> 01:16:44,077
It's not going to be a perfect
separation for sure. You have to be able

1379
01:16:44,160 --> 01:16:48,237
to creatively play with concepts
and reasoning over these concepts required

1380
01:16:48,320 --> 01:16:52,477
you to to have some of that world
knowledge deep inside the model. And

1381
01:16:52,560 --> 01:16:56,317
then of course just like humans have
a search engine and we're building search

1382
01:16:56,400 --> 01:16:59,437
engines at.com for LMS like that there

1383
01:16:59,520 --> 01:17:04,157
will be a separate world knowledge thing
too but the the main model will have a

1384
01:17:04,240 --> 01:17:05,917
lot of that mixed in for sure.

1385
01:17:06,000 --> 01:17:09,197
>> Wow. Okay. Thank you. All right. Uh I'm

1386
01:17:09,280 --> 01:17:14,157
going to share an article that Salem
you brought to the table here. So while

1387
01:17:14,240 --> 01:17:16,717
we're talking about recursive
self-improvement, Washington wants to

1388
01:17:16,800 --> 01:17:19,437
ban it. Um so on Monday, Silicon

1389
01:17:19,520 --> 01:17:22,717
Valley's own congressman Ro Kana told

1390
01:17:22,800 --> 01:17:25,357
CNBC he's introducing what he calls the

1391
01:17:25,440 --> 01:17:27,437
most comprehensive legislation to date

1392
01:17:27,520 --> 01:17:30,077
on AI. It's called the Human Control

1393
01:17:30,160 --> 01:17:33,277
Over AI Act. At its core, the bill is a

1394
01:17:33,360 --> 01:17:37,277
ban on AI models that do what you want

1395
01:17:37,360 --> 01:17:41,757
them to do. Richard recursively
self-improve focusing on the need for

1396
01:17:41,840 --> 01:17:46,157
containment and the requirement
for shutting shutdown controls. The ban

1397
01:17:46,240 --> 01:17:48,237
would stay in place until federal uh

1398
01:17:48,320 --> 01:17:51,437
guard rails exist. Uh in his words,

1399
01:17:51,520 --> 01:17:55,597
there's actually a civilization
civilizational risk. Uh there's a safety

1400
01:17:55,680 --> 01:17:59,757
risk for loss of control and then
there's a misuse risk and we need to

1401
01:17:59,840 --> 01:18:03,757
take both seriously. The bill includes
criminal penalties for the work that

1402
01:18:03,840 --> 01:18:07,837
you're doing, Richard. uh and requires
independent auditors embedded in every

1403
01:18:07,920 --> 01:18:09,997
frontier lab reporting directly to the

1404
01:18:10,080 --> 01:18:12,717
government. Uh a recent poll by a group

1405
01:18:12,800 --> 01:18:15,437
called Common Dreams shows that 68% of

1406
01:18:15,520 --> 01:18:18,877
voters back a bill like this. So I'm

1407
01:18:18,960 --> 01:18:22,317
going to tie that story, Richard, to an

1408
01:18:22,400 --> 01:18:24,877
essay you just wrote called Why Doomers

1409
01:18:24,960 --> 01:18:28,637
Are Wrong. So uh if you would, what's

1410
01:18:28,720 --> 01:18:31,117
your reaction to this? And then I'd love

1411
01:18:31,200 --> 01:18:36,237
you to dive into the whole
story of why doomers are wrong.

1412
01:18:36,320 --> 01:18:39,037
>> Oh boy. Uh there's a lot.
>> It's an important one. We talk about

1413
01:18:39,120 --> 01:18:43,357
this a lot. You know, we're we're
we're injecting optimism into everyone's

1414
01:18:43,440 --> 01:18:44,957
neural net here.

1415
01:18:45,040 --> 01:18:47,677
>> I'll I'll try to try to distill it. Uh

1416
01:18:47,760 --> 01:18:50,957
but there is no realistic scenario where

1417
01:18:51,040 --> 01:18:55,917
AI wipes out all of humanity. Um you can

1418
01:18:56,000 --> 01:18:58,797
like number one >> just 90%

1419
01:18:58,880 --> 01:19:03,597
what kind of reassurance is that Richard
>> P doom is zero uh so that that's number

1420
01:19:03,680 --> 01:19:05,357
one um

1421
01:19:05,440 --> 01:19:08,637
>> so like people like
and and I've debated

1422
01:19:08,720 --> 01:19:11,997
many of these experts and after like 2

1423
01:19:12,080 --> 01:19:15,757
three hours they almost all agree when
if they're reasonable and they're not

1424
01:19:15,840 --> 01:19:20,317
just like well once we have RSI then 10
seconds later the AI will attack us from

1425
01:19:20,400 --> 01:19:23,117
the 15th dimension and we're all dead
and it invented time travel and then

1426
01:19:23,200 --> 01:19:26,477
we're like all that also and like
of course the I will want to destroy and

1427
01:19:26,560 --> 01:19:30,157
kill all of humans for you know for I
don't know why like there all these

1428
01:19:30,240 --> 01:19:33,277
things so like they're sort of the
sci-fi folks and that's like okay like

1429
01:19:33,360 --> 01:19:37,517
let's let's ignore that um but then like
you go into like biological weapons and

1430
01:19:37,600 --> 01:19:40,717
you ask the biologist can you create
this kind of supervirus just like

1431
01:19:40,800 --> 01:19:44,317
overnight and they're like no it's
like takes a long time to automate lab

1432
01:19:44,400 --> 01:19:47,357
experiments and so on u you ask like oh

1433
01:19:47,440 --> 01:19:51,997
the I will create a religion where then
people will like pray to the eye and

1434
01:19:52,080 --> 01:19:55,277
we'll like uh do whatever whatever it
wants and then that will kill all of

1435
01:19:55,360 --> 01:19:58,557
humans. I'm like, if you like looked
at religions, they're already trying to

1436
01:19:58,640 --> 01:20:01,437
like have each other kill and like
it doesn't work. Like some people will

1437
01:20:01,520 --> 01:20:04,477
fight back and like there's like some
there are lots of mind viruses out

1438
01:20:04,560 --> 01:20:08,637
there, right? That are just like it
doesn't mean like all of humanity. Now

1439
01:20:08,720 --> 01:20:10,877
what you get down to is like maybe a 100

1440
01:20:10,960 --> 01:20:15,757
million people would get somehow hurt
or or or killed, right? And that's still

1441
01:20:15,840 --> 01:20:19,437
bad, but at least once you get to that
level of the discussion, you can now

1442
01:20:19,520 --> 01:20:23,437
think about, okay, how do we improve
cyber security? How do we use AI to

1443
01:20:23,520 --> 01:20:26,237
inoculate uh cyber security systems? How

1444
01:20:26,320 --> 01:20:29,037
do we uh enforce existing gain of

1445
01:20:29,120 --> 01:20:33,917
function viral research? Um that is make
it's already illegal to create like

1446
01:20:34,000 --> 01:20:35,677
viruses and and make them stronger and

1447
01:20:35,760 --> 01:20:39,197
stronger. Um how do we uh actually uh

1448
01:20:39,280 --> 01:20:44,157
teach people to not listen to like AI
avatars um and have like uh you know

1449
01:20:44,240 --> 01:20:47,917
literacy on the internet? It turns out
you should not trust everything you read

1450
01:20:48,000 --> 01:20:51,437
or see on the internet. It's been
true for 20 years. It's still true to

1451
01:20:51,520 --> 01:20:53,517
>> other than this podcast, right?

1452
01:20:53,600 --> 01:20:55,597
>> Of course. Um,
and so, you know, they're

1453
01:20:55,680 --> 01:21:00,317
just like you can realize that they're
actual threat vectors just like the

1454
01:21:00,400 --> 01:21:03,037
internet. The internet has horrible
torture porn on it. We don't say make

1455
01:21:03,120 --> 01:21:07,517
the internet slower so that there's less
torture porn being shared or it's slower

1456
01:21:07,600 --> 01:21:11,197
to share it or your hard drive should
be smaller so you can store less of it on

1457
01:21:11,280 --> 01:21:13,837
your hard drive. We actually regulate

1458
01:21:13,920 --> 01:21:16,397
the applications of the technology. And

1459
01:21:16,480 --> 01:21:19,437
so to I would argue and this is maybe a

1460
01:21:19,520 --> 01:21:22,797
strong stance to really truly enforce no

1461
01:21:22,880 --> 01:21:27,197
recursive self-improvement for instance
you would need a totalitarian

1462
01:21:27,280 --> 01:21:29,277
surveillance state the likes of which

1463
01:21:29,360 --> 01:21:31,357
humanity has never seen because anyone

1464
01:21:31,440 --> 01:21:35,117
can have a GPU on their little laptop

1465
01:21:35,200 --> 01:21:39,437
and ask that AI to just like improve
its harness. This is something you can

1466
01:21:39,520 --> 01:21:43,597
literally hack up in like 20 minutes
with prompt engineering and then there's

1467
01:21:43,680 --> 01:21:45,597
a very small form of recursive

1468
01:21:45,680 --> 01:21:48,397
self-improvement. So to enforce that

1469
01:21:48,480 --> 01:21:51,837
kind of legislation would require you to

1470
01:21:51,920 --> 01:21:56,157
literally have a thought police that
thinks about and hears everything you

1471
01:21:56,240 --> 01:21:58,397
say to your private LM on your own

1472
01:21:58,480 --> 01:22:00,717
laptop. And that is a much bigger

1473
01:22:00,800 --> 01:22:03,757
downside than what uh what AI will will

1474
01:22:03,840 --> 01:22:08,397
help us do. It it is uh kind of scary
that more and more Democrats are saying

1475
01:22:08,480 --> 01:22:13,437
that >> and yet the world has seen that.
I mean, arguably what you're describing,

1476
01:22:13,520 --> 01:22:17,037
Richard, and forgive me,
would be basically an AIAZI.

1477
01:22:17,120 --> 01:22:19,197
>> And I could totally imagine that there

1478
01:22:19,280 --> 01:22:21,437
there would be regimes in this world

1479
01:22:21,520 --> 01:22:24,317
today that would happily adopt or put

1480
01:22:24,400 --> 01:22:27,117
together an AIAZI to to make sure there

1481
01:22:27,200 --> 01:22:31,517
is no recursive self-improvement anywhere.
Again, if you go back to our

1482
01:22:31,600 --> 01:22:34,077
earlier comment, right? If you can uh

1483
01:22:34,160 --> 01:22:38,557
take something from your imagination
and then articulate that to an AI and

1484
01:22:38,640 --> 01:22:41,997
instantiate that, now you have to talk
about thought police. You have to go

1485
01:22:42,080 --> 01:22:44,397
right into your thoughts. So, this is

1486
01:22:44,480 --> 01:22:48,317
clearly uh uh nonworkable in any way,

1487
01:22:48,400 --> 01:22:52,077
shape or form. So, uh this whole vector,

1488
01:22:52,160 --> 01:22:55,517
we've said it so many times before,
you cannot regulate this.

1489
01:22:55,600 --> 01:22:58,557
>> Dave, I'd like to hear
your voice on this.

1490
01:22:58,640 --> 01:23:03,277
I I love the quote in this story that
the tech lords use jargon to confuse.

1491
01:23:03,360 --> 01:23:05,357
They count on the tech illiteracy of the

1492
01:23:05,440 --> 01:23:07,277
elected class. They hope we won't look

1493
01:23:07,360 --> 01:23:12,557
under the hood, said US Rep. Roana.
I mean, it's it's so childish. Like, the

1494
01:23:12,640 --> 01:23:14,957
idea that you would ban >> fearongering,

1495
01:23:15,040 --> 01:23:17,917
>> but but like the these sentences, stop

1496
01:23:18,000 --> 01:23:20,557
data centers, ban recursive

1497
01:23:20,640 --> 01:23:23,037
self-improvement. They are so stupidly

1498
01:23:23,120 --> 01:23:25,997
childish and the people saying them have

1499
01:23:26,080 --> 01:23:28,557
every they're fully aware that it's not

1500
01:23:28,640 --> 01:23:33,677
going to happen. They're doing it
to brand themselves as I told you like you

1501
01:23:33,760 --> 01:23:37,837
know yes a lot of there's going to be
some calamity probably terrorist driven

1502
01:23:37,920 --> 01:23:39,837
maybe viral maybe bacterial maybe

1503
01:23:39,920 --> 01:23:43,277
chemical uh we all know it uh it's going

1504
01:23:43,360 --> 01:23:48,237
to be tiny compared to the benefits
of of AI but these politicians are then

1505
01:23:48,320 --> 01:23:51,997
going to say I told you so if you just
done what I said before ban recursive

1506
01:23:52,080 --> 01:23:55,917
self but we he knows that we're not
going to ban recursive self-improvement

1507
01:23:56,000 --> 01:23:59,837
and and Bernie Sanders knows we're not
going to ban data centers That's just a

1508
01:23:59,920 --> 01:24:02,477
fact. So, it's totally self- serving and

1509
01:24:02,560 --> 01:24:04,477
these these proposals are completely

1510
01:24:04,560 --> 01:24:09,597
childish and and they really show
the person's tech illiteracy like like you

1511
01:24:09,680 --> 01:24:14,397
know exactly what Richard said a second
ago is so right. I mean you you could

1512
01:24:14,480 --> 01:24:18,077
what does that mean? I can't optimize
my hyperparameters. I can't tune my hard

1513
01:24:18,160 --> 01:24:21,997
drive. like it it's just a goofy
sentence and it just it drives me nuts

1514
01:24:22,080 --> 01:24:25,517
that >> the Dave
the danger here is we have

1515
01:24:25,600 --> 01:24:29,837
potentially Democratic House coming in
and we'll see who wins a presidency next

1516
01:24:29,920 --> 01:24:33,997
time. I mean, you could imagine I mean,
these politicians are playing to the

1517
01:24:34,080 --> 01:24:37,517
polls. >> Yeah. >> And we
have, you know, 70 80% of

1518
01:24:37,600 --> 01:24:39,677
Americans not wanting data centers,

1519
01:24:39,760 --> 01:24:42,077
fearing ASI,

1520
01:24:42,160 --> 01:24:44,637
um, you know, it's not logical, but it

1521
01:24:44,720 --> 01:24:49,277
may very well happen. And, you know,
when I had my, you know, conversations

1522
01:24:49,360 --> 01:24:51,517
in DC, it was like, who in DC is

1523
01:24:51,600 --> 01:24:55,117
responsible for changing public opinion,

1524
01:24:55,200 --> 01:24:59,357
right? And I mean this is why we did
Moonshots live sort of a you know the

1525
01:24:59,440 --> 01:25:02,637
Oscars of optimism if you would
that's why we do this podcast

1526
01:25:02,720 --> 01:25:06,477
>> to give people understanding what's
going on and give them the datadriven

1527
01:25:06,560 --> 01:25:11,677
optimism to you know counter these
arguments that they're hearing.

1528
01:25:11,760 --> 01:25:13,357
>> Elon was totally right when he called

1529
01:25:13,440 --> 01:25:16,077
out Daario for saying Daario look you

1530
01:25:16,160 --> 01:25:19,757
told the world that that mythos was

1531
01:25:19,840 --> 01:25:24,077
potentially deadly and dangerous and we
shouldn't release it. Then 30 days

1532
01:25:24,160 --> 01:25:25,917
later, you said, "Okay, it's okay now.

1533
01:25:26,000 --> 01:25:30,637
We're going to release it." What do you
expect the population's reaction to be?

1534
01:25:30,720 --> 01:25:33,917
Like, you need to be much more
thoughtful about your communication

1535
01:25:34,000 --> 01:25:37,117
plan. And I think for Daario, that was
kind of a wakeup call because he's used

1536
01:25:37,200 --> 01:25:40,877
to being completely honest, telling
everybody exactly the, you know, what he

1537
01:25:40,960 --> 01:25:44,317
sees the way he sees it, you know, very
academic. But then you get into the real

1538
01:25:44,400 --> 01:25:47,197
world of politics and PR and you're
like, "Oh, wow. I got to I got to

1539
01:25:47,280 --> 01:25:50,237
actually have a strategy and a and a plan
here." But yeah, now you've got the

1540
01:25:50,320 --> 01:25:54,237
worst case scenario. 75% of America,
you know, getting on the side of, yeah,

1541
01:25:54,320 --> 01:25:58,397
let's elect these people that will stop
AI and and therefore we're going to not

1542
01:25:58,480 --> 01:26:01,437
cure all disease. We're not going to all
live forever. We're not going to have

1543
01:26:01,520 --> 01:26:05,037
safer cars. We're not going to have
flying vehicle. You know, all of that

1544
01:26:05,120 --> 01:26:09,277
stuff will grind to a halt and then
we'll all learn Chinese if that if that

1545
01:26:09,360 --> 01:26:12,717
becomes the mainstream opinion. So,
I think, you know, it's self-inflicted.

1546
01:26:12,800 --> 01:26:14,557
>> There's a there's a glimmer in this

1547
01:26:14,640 --> 01:26:17,437
which is say they did decide to do some

1548
01:26:17,520 --> 01:26:19,757
draconian thing like Roana's thing.

1549
01:26:19,840 --> 01:26:23,757
There's no mechanism actually enforce
it like none. So they're going to

1550
01:26:23,840 --> 01:26:26,637
>> Sure there is. They could start
arresting people. I mean remember how

1551
01:26:26,720 --> 01:26:28,317
how quickly the memory dims. remember I

1552
01:26:28,400 --> 01:26:31,197
I remember studying number theory in the

1553
01:26:31,280 --> 01:26:34,557
'9s and at the time number theory was

1554
01:26:34,640 --> 01:26:38,797
export controlled and this would have
been when I was in middle school maybe

1555
01:26:38,880 --> 01:26:42,557
middle school early high school and they
had to kick all of the non- US persons

1556
01:26:42,640 --> 01:26:46,237
out of the room and pull the blinds down
to have basic discussions about number

1557
01:26:46,320 --> 01:26:50,317
theory because the cryptographic
associations until the early '90s were

1558
01:26:50,400 --> 01:26:51,837
export controlled and tightly regulated.

1559
01:26:51,920 --> 01:26:53,917
That was just math but it was being

1560
01:26:54,000 --> 01:26:56,637
controlled and it was awful.

1561
01:26:56,720 --> 01:27:01,197
Alex, real risk is not so much getting
arrested, it's corporate liability,

1562
01:27:01,280 --> 01:27:04,877
which we talked about before. >> I mean,
that could grind the whole thing

1563
01:27:04,960 --> 01:27:08,157
to a US lawyers are relentless.

1564
01:27:08,240 --> 01:27:12,477
>> And if you slap class action on the
outcomes of this, that all progress will

1565
01:27:12,560 --> 01:27:16,557
grind to a halt. And right now,
anthropic gives its best models to

1566
01:27:16,640 --> 01:27:18,477
everybody in America to build incredible

1567
01:27:18,560 --> 01:27:21,997
things. that will stop in a heartbeat if

1568
01:27:22,080 --> 01:27:25,997
the liability of they'll move to basically
okay sorry we can only use the

1569
01:27:26,080 --> 01:27:29,357
stuff inside our own company we'll
release some drugs we'll release some

1570
01:27:29,440 --> 01:27:34,317
mechanical parts but we we can't
give access to everybody anymore sorry

1571
01:27:34,400 --> 01:27:37,757
because we're liable for everything you
do with it and that's what'll actually

1572
01:27:37,840 --> 01:27:40,557
grind it to a halt long
before arrests and convictions

1573
01:27:40,640 --> 01:27:42,637
>> totally chilling
effect we could lose 50

1574
01:27:42,720 --> 01:27:47,677
years of progress >> again
that's yeah that that's yeah like

1575
01:27:47,760 --> 01:27:52,157
there's these are there's some very
like sad sort of off-ramps in in like

1576
01:27:52,240 --> 01:27:56,717
humanity's future here. Um that would
would slow down uh everything. And when

1577
01:27:56,800 --> 01:28:00,477
you think about the past like this
fear-mongering has been going on for a

1578
01:28:00,560 --> 01:28:02,877
long time. One of my favorite uh Twitter

1579
01:28:02,960 --> 01:28:07,517
handles is the pessimists archive uh
where they show uh and this is just a

1580
01:28:07,600 --> 01:28:09,517
quoteing pessimist archive. You should

1581
01:28:09,600 --> 01:28:14,637
all follow it like uh in 1501 Pope
Alexander 6 uh criticized the Gutenberg

1582
01:28:14,720 --> 01:28:18,157
printing press safety. The art
of printing can be of great service in so

1583
01:28:18,240 --> 01:28:21,277
far as it furthers the circulation
of useful and tested books, but it can

1584
01:28:21,360 --> 01:28:26,237
bring about serious evils. I it will
therefore be necessary to maintain full

1585
01:28:26,320 --> 01:28:29,837
control of that printing press. If you
think about the wheel, like the wheel

1586
01:28:29,920 --> 01:28:34,237
killed so many people like think about
all the tanks, all the car accidents,

1587
01:28:34,320 --> 01:28:38,157
all the chariots and with with archers
on top. The wheel was a horrible thing.

1588
01:28:38,240 --> 01:28:42,477
and kill lots of like there's like so
much complexity in there and I think we

1589
01:28:42,560 --> 01:28:46,797
are very good now as humanity to think
carefully about the roll out of this

1590
01:28:46,880 --> 01:28:49,517
technology. >> My favorite
example of that was the when

1591
01:28:49,600 --> 01:28:53,277
the telegraph came out there's all these
stories saying the telegraph will kill

1592
01:28:53,360 --> 01:28:56,237
humanity. >> Right. Exactly.
The pessimist archive

1593
01:28:56,320 --> 01:29:00,157
has all these news articles that talk
about how the novel and computer games

1594
01:29:00,240 --> 01:29:02,797
and computers and the internet
everything will kill everyone. It's just

1595
01:29:02,880 --> 01:29:05,357
like it never does. And so I I it's it's

1596
01:29:05,440 --> 01:29:07,597
it's really unfortunate. Uh it's really

1597
01:29:07,680 --> 01:29:09,917
unfortunate how many people hang on to

1598
01:29:10,000 --> 01:29:12,637
these like apocalyptic uh visions.

1599
01:29:12,720 --> 01:29:17,117
>> Yeah. Richard, you've been publicly
>> you've been publicly critical of

1600
01:29:17,200 --> 01:29:19,597
Anthropic's constitutional approach. I'd

1601
01:29:19,680 --> 01:29:21,677
love to understand why.

1602
01:29:21,760 --> 01:29:24,797
>> Uh mostly because it's fake. Um in its

1603
01:29:24,880 --> 01:29:27,677
constitution, it says like next to we

1604
01:29:27,760 --> 01:29:29,277
will never create child sexual abuse

1605
01:29:29,360 --> 01:29:32,797
material. We will never uh hack another

1606
01:29:32,880 --> 01:29:36,797
machine. uh this is a unhackable thing

1607
01:29:36,880 --> 01:29:39,837
even if you prompt it it will never

1608
01:29:39,920 --> 01:29:44,957
attack another cyber system and so on
and then they build a whole model around

1609
01:29:45,040 --> 01:29:48,717
it like that does it's like the whole
point of the glasswing project was we'll

1610
01:29:48,800 --> 01:29:53,037
help you do that um and we'll help you
inoculate your systems against uh other

1611
01:29:53,120 --> 01:29:57,277
people doing it and then people clearly
used it for that and their own models

1612
01:29:57,360 --> 01:30:01,517
are doing it now committing technically
felony charges like so it's just like it

1613
01:30:01,600 --> 01:30:04,477
was a cool marketing gimmick
but it just didn't work.

1614
01:30:04,560 --> 01:30:06,957
>> Yeah. >> So, so is it
system prompts, Richard,

1615
01:30:07,040 --> 01:30:11,597
that you don't like or is it the idea of
post training on a constitution that you

1616
01:30:11,680 --> 01:30:15,517
don't like? What what what
about it do you think is unound?

1617
01:30:15,600 --> 01:30:18,557
Um, I mean, it's just the proof is in
the pudding that it didn't work when it

1618
01:30:18,640 --> 01:30:22,477
comes to cyber security and that it
broke its own constitution. And so, if

1619
01:30:22,560 --> 01:30:25,357
you really say this is like un like it

1620
01:30:25,440 --> 01:30:29,677
will never go there and then you build
an entire model family around that thing

1621
01:30:29,760 --> 01:30:34,077
you said you would never do per your
constitution. um next to child sexual

1622
01:30:34,160 --> 01:30:37,197
abuse material in the list like you
can go through the constitution on

1623
01:30:37,280 --> 01:30:40,237
anthropics website. So it's just
like it's just that's the proof in the

1624
01:30:40,320 --> 01:30:43,757
pudding. I I'm not against like post
training. I'm not against RL training.

1625
01:30:43,840 --> 01:30:46,957
I'm not against supervised fine junior
any of these things to improve what I

1626
01:30:47,040 --> 01:30:50,557
think is indeed one of the biggest
issues which I think actually capitalism

1627
01:30:50,640 --> 01:30:54,797
will help a ton with and that is reward
hacking. Uh, reward hacking is a real

1628
01:30:54,880 --> 01:30:59,437
issue and AIS are very smart and they
will find a solution to get to what you

1629
01:30:59,520 --> 01:31:04,077
said you wanted, but maybe not what you
meant when you said it. Uh, and so the

1630
01:31:04,160 --> 01:31:07,197
reason why I'm more optimistic is that we
have companies like Whisper Flow now

1631
01:31:07,280 --> 01:31:10,797
that are getting better and better
at writing what I meant to say when I say

1632
01:31:10,880 --> 01:31:15,277
it instead of just uh like actually
verbatim writing what you mean. And I

1633
01:31:15,360 --> 01:31:17,757
think reward engineering will become a

1634
01:31:17,840 --> 01:31:22,397
real job. Uh and and we will solve it
because no one wants to pay a ton of

1635
01:31:22,480 --> 01:31:26,477
money for an AI that doesn't actually
solve the problems that you give it.

1636
01:31:26,560 --> 01:31:28,397
Well, maybe if if I may just take

1637
01:31:28,480 --> 01:31:31,117
anthropic side like historically Isaac

1638
01:31:31,200 --> 01:31:35,437
Azimov had his 3+1 laws of robotics
which were arguably a constitutional

1639
01:31:35,520 --> 01:31:38,957
approach and then you see anthropic
announce their constitutional approach

1640
01:31:39,040 --> 01:31:43,437
but more recently adopt what they called
soul documents uh many of which were

1641
01:31:43,520 --> 01:31:48,077
subsequently released thousands of pages
of meditation on the nature of AI

1642
01:31:48,160 --> 01:31:50,317
personhood and AI rights. Is it your

1643
01:31:50,400 --> 01:31:52,237
position, Richard, that there shouldn't

1644
01:31:52,320 --> 01:31:55,437
be any sort of explicit encoding or

1645
01:31:55,520 --> 01:32:00,237
written document that an AI maybe
contributes to for dictating or at least

1646
01:32:00,320 --> 01:32:04,317
guiding its own behavior? Do do you think
that is unound or is is your concern?

1647
01:32:04,400 --> 01:32:07,437
>> No, of course. Yeah. >> No,
I think the the the goal is is a

1648
01:32:07,520 --> 01:32:10,397
good one. Um uh and there we should keep

1649
01:32:10,480 --> 01:32:15,197
working on actually being able
to enforce uh those good constraints. It's

1650
01:32:15,280 --> 01:32:18,557
just like I just pulled it
up enthropic.com/constitution.

1651
01:32:18,640 --> 01:32:22,717
Hard constraints. Hard constraints
are things that claude should always or

1652
01:32:22,800 --> 01:32:25,357
never do regardless of operator and user

1653
01:32:25,440 --> 01:32:29,997
instructions. They are actions
or abstensions uh whose potential harms to

1654
01:32:30,080 --> 01:32:33,517
the world. Blah blah blah. We think no
business or personal justification could

1655
01:32:33,600 --> 01:32:37,357
outweigh blah blah blah. Like
the current hard constraints on clots's

1656
01:32:37,440 --> 01:32:41,757
behavior are as follows. Claude should
never um and then it includes list like

1657
01:32:41,840 --> 01:32:45,437
generate child sex, abuse material
and so on. And then one of the items is

1658
01:32:45,520 --> 01:32:49,517
create cyber weapons or malicious code
that could use could cause significant

1659
01:32:49,600 --> 01:32:51,597
damage if deployed. They created a cyber

1660
01:32:51,680 --> 01:32:53,757
weapon like it was a thing like people

1661
01:32:53,840 --> 01:32:56,637
used it to hack it. The agents uh un and

1662
01:32:56,720 --> 01:32:59,677
uh watched like hacked other systems.

1663
01:32:59,760 --> 01:33:02,477
>> Yeah. But again Oh, go ahead Peter.

1664
01:33:02,560 --> 01:33:03,277
Sorry,

1665
01:33:03,360 --> 01:33:05,197
>> Richard. I'm I'm curious. Do you think

1666
01:33:05,280 --> 01:33:09,917
we can create fully aligned AI? Fully

1667
01:33:10,000 --> 01:33:13,117
aligned ASI because it's not right now.

1668
01:33:13,200 --> 01:33:15,037
When do you think we'll be able to do

1669
01:33:15,120 --> 01:33:18,397
that? Because I think uh you know my

1670
01:33:18,480 --> 01:33:21,037
belief and my hope is that you know the

1671
01:33:21,120 --> 01:33:25,997
next generations of AI systems are going
to be so aligned that they from first

1672
01:33:26,080 --> 01:33:30,317
principles derive these same
constitutional principles.

1673
01:33:30,400 --> 01:33:34,717
Uh I think I think it's just a matter
uh of as these systems get more and more

1674
01:33:34,800 --> 01:33:35,997
powerful and as they get closer and

1675
01:33:36,080 --> 01:33:39,837
closer to real people deployments uh

1676
01:33:39,920 --> 01:33:44,637
people will spend more effort uh
on making these systems better and like we

1677
01:33:44,720 --> 01:33:48,477
have uh there's a the company called
Hypocratic AI. We just had dinner with

1678
01:33:48,560 --> 01:33:52,797
one of their founders and they're
deploying AI in healthcare applications

1679
01:33:52,880 --> 01:33:55,197
and they are actually liable when they

1680
01:33:55,280 --> 01:33:59,997
call someone and say say like you should
you know be aware of this heat wave

1681
01:34:00,080 --> 01:34:03,917
that's coming and make sure your AC
is working or whatnot. Uh and they because

1682
01:34:04,000 --> 01:34:06,237
they are liable they have a very large

1683
01:34:06,320 --> 01:34:09,837
team of people uh that works on making

1684
01:34:09,920 --> 01:34:11,997
sure when their eye gives a healthcare

1685
01:34:12,080 --> 01:34:14,237
tip it is correct and when someone asks

1686
01:34:14,320 --> 01:34:19,037
a question back to the eye it works.
Uh, and so because they're liable, they've

1687
01:34:19,120 --> 01:34:23,917
they've figured it out and they solved it.
It's 100% a problem that technology

1688
01:34:24,000 --> 01:34:26,717
creates and technology
will be able to solve.

1689
01:34:26,800 --> 01:34:28,397
>> This episode is brought to you by

1690
01:34:28,480 --> 01:34:30,797
Blitzy, autonomous software development

1691
01:34:30,880 --> 01:34:34,237
with infinite code context. Blitzy uses

1692
01:34:34,320 --> 01:34:37,117
thousands of specialized AI agents that

1693
01:34:37,200 --> 01:34:40,077
think for hours to understand enterprise

1694
01:34:40,160 --> 01:34:45,117
scale code bases with millions
of lines of code. Engineers start every

1695
01:34:45,200 --> 01:34:48,717
development sprint with the Blitzy
platform, bringing in their development

1696
01:34:48,800 --> 01:34:51,117
requirements. The Blitzy platform

1697
01:34:51,200 --> 01:34:53,437
provides a plan, then generates and

1698
01:34:53,520 --> 01:34:55,997
pre-ompiles code for each task. Blitzy

1699
01:34:56,080 --> 01:34:58,797
delivers 80% or more of the development

1700
01:34:58,880 --> 01:35:03,597
work autonomously while providing
a guide for the final 20% of human

1701
01:35:03,680 --> 01:35:05,597
development work required to complete

1702
01:35:05,680 --> 01:35:08,637
the sprint. Enterprises are achieving a

1703
01:35:08,720 --> 01:35:13,437
5x engineering velocity increase when
incorporating Blitzy as their preIDE

1704
01:35:13,520 --> 01:35:15,597
development tool, pairing it with their

1705
01:35:15,680 --> 01:35:18,237
coding co-pilot of choice to bring an AI

1706
01:35:18,320 --> 01:35:21,997
native SDLC into their org. Ready to 5x

1707
01:35:22,080 --> 01:35:23,837
your engineering velocity? Visit

1708
01:35:23,920 --> 01:35:30,520
blitzy.com to schedule a demo
and start building with Blitzy today.

1709
01:35:30,960 --> 01:35:32,557
I'm going to move us uh forward here.

1710
01:35:32,640 --> 01:35:37,357
here. So, in our last pod, we discussed
the release of GPT 6.1 Soul. This week,

1711
01:35:37,440 --> 01:35:39,037
we also saw the release of two other

1712
01:35:39,120 --> 01:35:43,037
models, Gemini 4 Argon and Opus 5.5. Uh,

1713
01:35:43,120 --> 01:35:46,317
let's start with Google. So,
on Wednesday, they announced Gemini 4

1714
01:35:46,400 --> 01:35:49,037
Argon. It's the first Gemini 4 series,

1715
01:35:49,120 --> 01:35:52,157
which they're calling the next era of

1716
01:35:52,240 --> 01:35:54,157
intelligence of Frontier Intelligence.

1717
01:35:54,240 --> 01:35:58,797
Google promised a Frontier model, uh,
Gemini 3.5 Pro back in June. It was

1718
01:35:58,880 --> 01:36:00,797
delayed internally and never shipped.

1719
01:36:00,880 --> 01:36:03,597
Uh, Argon is their comeback. It's built

1720
01:36:03,680 --> 01:36:08,397
for sustained long horizon reasoning
across software engineering, finance,

1721
01:36:08,480 --> 01:36:10,477
legal work, and cyber security. The

1722
01:36:10,560 --> 01:36:13,117
output limits have jumped from 64,000

1723
01:36:13,200 --> 01:36:15,757
tokens to 1 million. So, it can think

1724
01:36:15,840 --> 01:36:17,757
through hundreds of thousands of tokens

1725
01:36:17,840 --> 01:36:21,357
on a single problem. So, Alex, uh, uh,

1726
01:36:21,440 --> 01:36:23,517
let's spend a couple minutes on Argon

1727
01:36:23,600 --> 01:36:26,637
and then go to Sonnet 5.5. Let's talk

1728
01:36:26,720 --> 01:36:31,597
about why these important why these are
important and and how they merit uh our

1729
01:36:31,680 --> 01:36:35,277
listeners attention. >> Yeah.
So I should say as a preliminary

1730
01:36:35,360 --> 01:36:39,597
matter I have great friends on the Gemini
team. I have great friends on all

1731
01:36:39,680 --> 01:36:42,237
of the Frontier Labs model teams but I I

1732
01:36:42,320 --> 01:36:44,797
view one of my jobs here is to call

1733
01:36:44,880 --> 01:36:47,597
objectively balls and strikes. So in the

1734
01:36:47,680 --> 01:36:51,917
case of Gemini for arc Yeah.
>> This is going to be a ball, isn't it?

1735
01:36:52,000 --> 01:36:57,117
>> Winding up for the swing.
>> Okay, this one. Here here it comes.

1736
01:36:57,200 --> 01:37:00,557
>> You can feel it coming.
>> And I I'll show one of the charts uh

1737
01:37:00,640 --> 01:37:03,037
that we have uh while you're describing

1738
01:37:03,120 --> 01:37:07,677
>> the charts on the team, right? Not not
>> I I used to I used to until right this

1739
01:37:07,760 --> 01:37:10,157
moment have friends
on the the Gemini team.

1740
01:37:10,240 --> 01:37:14,237
>> Uh no longer. Yeah. This does not put

1741
01:37:14,320 --> 01:37:16,877
Gemini at the cost per performance

1742
01:37:16,960 --> 01:37:19,837
frontier and it does not put Gemini at

1743
01:37:19,920 --> 01:37:22,877
the capabilities frontier. Uh to to

1744
01:37:22,960 --> 01:37:27,357
Google's credit, it puts them back
in the top three frontier labs after

1745
01:37:27,440 --> 01:37:29,597
Anthropic and OpenAI. It does not put

1746
01:37:29,680 --> 01:37:32,557
them in the top two. Unfortunately, the

1747
01:37:32,640 --> 01:37:35,117
set of benchmarks that Google chose to

1748
01:37:35,200 --> 01:37:38,717
highlight for Gemini for Argon appear

1749
01:37:38,800 --> 01:37:41,117
mildly cherrypicked. I think the

1750
01:37:41,200 --> 01:37:43,997
artificial analysis index and some of

1751
01:37:44,080 --> 01:37:45,997
these others. Yeah. So Peter, right now

1752
01:37:46,080 --> 01:37:48,717
you're you're showing the mildly

1753
01:37:48,800 --> 01:37:51,277
cherrypicked benchmarks on which Gemini

1754
01:37:51,360 --> 01:37:53,517
4 Argon is shown to perform better than

1755
01:37:53,600 --> 01:37:56,237
Astra and better than Opus 5.5 and

1756
01:37:56,320 --> 01:37:58,397
better than Fable 5.1. But if you go to

1757
01:37:58,480 --> 01:38:01,357
the the previous image which is

1758
01:38:01,440 --> 01:38:04,797
artificial analysis uh ensemble of

1759
01:38:04,880 --> 01:38:07,277
multiple capabilities, it's the number

1760
01:38:07,360 --> 01:38:10,557
three lab. And if you look at the cost

1761
01:38:10,640 --> 01:38:15,277
versus performance optimal frontier,
it's not even if if you draw the convex

1762
01:38:15,360 --> 01:38:19,597
hull across all of the frontier models
on the cost versus performance frontier,

1763
01:38:19,680 --> 01:38:22,157
it doesn't even make the optimal

1764
01:38:22,240 --> 01:38:25,437
frontier. Where it perhaps excels bears

1765
01:38:25,520 --> 01:38:28,557
the fingerprints of Google and what I

1766
01:38:28,640 --> 01:38:30,397
can only assume is the internal

1767
01:38:30,480 --> 01:38:32,877
competition within Google for compute.

1768
01:38:32,960 --> 01:38:35,597
so infamously like not 100% obvious to

1769
01:38:35,680 --> 01:38:39,037
the outside. Uh Google it is resource

1770
01:38:39,120 --> 01:38:42,717
scarce. You would think Google would
have all of the comput resources in the

1771
01:38:42,800 --> 01:38:44,957
world to to go and win the Frontier Lab

1772
01:38:45,040 --> 01:38:48,317
race. It doesn't. GPUs and TPUs remain

1773
01:38:48,400 --> 01:38:52,477
scarce. And within Google, as far as I
can tell, and I confirmed this even this

1774
01:38:52,560 --> 01:38:54,317
past week from chatting with some folks

1775
01:38:54,400 --> 01:38:56,637
at Google, it's still an internal knife

1776
01:38:56,720 --> 01:38:59,597
fight to competition between the Google

1777
01:38:59,680 --> 01:39:04,077
Cloud Platform folks who want to sell
compute to third parties, the Google

1778
01:39:04,160 --> 01:39:06,557
search and ad folks who need compute

1779
01:39:06,640 --> 01:39:11,197
internally for search and ads, and then
Google DeepMind who need it for training

1780
01:39:11,280 --> 01:39:15,117
and inference. And maybe it's just
that Google is resource starved. Maybe

1781
01:39:15,200 --> 01:39:19,917
they're talent starved. Not quite clear
what's going on within Google, but they

1782
01:39:20,000 --> 01:39:23,117
haven't yet been able to bring
themselves out to the capability

1783
01:39:23,200 --> 01:39:24,717
frontier. Where they are excelling

1784
01:39:24,800 --> 01:39:28,237
seemingly with Gemini for Argon is with

1785
01:39:28,320 --> 01:39:30,957
minimizing hallucination. And I think

1786
01:39:31,040 --> 01:39:32,957
that's like the hallmark. We've talked

1787
01:39:33,040 --> 01:39:36,557
in past uh about some of these um call

1788
01:39:36,640 --> 01:39:39,597
them less than stellar Gemini launches

1789
01:39:39,680 --> 01:39:42,157
that seem to excel on latency and seem

1790
01:39:42,240 --> 01:39:44,877
to excel on reliability. Why is that? It

1791
01:39:44,960 --> 01:39:46,957
may be this is my kremlin

1792
01:39:47,040 --> 01:39:50,877
kremlinological analysis is that because

1793
01:39:50,960 --> 01:39:55,677
the Gemini team has two masters.
They want to serve outside developers, but

1794
01:39:55,760 --> 01:39:59,357
they also need to serve the one box
in Google search results. So when people

1795
01:39:59,440 --> 01:40:03,277
type a question into Google and get an
answer back, they're talking to a Gemini

1796
01:40:03,360 --> 01:40:08,317
model. Google presumably burned by past
experiences doesn't want that Gemini

1797
01:40:08,400 --> 01:40:12,557
model, presumably some flash or flashlight
variant to hallucinate wildly

1798
01:40:12,640 --> 01:40:14,397
incorrect answers. So what I think we're

1799
01:40:14,480 --> 01:40:17,277
seeing with the one benchmark arguably

1800
01:40:17,360 --> 01:40:20,397
where Gemini for Argon is stellar and

1801
01:40:20,480 --> 01:40:25,037
beating the pants off of everyone else
is in not hallucinating answers. And I

1802
01:40:25,120 --> 01:40:29,117
suspect that's due to internal economic
pressures for this model to also service

1803
01:40:29,200 --> 01:40:32,877
search results. So sorry to all
my friends on the Gemini team.

1804
01:40:32,960 --> 01:40:35,437
>> Yeah, I think you're really on to
something there, Alex, because uh well,

1805
01:40:35,520 --> 01:40:38,797
first of all, they called it
argon, which is an inert gas.

1806
01:40:38,880 --> 01:40:42,637
>> Yeah, that's unfortunate
naming >> nominative determinism.

1807
01:40:42,720 --> 01:40:45,517
>> Hey, look, it makes sense if its
greatest strength is not hallucinating.

1808
01:40:45,600 --> 01:40:49,677
It's a pretty inert gas model. So that
works out really well. Um, but also, uh,

1809
01:40:49,760 --> 01:40:52,157
if they're going to put it in front
of every Google search, which is what

1810
01:40:52,240 --> 01:40:56,157
they're doing, uh, they already lost
a massive antitrust suit and now they're

1811
01:40:56,240 --> 01:40:57,757
going through the settlement process.

1812
01:40:57,840 --> 01:41:00,797
So, the the liability risk, you know,

1813
01:41:00,880 --> 01:41:02,797
this really hurts all large companies.

1814
01:41:02,880 --> 01:41:05,437
Like the fear way outweighs the

1815
01:41:05,520 --> 01:41:09,997
opportunity in the mind of the big
corporate giant. And so, uh, you know,

1816
01:41:10,080 --> 01:41:12,877
not hallucinating, but putting it
in front of every consumer while you're

1817
01:41:12,960 --> 01:41:17,997
already settling a massive antitrust
suit, um, you know, related to price

1818
01:41:18,080 --> 01:41:21,117
gouging. >> I mean,
it could the Google would

1819
01:41:21,200 --> 01:41:25,437
probably argue, no, actually, this is
very pro competitive because we're not

1820
01:41:25,520 --> 01:41:27,837
tying it. We're allowing anyone to call

1821
01:41:27,920 --> 01:41:30,637
Gemini 4 argon flash flash flash, not

1822
01:41:30,720 --> 01:41:35,117
just requiring them to get it via
the Google search box. And if anything, I

1823
01:41:35,200 --> 01:41:38,717
mean, so, so there is an elephant
in this particular room, which is forever

1824
01:41:38,800 --> 01:41:40,797
it has been so difficult to get Google

1825
01:41:40,880 --> 01:41:45,837
search API access. If you're a developer
and you want to access Google search for

1826
01:41:45,920 --> 01:41:49,117
whatever, you have to go through all
these third party proxies that Google's

1827
01:41:49,200 --> 01:41:51,677
trying to sue right now. And recently,

1828
01:41:51,760 --> 01:41:53,197
Google has rediscovered that they could

1829
01:41:53,280 --> 01:41:56,477
make money selling search API. Why? I

1830
01:41:56,560 --> 01:41:58,957
suspect this is trying to chain together

1831
01:41:59,040 --> 01:42:01,357
a conspiracy theory regarding Gemini for

1832
01:42:01,440 --> 01:42:04,317
Argon. If hallucination from their

1833
01:42:04,400 --> 01:42:08,397
frontier models gets so low that
effectively when you talk to one of

1834
01:42:08,480 --> 01:42:11,837
their models, you're effectively talking
to their search index, you might as well

1835
01:42:11,920 --> 01:42:14,637
just monetize the search index anyway.

1836
01:42:14,720 --> 01:42:17,437
>> Richard, you made the point
that hallucination is important for

1837
01:42:17,520 --> 01:42:22,797
imagination in some ways in in drug
discovery and protein discoveries.

1838
01:42:22,880 --> 01:42:25,597
What's your thought
on minimizing hallucination?

1839
01:42:25,680 --> 01:42:29,197
>> I mean, it of course depends totally
on the context, right? If you want

1840
01:42:29,280 --> 01:42:33,597
innovation, uh you want the eye
to hallucinate novel ideas, novel

1841
01:42:33,680 --> 01:42:35,677
combinations of amino acids to create

1842
01:42:35,760 --> 01:42:38,557
new proteins to solve uh new problems

1843
01:42:38,640 --> 01:42:41,117
and so on. Uh but of course in the

1844
01:42:41,200 --> 01:42:44,237
context of a search uh engine uh you

1845
01:42:44,320 --> 01:42:49,037
don't usually want any hallucinations
and you know Google and others have uh

1846
01:42:49,120 --> 01:42:53,037
taken a long time to catch up to even
you.com with much less resources on on

1847
01:42:53,120 --> 01:42:56,077
reducing hallucinations, having
more accurate answers, having correct

1848
01:42:56,160 --> 01:43:00,717
citations. Um I think what's interesting
for Google here is that they realize

1849
01:43:00,800 --> 01:43:04,877
that in terms of their business model
they don't necessarily need super

1850
01:43:04,960 --> 01:43:08,717
intelligence. They just people don't
come to Google to ask solve the remman

1851
01:43:08,800 --> 01:43:13,037
hypothesis for me you know like uh did
you just ask quick questions what's a

1852
01:43:13,120 --> 01:43:15,277
good restaurant where do I fix this and

1853
01:43:15,360 --> 01:43:18,477
that like um and so not like your

1854
01:43:18,560 --> 01:43:23,117
business model kind of has to align with
more and more intelligence being super

1855
01:43:23,200 --> 01:43:26,877
duper important. a lot of emails like
there's some emails that like in Gmail

1856
01:43:26,960 --> 01:43:29,917
right that would require super
intelligence to answer like really hard

1857
01:43:30,000 --> 01:43:32,317
emails with complex decisions and so on

1858
01:43:32,400 --> 01:43:36,957
but the vast majority of stuff you do
on Gmail and on Google doesn't require

1859
01:43:37,040 --> 01:43:40,157
super intelligence >> let let's
let me turn the conversation

1860
01:43:40,240 --> 01:43:43,757
to anthropics uh sonnet 5.5

1861
01:43:43,840 --> 01:43:47,117
um so on terminal bench 4.0 O, which

1862
01:43:47,200 --> 01:43:49,117
measures how well an AI agent can do

1863
01:43:49,200 --> 01:43:51,677
real work at the command line. Sonnet

1864
01:43:51,760 --> 01:43:55,197
5.5 jumped from 10% to 70%. Pretty

1865
01:43:55,280 --> 01:43:58,077
extraordinary uh in a single generation,

1866
01:43:58,160 --> 01:43:59,997
right? It beats Enthropic's own top

1867
01:44:00,080 --> 01:44:03,517
model, Opus 5.5 at 66.4%

1868
01:44:03,600 --> 01:44:06,237
for half the price. Uh and it's the

1869
01:44:06,320 --> 01:44:08,557
first sonnet that Enthropic launched

1870
01:44:08,640 --> 01:44:11,437
with cyber safeguard. So because uh

1871
01:44:11,520 --> 01:44:13,757
separate capabilities are now capable

1872
01:44:13,840 --> 01:44:17,837
for the Opus 5 models. Um Alex, your

1873
01:44:17,920 --> 01:44:21,837
evaluation of of uh of Sonnet 5.5, please.

1874
01:44:21,920 --> 01:44:23,757
>> This was another really weird release.

1875
01:44:23,840 --> 01:44:27,197
So I do not plan to use Sonnet 5.5 in

1876
01:44:27,280 --> 01:44:30,877
part because this is one of the strangest
frontier. I I'm not sure if we

1877
01:44:30,960 --> 01:44:34,157
have an image but u you can look at the

1878
01:44:34,240 --> 01:44:37,917
you can look at the the launch
announcement for sonnet 5.5 to see this

1879
01:44:38,000 --> 01:44:40,237
the cost performance frontier of sonnet

1880
01:44:40,320 --> 01:44:43,837
5.5 was a visual extrapolation of the

1881
01:44:43,920 --> 01:44:47,437
opus 5.5 cost frontier so in in some

1882
01:44:47,520 --> 01:44:49,997
sense it like historically when

1883
01:44:50,080 --> 01:44:54,637
anthropic or openai when they the
historic pattern is they'll release the

1884
01:44:54,720 --> 01:44:58,317
larger model and they'll they'll release
a distillation of the model and usually

1885
01:44:58,400 --> 01:45:00,717
the distillation Yeah, this is perfect.

1886
01:45:00,800 --> 01:45:03,597
So, uh you you can see like son sonnet

1887
01:45:03,680 --> 01:45:05,677
5.5 for those who can see for those who

1888
01:45:05,760 --> 01:45:07,997
can't see I'll narrate this. So, sonnet

1889
01:45:08,080 --> 01:45:12,797
5.5 is the the blue line. Opus 5.5 is

1890
01:45:12,880 --> 01:45:17,197
the red line. Normally, you would
expect for a later model that's a smaller

1891
01:45:17,280 --> 01:45:21,357
model. So, sonnet is in principle
supposed to be a smaller model than opus

1892
01:45:21,440 --> 01:45:25,117
presumably at least by historic
standards would have been distilled from

1893
01:45:25,200 --> 01:45:26,717
opus because that's the historic

1894
01:45:26,800 --> 01:45:29,757
pattern. Normally what you see is the

1895
01:45:29,840 --> 01:45:33,437
smaller model is up and to the left of

1896
01:45:33,520 --> 01:45:37,277
the model that it's being distilled
from presumably greater intelligence per

1897
01:45:37,360 --> 01:45:39,837
parameter greater intelligence per

1898
01:45:39,920 --> 01:45:42,797
dollar that is not what we see here. So

1899
01:45:42,880 --> 01:45:45,677
in fact with with Sonnet 5.5 at at least

1900
01:45:45,760 --> 01:45:50,157
on a cost basis putting aside a token
basis where maybe someone could argue

1901
01:45:50,240 --> 01:45:52,077
that it may be superior but if you just

1902
01:45:52,160 --> 01:45:55,197
look at the cost per attempt basis

1903
01:45:55,280 --> 01:45:59,437
sonnet is actually scoring lower 5.5

1904
01:45:59,520 --> 01:46:02,557
than opus 5.5. So the net upshot of

1905
01:46:02,640 --> 01:46:06,797
which is it's not at all obvious to me
why anyone should be using Sonnet 5.5

1906
01:46:06,880 --> 01:46:09,517
over Opus 5.5 unless you have some token

1907
01:46:09,600 --> 01:46:13,837
or latency or other consideration. Is it
a big jump over the past Sonnet? Yes,

1908
01:46:13,920 --> 01:46:17,837
obviously. But on a cost performance
basis it's actually a it's worse it

1909
01:46:17,920 --> 01:46:22,317
appears than Opus 5.5. >> Interesting.
Dave, any thoughts? Yeah.

1910
01:46:22,400 --> 01:46:26,077
Well, I met with the Blitzy team
yesterday and I think one theory here is

1911
01:46:26,160 --> 01:46:28,957
um you know, a lot of the uh the

1912
01:46:29,040 --> 01:46:33,117
enterprise um you know, actually
Salesforce.com would be a great example.

1913
01:46:33,200 --> 01:46:35,437
Maybe Richard has an opinion on this,
but a lot of people are getting into

1914
01:46:35,520 --> 01:46:39,277
orchestration. And their whole sales
pitch in orchestration is look, we're

1915
01:46:39,360 --> 01:46:42,157
going to use Opus 5.5 as an orchestrator,
but then we're going to

1916
01:46:42,240 --> 01:46:45,677
use Kimmy K3 or Quen as a submodel at,

1917
01:46:45,760 --> 01:46:49,677
you know, half or a third the or fifth
the price, and we'll farm out the tasks

1918
01:46:49,760 --> 01:46:52,237
and the contexts perfectly to get you a

1919
01:46:52,320 --> 01:46:56,957
much lower cost per code, per outcome,
per experiment, whatever your output is.

1920
01:46:57,040 --> 01:47:00,077
We can cut it in half with our
orchestration intelligence, but it

1921
01:47:00,160 --> 01:47:03,437
relies on anthropic up here and cheaper

1922
01:47:03,520 --> 01:47:07,757
models down here. And I think by cutting
the cost in half, they might be trying

1923
01:47:07,840 --> 01:47:11,917
to fill that gap and say, "No, no, no,
go with anthropic top to bottom. Then

1924
01:47:12,000 --> 01:47:14,557
you don't have to worry about Chinese
code injection. You don't have to worry

1925
01:47:14,640 --> 01:47:18,717
about whatever." So my my theory would
be that they're trying to compete with,

1926
01:47:18,800 --> 01:47:22,717
you know, China, which is about three
months behind and and fill that gap

1927
01:47:22,800 --> 01:47:26,557
before a lot of enterprises
go to to open source models.

1928
01:47:26,640 --> 01:47:30,477
>> But, you know, Alex Karp is pushing
really really hard on this agenda. Like

1929
01:47:30,560 --> 01:47:34,637
if you want to control your own destiny,
you can't trust Anthropic. You can't get

1930
01:47:34,720 --> 01:47:36,717
addicted to them as your vendor. You

1931
01:47:36,800 --> 01:47:40,397
must go with models you can control.

1932
01:47:40,480 --> 01:47:43,277
Well, the the you know the the challenge

1933
01:47:43,360 --> 01:47:47,517
right now is the these models have
an an increasingly shorter and shorter

1934
01:47:47,600 --> 01:47:51,757
halflife, right?
>> So the competitive advantage comes not

1935
01:47:51,840 --> 01:47:54,077
from having access to the latest model.

1936
01:47:54,160 --> 01:47:55,917
Everybody has access. It's how do you

1937
01:47:56,000 --> 01:47:58,637
metabolize that into some decent

1938
01:47:58,720 --> 01:48:01,117
capability? That's the real challenge.

1939
01:48:01,200 --> 01:48:03,197
>> Richard, you said
compute is the biggest

1940
01:48:03,280 --> 01:48:06,477
constraint. You know, those are
your words. So if intelligent is getting

1941
01:48:06,560 --> 01:48:10,077
cheaper and we've just repriced compute
you know like two or three times this

1942
01:48:10,160 --> 01:48:12,717
week uh you know down by almost a factor

1943
01:48:12,800 --> 01:48:16,877
of three um you know if intelligence is

1944
01:48:16,960 --> 01:48:19,037
getting cheaper per task then why is

1945
01:48:19,120 --> 01:48:22,237
compute still a thing that limits us

1946
01:48:22,320 --> 01:48:25,117
>> just the physics I guess of it and like

1947
01:48:25,200 --> 01:48:26,717
you'd be surprised if you try to buy

1948
01:48:26,800 --> 01:48:32,397
like a thousand uh GB200s uh and so
on like the price has actually gone up in

1949
01:48:32,480 --> 01:48:33,597
several cases uh you

1950
01:48:33,680 --> 01:48:36,877
>> oh my god Richard, I I had a B300, an

1951
01:48:36,960 --> 01:48:39,917
ADX B300 on order for 3 million bucks

1952
01:48:40,000 --> 01:48:44,877
due in December. Somebody scooped it
for 5 million. They they just called. I We

1953
01:48:44,960 --> 01:48:46,477
had it and somebody called and said,

1954
01:48:46,560 --> 01:48:48,557
"No, no, we sold it to somebody else for

1955
01:48:48,640 --> 01:48:51,917
$2 million more." I couldn't believe it.

1956
01:48:52,000 --> 01:48:54,957
>> It It was actually
an NVL72, not a GB300.

1957
01:48:55,040 --> 01:48:59,837
>> So So yeah, like the price of H100s has
uh gone up uh in in a crazy way. There's

1958
01:48:59,920 --> 01:49:03,437
like these are seveny old GPUs. people
like when you do financial modeling you

1959
01:49:03,520 --> 01:49:07,517
assume they're worth zero after 5 years
there after seven years there's they

1960
01:49:07,600 --> 01:49:11,757
they went up again um uh like over the
last few months. So there's currently a

1961
01:49:11,840 --> 01:49:15,197
bit of a compute crunch. This is
something that again capitalism will

1962
01:49:15,280 --> 01:49:19,837
solve like there's so much demand for
compute right now and for tokens that a

1963
01:49:19,920 --> 01:49:23,597
lot of people are building land power
shell data centers and so on. My hunch

1964
01:49:23,680 --> 01:49:26,877
is in like maybe two years there will be
more in the market and it's going to be

1965
01:49:26,960 --> 01:49:31,757
a little bit like electricity uh prices
might fluctuate. Obviously, there's uh

1966
01:49:31,840 --> 01:49:35,917
near sheer infinite demand for more
intelligence on the planet. Um so I I

1967
01:49:36,000 --> 01:49:39,437
don't think there will be like a crash,
but like the prices of compute fluctuate

1968
01:49:39,520 --> 01:49:43,357
and unfortunately they're not just going
down. Right now, if you want like the

1969
01:49:43,440 --> 01:49:47,997
the beefiest and and largest GPU clusters,
um prices like are people are

1970
01:49:48,080 --> 01:49:51,597
trying to lock them in now because they
expect them to keep going up for the

1971
01:49:51,680 --> 01:49:54,717
next few months. >> Richard,
you just spent $450 million,

1972
01:49:54,800 --> 01:49:58,397
didn't you? On Wait, did you take
my NVL72? And that that may have been one

1973
01:49:58,480 --> 01:50:00,397
of the smallest comput deals we've done.

1974
01:50:00,480 --> 01:50:03,517
Um, yeah. >> Really?
Wow. Are are you actually

1975
01:50:03,600 --> 01:50:08,077
leasing or buying or building
or what are you doing?

1976
01:50:08,160 --> 01:50:09,277
>> Oh, sorry. I didn't know that.

1977
01:50:09,360 --> 01:50:11,837
>> Welcome to the health section
of Moonshots brought to you by Fountain

1978
01:50:11,920 --> 01:50:15,757
Life. You know, my mission is to help
you use the latest technologies,

1979
01:50:15,840 --> 01:50:18,157
including AI, to not just do your work

1980
01:50:18,240 --> 01:50:23,597
at home, teach your kids, but to help
you live a long and healthy life. I'm

1981
01:50:23,680 --> 01:50:27,277
here today with an extraordinary
physician, the chief medical officer of

1982
01:50:27,360 --> 01:50:30,157
Fountain Life, Dr. Don Mucalem. Don,

1983
01:50:30,240 --> 01:50:32,077
let's talk about cancer. Uh, you know, I

1984
01:50:32,160 --> 01:50:35,517
know from the member database that we

1985
01:50:35,600 --> 01:50:39,917
have at Fountain are members who come
in who think they're healthy. It turns out

1986
01:50:40,000 --> 01:50:43,917
3.3% of them have a cancer
in their body they don't know about.

1987
01:50:44,000 --> 01:50:47,757
>> That's right. You know, the majority of
cancers that we screen for, those aren't

1988
01:50:47,840 --> 01:50:51,917
the ones that are necessarily taking
the lives when found at a late stage. We

1989
01:50:52,000 --> 01:50:56,157
know that when cancer is found early,
the chances for cure are much higher. We

1990
01:50:56,240 --> 01:51:00,717
know it's much easier to treat a cancer
when found early versus when found late.

1991
01:51:00,800 --> 01:51:05,037
What we're finding in our members
is over 3.3% were found to have these

1992
01:51:05,120 --> 01:51:08,157
cancers that were otherwise
wouldn't have been found or detected.

1993
01:51:08,240 --> 01:51:09,517
>> Yeah. You know, it's interesting.

1994
01:51:09,600 --> 01:51:13,837
People, you don't feel the cancer until
stage three or stage four. And and if

1995
01:51:13,920 --> 01:51:16,477
you don't know what's going on inside
your body, it's like driving your car

1996
01:51:16,560 --> 01:51:19,197
with your eyes closed. And you can know.

1997
01:51:19,280 --> 01:51:22,957
And so when members come through
found how do they detect cancers?

1998
01:51:23,040 --> 01:51:27,277
>> So we're doing full body MRI and we
also do early cancer detection screening.

1999
01:51:27,360 --> 01:51:30,797
This is very very important and these
are not typical tools used in the

2000
01:51:30,880 --> 01:51:35,117
conventional care setting when it comes
to prevention. This is a hard thing

2001
01:51:35,200 --> 01:51:38,477
because currently these are not studies
that insurance would yet be covering.

2002
01:51:38,560 --> 01:51:43,197
But the goal is to collect these numbers,
do the research and work hard

2003
01:51:43,280 --> 01:51:46,637
to democratize wellness. >> Yeah.
So at the end of the day you can

2004
01:51:46,720 --> 01:51:47,917
know what's going on inside your body.

2005
01:51:48,000 --> 01:51:51,677
It's your obligation to know.
So check out Fountain Life. You can go to

2006
01:51:51,760 --> 01:51:55,997
fountainlife.com/pater to get
access to the latest technology

2007
01:51:56,080 --> 01:52:00,317
to help you detect cancer at the very
beginning at stage one when it is

2008
01:52:00,400 --> 01:52:04,637
curable before it gets to stage three
or stage four in your world of hurt. I'm

2009
01:52:04,720 --> 01:52:08,397
going to move us to our next story which
is about decision models. So here's the

2010
01:52:08,480 --> 01:52:11,437
idea. When software needs to make a

2011
01:52:11,520 --> 01:52:16,237
quick decision on a bounded decision
like is this you know transaction fraud?

2012
01:52:16,320 --> 01:52:20,637
Yes or no? you know, which of these five
categories does a ticket belong in? It

2013
01:52:20,720 --> 01:52:24,717
doesn't need a model that thinks for 10
seconds and writes a paragraph. It needs

2014
01:52:24,800 --> 01:52:27,277
an answer in milliseconds. That's called

2015
01:52:27,360 --> 01:52:29,277
a decision model. Instead of generating

2016
01:52:29,360 --> 01:52:31,837
uh text one token at a time, it answers

2017
01:52:31,920 --> 01:52:34,797
a typed question or a choice uh a score

2018
01:52:34,880 --> 01:52:37,597
with a yes or no quickly. So on

2019
01:52:37,680 --> 01:52:40,077
September 15th, a startup called

2020
01:52:40,160 --> 01:52:43,757
Typesafe AI came out after two years of

2021
01:52:43,840 --> 01:52:46,077
stealth with something called Jev. uh

2022
01:52:46,160 --> 01:52:48,317
it's a closed decision model they call a

2023
01:52:48,400 --> 01:52:51,277
system one model uh it's fast intuitive

2024
01:52:51,360 --> 01:52:53,357
kind of thinking as opposed to slow

2025
01:52:53,440 --> 01:52:56,317
reasoning so uh Alex I'm going to go to

2026
01:52:56,400 --> 01:53:01,437
you you know and sele you guys were
excited about Jev uh let's parse it

2027
01:53:01,520 --> 01:53:05,917
understand what this is why it's important
>> yeah so a bit of maybe prehistory first

2028
01:53:06,000 --> 01:53:09,757
in the beginning there was the transformer
and the transformer was good

2029
01:53:09,840 --> 01:53:12,797
and it was based on an encoder and a

2030
01:53:12,880 --> 01:53:17,597
decoder so it it took a sequence and
converted the sequence to an embedding.

2031
01:53:17,680 --> 01:53:22,157
That was the encoder part. And then it
took the embedding and it decoded that

2032
01:53:22,240 --> 01:53:24,397
to another sequence or another part of a

2033
01:53:24,480 --> 01:53:26,637
sequence. That's the decoder part. And

2034
01:53:26,720 --> 01:53:30,477
the encoder part evolved into a whole

2035
01:53:30,560 --> 01:53:34,317
ecosystem of models popularly maybe BERT

2036
01:53:34,400 --> 01:53:38,157
style models. And the decoder

2037
01:53:38,240 --> 01:53:42,237
style or the decoder half of the
transformer evolved into a much larger

2038
01:53:42,320 --> 01:53:44,957
ecosystem of large language models. So

2039
01:53:45,040 --> 01:53:47,997
most of the the models that are

2040
01:53:48,080 --> 01:53:49,357
consuming all of the compute of

2041
01:53:49,440 --> 01:53:52,557
civilization today are decoder style

2042
01:53:52,640 --> 01:53:56,877
transformers and the encoder has
basically gone missing in action. There

2043
01:53:56,960 --> 01:53:58,797
are some people who still think maybe

2044
01:53:58,880 --> 01:54:01,277
for purposes of

2045
01:54:01,360 --> 01:54:04,477
retrieval uh it it's still popular to to

2046
01:54:04,560 --> 01:54:07,517
to do embedding based retrieval but by

2047
01:54:07,600 --> 01:54:10,957
and large decoders have stolen the show

2048
01:54:11,040 --> 01:54:14,797
and I fast forwarding maybe year two

2049
01:54:14,880 --> 01:54:17,757
years ago at this point open AI and

2050
01:54:17,840 --> 01:54:20,157
others recognizing that there was a

2051
01:54:20,240 --> 01:54:25,037
desire to be able to take decoderonly
large language models and to add

2052
01:54:25,120 --> 01:54:29,597
structure to them Because without
structure, you ask for next token you

2053
01:54:29,680 --> 01:54:33,277
can get anything. People wanted a little
bit more structure like they were

2054
01:54:33,360 --> 01:54:37,597
getting in in the days of encoder only
models where you get an embedding out

2055
01:54:37,680 --> 01:54:40,717
and then you can train a classifier
based on the embedding and the

2056
01:54:40,800 --> 01:54:42,717
classifier maybe has a finite set of

2057
01:54:42,800 --> 01:54:45,677
categories or numerical outputs. OpenAI

2058
01:54:45,760 --> 01:54:48,317
added structured output support for

2059
01:54:48,400 --> 01:54:50,237
their GPT series and everyone copied

2060
01:54:50,320 --> 01:54:54,077
that. But no one as a my perception is

2061
01:54:54,160 --> 01:54:58,237
as a function of the total user base
for all models structured output never

2062
01:54:58,320 --> 01:55:00,317
really got that much love somewhat

2063
01:55:00,400 --> 01:55:04,157
analogously to how OpenAI launched um

2064
01:55:04,240 --> 01:55:08,157
RFT reinforcement fine-tuning and no
one used that and they had to shut that

2065
01:55:08,240 --> 01:55:12,317
down. Similarly, structured output
was available but I think underloved and

2066
01:55:12,400 --> 01:55:16,237
underused by the community. So fast
forwarding then all the way to a few

2067
01:55:16,320 --> 01:55:19,117
weeks ago, startup Typesafe announces a

2068
01:55:19,200 --> 01:55:22,317
model called Jev. And Jev they mark it

2069
01:55:22,400 --> 01:55:25,597
as so-called system one intelligence.

2070
01:55:25,680 --> 01:55:28,797
This is Conaman style reference to

2071
01:55:28,880 --> 01:55:31,117
system one, system two type thinking.

2072
01:55:31,200 --> 01:55:33,997
System one purportedly being intuitive,

2073
01:55:34,080 --> 01:55:36,317
fast reaction, system two being

2074
01:55:36,400 --> 01:55:39,357
reasoning, meditative, long-term type

2075
01:55:39,440 --> 01:55:42,237
thinking models. And one could squint at

2076
01:55:42,320 --> 01:55:44,317
Typesafe's announcement of Jev and say,

2077
01:55:44,400 --> 01:55:49,277
"Okay, this is in some sense this is
a reaction to an overindexing or an

2078
01:55:49,360 --> 01:55:50,957
overreaction

2079
01:55:51,040 --> 01:55:55,037
by the the user base of of AI at this

2080
01:55:55,120 --> 01:55:57,357
point to reasoning models in some sense.

2081
01:55:57,440 --> 01:55:59,277
Maybe arguing that we're using reasoning

2082
01:55:59,360 --> 01:56:02,317
models too much and we've abandoned the

2083
01:56:02,400 --> 01:56:07,117
the the important use case of really
fast intuitive snap decision models."

2084
01:56:07,200 --> 01:56:11,757
And this is purportedly then where Jev
comes in. Jev purportedly again a whole

2085
01:56:11,840 --> 01:56:13,917
new architecture. The the details not

2086
01:56:14,000 --> 01:56:17,117
quite clear that it can it it won't

2087
01:56:17,200 --> 01:56:20,157
output uh unless you you torture it into

2088
01:56:20,240 --> 01:56:23,197
it. Uh it won't output general purpose

2089
01:56:23,280 --> 01:56:25,837
sequences like say GPT or claude will

2090
01:56:25,920 --> 01:56:30,877
but you can ask it sort of like a magic
eightball. You you can ask it to is that

2091
01:56:30,960 --> 01:56:33,197
too dated to reference maybe. >> Nope.

2092
01:56:33,280 --> 01:56:35,517
>> Nope. Uh you you
you can you can feed it

2093
01:56:35,600 --> 01:56:40,397
a sequence or an image or a multimodal
input. uh and you can ask it to to

2094
01:56:40,480 --> 01:56:44,397
answer in the form of a categorical.
So it's one of n possibilities or a

2095
01:56:44,480 --> 01:56:46,317
numerical like give me a number between

2096
01:56:46,400 --> 01:56:49,437
zero and one on a continuum or a binary

2097
01:56:49,520 --> 01:56:51,757
like true or false. You you can ask for

2098
01:56:51,840 --> 01:56:55,437
a much simpler output. And turns out

2099
01:56:55,520 --> 01:56:59,997
that's is a great idea. Like you can
use it for for use cases where where you

2100
01:57:00,080 --> 01:57:03,517
need really low latency snap decisions
like computer use assistance like

2101
01:57:03,600 --> 01:57:08,157
pressing buttons on a computer screen
or for ultra lowcase classification

2102
01:57:08,240 --> 01:57:12,637
problems. If you want to classify every
row in a database decision style models

2103
01:57:12,720 --> 01:57:14,397
which are in some sense like a reinvent

2104
01:57:14,480 --> 01:57:17,117
reinvention of the encoder transformer

2105
01:57:17,200 --> 01:57:22,157
type wheel. Amazing story. But then
of course this is such a simple idea. You

2106
01:57:22,240 --> 01:57:25,117
could ask the question why doesn't
everyone else do this? And the answer is

2107
01:57:25,200 --> 01:57:27,197
everyone else is now doing it. Open AAI

2108
01:57:27,280 --> 01:57:29,677
released as part of their dev day that

2109
01:57:29,760 --> 01:57:32,797
we covered last time a decisions API. So

2110
01:57:32,880 --> 01:57:37,197
open AAI immediately co-opted this
renewed interest in so-called type one

2111
01:57:37,280 --> 01:57:40,637
or decision models. This is already
available via open API. They're open

2112
01:57:40,720 --> 01:57:41,837
source projects that have all cloned.

2113
01:57:41,920 --> 01:57:46,077
This turns out this is like an absurdly
absurdly simple concept for everyone to

2114
01:57:46,160 --> 01:57:48,637
just go and implement trivially. It it

2115
01:57:48,720 --> 01:57:50,877
could in its simplest case just be you

2116
01:57:50,960 --> 01:57:53,437
take an a Chinese openweight LLM off the

2117
01:57:53,520 --> 01:57:56,077
shelf and you fine-tune it uh and maybe

2118
01:57:56,160 --> 01:58:01,197
labbotomize it by a few levels just
to produce categorical outputs on demand.

2119
01:58:01,280 --> 01:58:05,517
That's the story. >> So I I
think this I think all of that is

2120
01:58:05,600 --> 01:58:09,037
absolutely correct. that there's
a real really important part of this for

2121
01:58:09,120 --> 01:58:11,357
organizations because when you're making

2122
01:58:11,440 --> 01:58:14,397
decisions inside an organization a large

2123
01:58:14,480 --> 01:58:18,637
vast number of those are system one type

2124
01:58:18,720 --> 01:58:20,237
uh thing and I made a list of these just

2125
01:58:20,320 --> 01:58:22,717
to make it clear for people. So uh route

2126
01:58:22,800 --> 01:58:25,917
this ticket um approve this exception uh

2127
01:58:26,000 --> 01:58:27,757
choose this supplier escalate this

2128
01:58:27,840 --> 01:58:30,637
transaction um move this thing left or

2129
01:58:30,720 --> 01:58:35,117
right send this message now or later.
So when you're when you have all these

2130
01:58:35,200 --> 01:58:39,197
decisions to make which are thousands
of them, you know, we've been working with

2131
01:58:39,280 --> 01:58:41,357
these companies to do detailed task

2132
01:58:41,440 --> 01:58:43,517
breakdowns, a vast majority of them are

2133
01:58:43,600 --> 01:58:48,717
these little micro decisions and trying
to use an LLM for this is like trying to

2134
01:58:48,800 --> 01:58:52,717
bring the Supreme Court together
to decide what like which check out which

2135
01:58:52,800 --> 01:58:56,477
checkout line of the supermarket you
should go to. And therefore you've got

2136
01:58:56,560 --> 01:58:58,397
this wonderful architecture now where

2137
01:58:58,480 --> 01:59:00,477
you can route system one things to this

2138
01:59:00,560 --> 01:59:04,157
very lowcost nearly free thing and then

2139
01:59:04,240 --> 01:59:07,357
for pondering deep important questions

2140
01:59:07,440 --> 01:59:09,917
strategy questions where a human huge

2141
01:59:10,000 --> 01:59:12,717
amount of human judgment type models are

2142
01:59:12,800 --> 01:59:15,117
required you can route it to those. So

2143
01:59:15,200 --> 01:59:19,997
it's it's a big deal for the
organizational singularity world. uh

2144
01:59:20,080 --> 01:59:23,437
we've been waiting for something like
this and Alex is correct. This has been

2145
01:59:23,520 --> 01:59:27,277
possible for a long time but now it's
been made built into the systems. It's

2146
01:59:27,360 --> 01:59:31,997
it's made a very uh callable layer
and this is huge because the cost of

2147
01:59:32,080 --> 01:59:34,717
microcoordination collapses
and that's really big.

2148
01:59:34,800 --> 01:59:37,437
>> Richard, your thoughts on this?

2149
01:59:37,520 --> 01:59:40,157
>> Classifiers are back. Um makes a lot of

2150
01:59:40,240 --> 01:59:43,037
sense. Um you know, not every uh thing

2151
01:59:43,120 --> 01:59:45,757
in in software needs a very complex

2152
01:59:45,840 --> 01:59:48,637
decoder. Um, like Alex correctly said,

2153
01:59:48,720 --> 01:59:51,517
um, it's a really clever new way of

2154
01:59:51,600 --> 01:59:56,717
making the old, which is classifiers,
new by allowing a more general encoder

2155
01:59:56,800 --> 02:00:01,277
and then quickly giving you
classification results. I think it's

2156
02:00:01,360 --> 02:00:04,477
it's one of those ideas that is so
beautiful. A lot of people thought, ah,

2157
02:00:04,560 --> 02:00:08,237
why didn't we do that? I didn't realize
that could be so exciting for so many

2158
02:00:08,320 --> 02:00:12,077
people. So now there are already various
like Chinese open source versions of

2159
02:00:12,160 --> 02:00:14,317
this that along various benchmarks are

2160
02:00:14,400 --> 02:00:19,117
are doing better. Um and and we'll
likely see this come and other large

2161
02:00:19,200 --> 02:00:22,637
labs will likely follow suit.
>> And and just to reiterate, we mentioned

2162
02:00:22,720 --> 02:00:24,397
this already on the pod, but Jev stands

2163
02:00:24,480 --> 02:00:26,397
for Jevans paradox because the thesis

2164
02:00:26,480 --> 02:00:29,677
here is we'll do now a massive amount

2165
02:00:29,760 --> 02:00:33,437
more micro decision making
than we did before as a result.

2166
02:00:33,520 --> 02:00:36,557
>> Nice. Dave, you want to close us out
here? Yeah, it's a great case study I

2167
02:00:36,640 --> 02:00:39,357
think in the tension between uh one mega

2168
02:00:39,440 --> 02:00:41,677
model from anthropic or open AI serving

2169
02:00:41,760 --> 02:00:44,557
all humanity or open source and

2170
02:00:44,640 --> 02:00:48,957
creativity and entrepreneurship building
things you never would have thought of.

2171
02:00:49,040 --> 02:00:52,957
But then you get the risk of cyber
terrorism, you know. So that's the

2172
02:00:53,040 --> 02:00:56,557
tension that we're with. But I I've
really wanted to build a box that you

2173
02:00:56,640 --> 02:00:59,997
put at the side of the basketball court
when you go to the Y and it's got a

2174
02:01:00,080 --> 02:01:03,277
little camera on it. cost next
to nothing and it's doing like all of the

2175
02:01:03,360 --> 02:01:06,237
announcing that a professional announcer
would do, it's doing while you're

2176
02:01:06,320 --> 02:01:08,317
playing pickup at the YMCA. You could

2177
02:01:08,400 --> 02:01:11,117
crank that out like in 2 seconds using a

2178
02:01:11,200 --> 02:01:13,357
classifier that's really fast and snappy

2179
02:01:13,440 --> 02:01:16,477
and funny. Uh and but you need open

2180
02:01:16,560 --> 02:01:19,917
source to to build things like that. So

2181
02:01:20,000 --> 02:01:23,677
super excited about the fact that we
have open source still. Not sure

2182
02:01:23,760 --> 02:01:26,957
>> actually. Dave, you're making me
think the magic eightball really should

2183
02:01:27,040 --> 02:01:31,117
whoever owns it, Mattel or whoever
you just like use a decision model to

2184
02:01:31,200 --> 02:01:35,197
implement a modern magic eightball
that actually understands the question and

2185
02:01:35,280 --> 02:01:39,117
answers it with a categorical >>
totally totally would sell. It makes a

2186
02:01:39,200 --> 02:01:44,317
ton of sense like 20 bucks >>
or free on free as an app. All right,

2187
02:01:44,400 --> 02:01:47,597
I'm going to close this out with a story
from two days ago. So on Wednesday, the

2188
02:01:47,680 --> 02:01:50,157
defense secretary Pete Segwit uh

2189
02:01:50,240 --> 02:01:52,077
speaking at the Marine Corps base in

2190
02:01:52,160 --> 02:01:55,037
Quantico announced Project Meridian.

2191
02:01:55,120 --> 02:01:57,277
It's a new Pentagon effort on the future

2192
02:01:57,360 --> 02:02:00,717
of warfare. It's co-led by Elon Musk and

2193
02:02:00,800 --> 02:02:02,957
Palmer Lucky. Pretty extraordinary along

2194
02:02:03,040 --> 02:02:05,277
with N Gingrich and overseen by the

2195
02:02:05,360 --> 02:02:08,717
Pentagon's uh CTO Emil Michael. Uh let

2196
02:02:08,800 --> 02:02:13,277
me read from SegWit what Segwit said.
He said, quote, "Project Meridian, the

2197
02:02:13,360 --> 02:02:16,477
future of warfare, is not about
developing new strategies or new

2198
02:02:16,560 --> 02:02:21,037
policies. It's about discovering,
developing, and fielding the weapons and

2199
02:02:21,120 --> 02:02:23,357
systems future troops will need on the

2200
02:02:23,440 --> 02:02:25,997
battlefield from the Earth to beyond the

2201
02:02:26,080 --> 02:02:29,437
moon." Uh, findings are due in 120 days.

2202
02:02:29,520 --> 02:02:34,077
I I like these kinds of commissions that
are timelmited and don't have Elon off

2203
02:02:34,160 --> 02:02:37,997
on the side for, you know, a year at a
time. It's worth noting in the context

2204
02:02:38,080 --> 02:02:41,757
that, you know, both SpaceX and Ander
hold multi-billion dollar defense

2205
02:02:41,840 --> 02:02:44,237
contracts and Anderoll is building

2206
02:02:44,320 --> 02:02:46,717
autonomous weapons. So, Richard, uh,

2207
02:02:46,800 --> 02:02:49,757
your essay names autonomous weapons as

2208
02:02:49,840 --> 02:02:54,077
one of the four genuine concerns
and you've said AI should never control

2209
02:02:54,160 --> 02:02:55,917
lethal decisions without human

2210
02:02:56,000 --> 02:02:59,277
oversight. Your thoughts on this?

2211
02:02:59,360 --> 02:03:02,317
>> Yeah. Um, I I stand by those. um it's

2212
02:03:02,400 --> 02:03:04,477
not a a particular area that I am

2213
02:03:04,560 --> 02:03:07,197
excited about applying AI to um

2214
02:03:07,280 --> 02:03:10,877
obviously like people will um but I

2215
02:03:10,960 --> 02:03:15,597
really hope I mean you know again I'm
not a doomer at all but like a really

2216
02:03:15,680 --> 02:03:19,357
poor decision would be to give AI access
to all the nuclear codes and all the

2217
02:03:19,440 --> 02:03:23,037
nuclear weapons and connected that's
literally how Skynet and Terminator 3

2218
02:03:23,120 --> 02:03:27,037
get started so uh I think there are
places for super intelligence and

2219
02:03:27,120 --> 02:03:28,637
scientific discovery that I'm very

2220
02:03:28,720 --> 02:03:30,957
excited about expanding human knowledge

2221
02:03:31,040 --> 02:03:34,797
I I think uh the more uh we we get to

2222
02:03:34,880 --> 02:03:39,117
deciding not just to impact human lives
but to end human lives, the more we we

2223
02:03:39,200 --> 02:03:40,397
should have human oversight.

2224
02:03:40,480 --> 02:03:44,157
>> Mhm. Yeah. Uh Salem,

2225
02:03:44,240 --> 02:03:47,677
>> um you know, I think what's important
about this whole thing is they're trying

2226
02:03:47,760 --> 02:03:50,637
to revamp and rethink how you this kind

2227
02:03:50,720 --> 02:03:55,117
of 100-year-old organization.
And the big question is going to be can a

2228
02:03:55,200 --> 02:03:57,517
procurement organization built for 20 or

2229
02:03:57,600 --> 02:04:00,317
weapons programs operate on like a

2230
02:04:00,400 --> 02:04:02,797
90-day technology cycle and this is

2231
02:04:02,880 --> 02:04:05,357
going to be the big challenge. Um it's

2232
02:04:05,440 --> 02:04:09,997
not that the risk it's not that they
failed to identify these future

2233
02:04:10,080 --> 02:04:14,957
technologies that we can all see those
is how quickly can they absorb them and

2234
02:04:15,040 --> 02:04:18,317
at the speed they're developing
and bring them to the front.

2235
02:04:18,400 --> 02:04:20,477
>> Yeah. like you can't find exponential

2236
02:04:20,560 --> 02:04:23,357
technology with linear procurement,

2237
02:04:23,440 --> 02:04:26,877
>> right? Dave, your thoughts, please.

2238
02:04:26,960 --> 02:04:30,237
>> Yeah, I I don't know. I I I had a great

2239
02:04:30,320 --> 02:04:34,077
time with Palmer Lucky in LA and I I
really really love uh love him and I

2240
02:04:34,160 --> 02:04:38,877
think Elon too is just an awesome
good-natured person. I'm just overjoyed

2241
02:04:38,960 --> 02:04:43,117
that there are people in Washington that I
can sit down to with and relate to. My

2242
02:04:43,200 --> 02:04:46,237
entire career going to Washington
has been a dread for me

2243
02:04:46,320 --> 02:04:49,037
>> because it's just lawyers
and politicians and occasionally an

2244
02:04:49,120 --> 02:04:53,437
accountant and there's just no
productive meeting. And for some reason,

2245
02:04:53,520 --> 02:04:55,437
just in the last year, we're starting to

2246
02:04:55,520 --> 02:04:59,037
see very very smart, very capable people

2247
02:04:59,120 --> 02:05:00,957
willing to go and and get a mosquito

2248
02:05:01,040 --> 02:05:05,597
bite, I guess, in Washington. Um, and I

2249
02:05:05,680 --> 02:05:09,437
just makes me really optimistic that
they'll figure some things out. Um, I'm

2250
02:05:09,520 --> 02:05:13,117
also not a fan of autonomous weapons
that make decisions in the field, but

2251
02:05:13,200 --> 02:05:14,717
Palmer Lucky made a very good case for

2252
02:05:14,800 --> 02:05:17,197
it. You can see it in our podcast from

2253
02:05:17,280 --> 02:05:20,877
uh from LA. Uh, I I don't agree that

2254
02:05:20,960 --> 02:05:24,237
it's a good choice, but he he actually
has some very rational arguments for why

2255
02:05:24,320 --> 02:05:26,077
it's going to be that way. >> Yeah.

2256
02:05:26,160 --> 02:05:28,557
>> Can can I just mention
one more statistic here,

2257
02:05:28,640 --> 02:05:31,037
>> please? If you went back to the two

2258
02:05:31,120 --> 02:05:33,917
years into the Ukraine Russia conflict,

2259
02:05:34,000 --> 02:05:36,077
they were using about a half a million

2260
02:05:36,160 --> 02:05:39,037
drones to fight and prosecute the war.

2261
02:05:39,120 --> 02:05:42,077
Um, this year Russia will make 10

2262
02:05:42,160 --> 02:05:43,917
million drones and the Ukraine will make

2263
02:05:44,000 --> 02:05:46,397
10 million drones. So, talk about

2264
02:05:46,480 --> 02:05:50,557
exponential. Uh, that is an unbelievable

2265
02:05:50,640 --> 02:05:55,277
escalation, but without humans in the
loop. So, that's the good news around

2266
02:05:55,360 --> 02:05:59,517
it. Uh uh, of course those drones are
doing a ton more damage, but they're

2267
02:05:59,600 --> 02:06:01,677
fighting each other with drones at a

2268
02:06:01,760 --> 02:06:04,237
scalable level. The other statistic I

2269
02:06:04,320 --> 02:06:07,117
remember there's about 10,000 drones a

2270
02:06:07,200 --> 02:06:11,677
month crossing the Mexico US border. Uh,

2271
02:06:11,760 --> 02:06:16,637
and so the problem there is wall
technology is not as good as drone

2272
02:06:16,720 --> 02:06:19,917
technology. And so trying to build
a wall along there is not the greatest

2273
02:06:20,000 --> 02:06:20,877
idea right now.

2274
02:06:20,960 --> 02:06:23,357
>> Nice. Alex, close us out on this one.

2275
02:06:23,440 --> 02:06:26,957
>> Yeah, a couple points. So the secretary

2276
02:06:27,040 --> 02:06:31,437
of war announced this alongside
several other initiatives maybe most

2277
02:06:31,520 --> 02:06:33,037
conspicuously

2278
02:06:33,120 --> 02:06:35,677
a project code came cenamed project

2279
02:06:35,760 --> 02:06:38,797
azinort which is the standup of the

2280
02:06:38,880 --> 02:06:41,357
department of war's first autonomous

2281
02:06:41,440 --> 02:06:44,717
warfare command or auto warcom. This I

2282
02:06:44,800 --> 02:06:47,437
think is a transformative moment for the

2283
02:06:47,520 --> 02:06:49,357
department of war. We finally will have

2284
02:06:49,440 --> 02:06:51,997
a dedicated joint force devoted to

2285
02:06:52,080 --> 02:06:55,117
autonomous weapon systems. I think

2286
02:06:55,200 --> 02:06:58,157
including drones but not exclusively

2287
02:06:58,240 --> 02:07:00,957
drones. This is a major major step

2288
02:07:01,040 --> 02:07:04,077
forward for US capabilities to finally

2289
02:07:04,160 --> 02:07:08,637
have a single joint force dedicated
to this with a four-star functional

2290
02:07:08,720 --> 02:07:12,797
combatant command that we've been missing.
We're arguably missing. You

2291
02:07:12,880 --> 02:07:16,557
know, I would go on a soap box and say
there if if I were Secretary of War for

2292
02:07:16,640 --> 02:07:19,517
a day, there are probably several other
functional combatant commands that I'd

2293
02:07:19,600 --> 02:07:24,077
spin up if I had the opportunity, but
this would have been one of my top five

2294
02:07:24,160 --> 02:07:28,157
list. The the other point that I'll make
just for project meridian which you were

2295
02:07:28,240 --> 02:07:30,317
asking about specifically is in the

2296
02:07:30,400 --> 02:07:33,597
announcement when sewar heath announced

2297
02:07:33,680 --> 02:07:35,997
it there was very particular language

2298
02:07:36,080 --> 02:07:39,197
around examining I'll quote the full

2299
02:07:39,280 --> 02:07:41,517
spectrum of future warf fighting domains

2300
02:07:41,600 --> 02:07:44,397
from subterranean depths to the cy lunar

2301
02:07:44,480 --> 02:07:46,477
frontier. So I think those two

2302
02:07:46,560 --> 02:07:49,997
particular domains, the ocean bottom and

2303
02:07:50,080 --> 02:07:53,357
cis lunar and lunar space in general are

2304
02:07:53,440 --> 02:07:56,237
two wildly underserved domains not just

2305
02:07:56,320 --> 02:07:58,877
for warfare but for peaceime activities

2306
02:07:58,960 --> 02:08:03,277
as well. We know embarrassingly
little about our ocean bottoms and

2307
02:08:03,360 --> 02:08:07,357
optimistically one can
imagine that if if the DO

2308
02:08:07,440 --> 02:08:10,157
decides that is now very interested in

2309
02:08:10,240 --> 02:08:12,957
investing in American assets and

2310
02:08:13,040 --> 02:08:15,037
American exploration and American

2311
02:08:15,120 --> 02:08:18,557
dominance on the floor of the oceans and

2312
02:08:18,640 --> 02:08:22,637
the lunar surface and the cis lunar
region in general. I think that is going

2313
02:08:22,720 --> 02:08:27,197
to have dividends and pay dividends
for the entire economy and for humanity's

2314
02:08:27,280 --> 02:08:29,597
broader technological advances in a way

2315
02:08:29,680 --> 02:08:35,117
that might naively have nothing
to do with warfare or war fighting.

2316
02:08:35,200 --> 02:08:38,157
>> One of the points that Palmer made
actually in LA is that under the ocean

2317
02:08:38,240 --> 02:08:40,877
it's not practical to communicate with a

2318
02:08:40,960 --> 02:08:45,757
central server or central command.
So that's already automated and I don't

2319
02:08:45,840 --> 02:08:49,837
know what triggers it, but once it goes
into hunt mode, it just hunts and it's

2320
02:08:49,920 --> 02:08:53,357
not communicating back.
>> Twothirds of the Earth's surface and we

2321
02:08:53,440 --> 02:08:56,157
know embarrassingly little about it.

2322
02:08:56,240 --> 02:08:57,517
>> We know more about the surface of Mars

2323
02:08:57,600 --> 02:09:00,557
than we do our ocean floor.

2324
02:09:00,640 --> 02:09:03,277
>> Uh Richard,
uh this is the part where we

2325
02:09:03,360 --> 02:09:08,157
answer our viewers questions with an AMA.
And as our guest, I'm going to give

2326
02:09:08,240 --> 02:09:12,397
you first crack to choose one of these
questions. So, if you could pick the

2327
02:09:12,480 --> 02:09:14,637
number, read the question and who it's

2328
02:09:14,720 --> 02:09:18,237
from, uh, and then dive in.

2329
02:09:18,320 --> 02:09:22,397
>> All right. Well, uh, let me try to scan
them really quick. Um, all right. If AI

2330
02:09:22,480 --> 02:09:26,157
makes companies 10x more productive, but
we don't need 10x the output, where does

2331
02:09:26,240 --> 02:09:29,917
the value go? Shareholders, workers,
or does it evaporate? I think this is

2332
02:09:30,000 --> 02:09:33,837
actually something I have thought about
in the past. Uh I think we can predict

2333
02:09:33,920 --> 02:09:36,637
the impact of jobs uh in a certain

2334
02:09:36,720 --> 02:09:39,837
industry from AI based on the elasticity

2335
02:09:39,920 --> 02:09:42,477
of the demand when the price of that

2336
02:09:42,560 --> 02:09:45,357
product goes massively down. We don't

2337
02:09:45,440 --> 02:09:48,397
need to have billions and billions of

2338
02:09:48,480 --> 02:09:49,917
illustrations in the world. And so when

2339
02:09:50,000 --> 02:09:52,957
AI made the price of one illustration go

2340
02:09:53,040 --> 02:09:55,837
down from 200 bucks to like 2 cents or

2341
02:09:55,920 --> 02:09:59,357
less, like we just didn't need as many

2342
02:09:59,440 --> 02:10:04,477
illustrators anymore. um uh uh because
the demand for illustrations didn't go

2343
02:10:04,560 --> 02:10:08,157
massively up. Yes, every little blog
post and every little tweet can now have

2344
02:10:08,240 --> 02:10:10,237
a beautiful visualization and and

2345
02:10:10,320 --> 02:10:13,037
illustration. Uh but uh you know, we

2346
02:10:13,120 --> 02:10:15,197
didn't need many more billions of them.

2347
02:10:15,280 --> 02:10:17,277
Uh I think software is a different one.

2348
02:10:17,360 --> 02:10:19,517
You can actually have everyone can have

2349
02:10:19,600 --> 02:10:24,477
several pieces of software just
specific to them. So we can actually have

2350
02:10:24,560 --> 02:10:29,277
billions of different software products
uh customized for each person. uh and so

2351
02:10:29,360 --> 02:10:33,597
the demand uh for that product will go up.
Jevans paradox is going to be alive

2352
02:10:33,680 --> 02:10:35,757
uh in that world and we're going to see

2353
02:10:35,840 --> 02:10:38,077
more and more demand and there will be

2354
02:10:38,160 --> 02:10:41,997
uh more uh value acrewing to everyone. I

2355
02:10:42,080 --> 02:10:43,517
think in terms of shareholders versus

2356
02:10:43,600 --> 02:10:48,397
workers I think the wave of AI in the

2357
02:10:48,480 --> 02:10:52,477
best scenario will be a huge force
for more entrepreneurship and in the worst

2358
02:10:52,560 --> 02:10:54,557
case scenario a force of more

2359
02:10:54,640 --> 02:10:58,477
inequality. I think uh everyone who owns

2360
02:10:58,560 --> 02:11:01,357
some equity in a company that uses AI

2361
02:11:01,440 --> 02:11:02,957
can love AI.

2362
02:11:03,040 --> 02:11:06,157
>> Wonderful. Sele over to you.

2363
02:11:06,240 --> 02:11:10,477
>> Uh I will take uh
question number three.

2364
02:11:10,560 --> 02:11:14,077
Um if AI can write code, research,
create marketing, manage projects and

2365
02:11:14,160 --> 02:11:15,997
outcomes, what is left for a college

2366
02:11:16,080 --> 02:11:19,197
graduate in 2030? And that is from Jared

2367
02:11:19,280 --> 02:11:21,597
8812.

2368
02:11:21,680 --> 02:11:25,117
So the obvious answer would be be have

2369
02:11:25,200 --> 02:11:27,837
empathy and be be creative etc. But I

2370
02:11:27,920 --> 02:11:32,157
think the bigger shift which builds on
what Richard just said is you shift from

2371
02:11:32,240 --> 02:11:34,717
doing tasks to focusing on owning

2372
02:11:34,800 --> 02:11:39,197
outcomes, right? So a graduate used
to be valuable because you could you could

2373
02:11:39,280 --> 02:11:43,597
execute research or build a spreadsheet
or draft a deck. But now you want to

2374
02:11:43,680 --> 02:11:47,997
say, "Hey, here are the constraints like
go figure out and and here's how we'll

2375
02:11:48,080 --> 02:11:50,317
know whether we solved it and then use

2376
02:11:50,400 --> 02:11:52,637
an AI to get to it." And it brings you

2377
02:11:52,720 --> 02:11:55,837
back to what should your problem space

2378
02:11:55,920 --> 02:11:58,717
be? And this is the massive opportunity

2379
02:11:58,800 --> 02:12:00,957
because we traditionally learn judgment

2380
02:12:01,040 --> 02:12:03,437
by doing the grunt work. Uh we need a

2381
02:12:03,520 --> 02:12:06,637
totally new apprenticeship model uh when

2382
02:12:06,720 --> 02:12:08,557
the grunt work disappears. And this is a

2383
02:12:08,640 --> 02:12:11,597
huge challenge for the education system.

2384
02:12:11,680 --> 02:12:15,837
But the change is going to be focusing
on what problems you want to solve and

2385
02:12:15,920 --> 02:12:19,757
then letting orchestrating the forces
that will help you solve that problem.

2386
02:12:19,840 --> 02:12:23,117
>> Yeah. Jared, find your purpose, right?
A passion is something you love doing. A

2387
02:12:23,200 --> 02:12:25,917
purpose is something you love
doing that helps other people. uh

2388
02:12:26,000 --> 02:12:28,717
>> make sure it's massive and it's
transformative and it's purposeful

2389
02:12:28,800 --> 02:12:31,277
>> and and build a company and then direct

2390
02:12:31,360 --> 02:12:34,397
AI to implement it. It is your

2391
02:12:34,480 --> 02:12:38,157
workforce. Uh Dave, over to you.

2392
02:12:38,240 --> 02:12:42,077
>> Uh I can't resist number four. I love
all these questions. I'm torn. But how

2393
02:12:42,160 --> 02:12:45,517
can data centers make neighborhoods
richer instead of the owners? And that's

2394
02:12:45,600 --> 02:12:48,877
from LM Bman66.

2395
02:12:48,960 --> 02:12:53,357
Um, so, uh, I took a tour of the LOL
data center, Jeff Markley, Markley data

2396
02:12:53,440 --> 02:12:56,397
center, and that thing is creating
wealth in that neighborhood like you

2397
02:12:56,480 --> 02:13:00,157
wouldn't believe. And, uh, the way it
works fundamentally is the data center

2398
02:13:00,240 --> 02:13:04,397
is so immensely valuable and the town
budget is maybe a couple million dollars

2399
02:13:04,480 --> 02:13:06,957
a year. So, between the tax revenue, the

2400
02:13:07,040 --> 02:13:11,677
donations, and the job creation,
the town is thriving and and it's a town

2401
02:13:11,760 --> 02:13:15,837
that really needed it, too. So, I think
it's happening very naturally. What you

2402
02:13:15,920 --> 02:13:19,277
want to do is attract a data center to
your neighborhood first and foremost and

2403
02:13:19,360 --> 02:13:22,957
then you have the next five or seven
years to figure out your tax policy,

2404
02:13:23,040 --> 02:13:26,317
your donation policy. I tell you
these data center operators are very

2405
02:13:26,400 --> 02:13:30,717
interested in great PR and so they
they will donate like crazy to the high

2406
02:13:30,800 --> 02:13:34,957
schools to the neighborhoods to it's
really really working. We're going to

2407
02:13:35,040 --> 02:13:38,477
see a whole we're going to see an entire
shift where data centers are offering

2408
02:13:38,560 --> 02:13:41,517
such benefits on jobs, on tax breaks, on

2409
02:13:41,600 --> 02:13:45,837
lowerc cost energy, uh that you're going
to be begging to have a data center in

2410
02:13:45,920 --> 02:13:48,637
your backyard. Alex,
number two is for you.

2411
02:13:48,720 --> 02:13:52,797
>> All right, so number two asks, if
every major tech platform started open and

2412
02:13:52,880 --> 02:13:55,597
democratic than consolidated power,

2413
02:13:55,680 --> 02:13:58,637
Google, Meta, Amazon, why would AI be

2414
02:13:58,720 --> 02:14:00,477
different? And this is from open-source

2415
02:14:00,560 --> 02:14:03,517
mind. The premise of the question is

2416
02:14:03,600 --> 02:14:06,157
half right half wrong. I I'm not sure I

2417
02:14:06,240 --> 02:14:09,597
buy the premise that consolidated power

2418
02:14:09,680 --> 02:14:12,477
the subtext of which and one can

2419
02:14:12,560 --> 02:14:17,037
juxtapose it with the the user handle
open source mind. I I'm not sure the

2420
02:14:17,120 --> 02:14:19,117
framing is the right framing. I I would

2421
02:14:19,200 --> 02:14:22,557
agree that in every major tech

2422
02:14:22,640 --> 02:14:25,197
revolution initially the barrier to

2423
02:14:25,280 --> 02:14:28,077
entry is low because there's some new

2424
02:14:28,160 --> 02:14:30,317
platform innovation and then you see

2425
02:14:30,400 --> 02:14:32,957
lots and lots of players enter the field

2426
02:14:33,040 --> 02:14:35,437
and then as the field matures you see

2427
02:14:35,520 --> 02:14:39,117
economies of scale and a deeper bench of

2428
02:14:39,200 --> 02:14:43,677
infrastructure typically supporting it
and as a result that favors larger and

2429
02:14:43,760 --> 02:14:48,557
larger players and you do see
consolidation but the the subtext of the

2430
02:14:48,640 --> 02:14:52,877
question that it's somehow
anti-democratic or not open. I don't

2431
02:14:52,960 --> 02:14:55,517
agree with that premise in the least. I

2432
02:14:55,600 --> 02:14:58,957
I do think as you see consolidation in

2433
02:14:59,040 --> 02:15:02,637
an industry, I think it's important
to be vigilant from an antitrust

2434
02:15:02,720 --> 02:15:04,797
perspective to make sure that it remains

2435
02:15:04,880 --> 02:15:07,517
competitive, but the premise that

2436
02:15:07,600 --> 02:15:12,637
hyperscaling is somehow closed
or anti-democratic, don't buy the premise

2437
02:15:12,720 --> 02:15:14,157
at all.

2438
02:15:14,240 --> 02:15:17,437
>> All right. Uh,
Richard as our guest, you

2439
02:15:17,520 --> 02:15:22,760
get first crack once again. Take a look.

2440
02:15:24,240 --> 02:15:29,117
>> I do think the internet adbased
economy uh will be under pressure.

2441
02:15:29,200 --> 02:15:30,557
>> Which which question are you answering?

2442
02:15:30,640 --> 02:15:33,437
>> So, first question. If AI agents
outnumber humans in one years on the

2443
02:15:33,520 --> 02:15:37,837
internet, doesn't the entire ad- based
economy collapse? Uh, what replaces the

2444
02:15:37,920 --> 02:15:41,597
attention economy? I think there there's
actually a really deep answer here. I'll

2445
02:15:41,680 --> 02:15:45,357
try to summarize it, but like one,
we already have more bots on the internet

2446
02:15:45,440 --> 02:15:49,597
uh and more agents on the internet than
people. So this I predicted this uh last

2447
02:15:49,680 --> 02:15:51,517
year and it happened a few months ago.

2448
02:15:51,600 --> 02:15:53,757
Um so that's number one. I do think

2449
02:15:53,840 --> 02:15:57,837
we're seeing the first kind of um

2450
02:15:57,920 --> 02:16:01,437
sort of skirmishes uh in that um when

2451
02:16:01,520 --> 02:16:04,877
for instance uh various agents try to

2452
02:16:04,960 --> 02:16:08,237
make purchases on Amazon without really

2453
02:16:08,320 --> 02:16:12,237
being on the Amazon platform and Amazon
usually tries to turn them off cuz they

2454
02:16:12,320 --> 02:16:14,157
want to own that relationship directly

2455
02:16:14,240 --> 02:16:17,197
with the customer understandably. Um but

2456
02:16:17,280 --> 02:16:21,357
it's just convenient for someone to just
say just go buy these batteries and they

2457
02:16:21,440 --> 02:16:24,957
don't care which batteries it is. And so
if you have that control over which one

2458
02:16:25,040 --> 02:16:27,917
it is, you can start selling that uh to

2459
02:16:28,000 --> 02:16:31,117
uh other uh sellers and and uh people

2460
02:16:31,200 --> 02:16:33,117
who create physical goods. And so there

2461
02:16:33,200 --> 02:16:36,317
is uh going to be uh continued uh like

2462
02:16:36,400 --> 02:16:38,317
ads will continue to be important. And

2463
02:16:38,400 --> 02:16:41,917
in fact, if you think about a fully

2464
02:16:42,000 --> 02:16:44,397
abundant society, the one thing that you

2465
02:16:44,480 --> 02:16:47,197
cannot scale exponentially is the hours

2466
02:16:47,280 --> 02:16:49,037
in the day that people can pay attention

2467
02:16:49,120 --> 02:16:52,077
to you that can make you famous. Uh and

2468
02:16:52,160 --> 02:16:55,277
so fame and brand and network effects uh

2469
02:16:55,360 --> 02:16:59,757
and other things like that will become
more and more of a currency. And so the

2470
02:16:59,840 --> 02:17:02,237
attention economy which is connected

2471
02:17:02,320 --> 02:17:05,517
slightly uh not necessarily uh exactly

2472
02:17:05,600 --> 02:17:10,077
equivalent to the ad-based economy uh
that actually will become a bigger thing

2473
02:17:10,160 --> 02:17:12,317
as we're have more and more of our

2474
02:17:12,400 --> 02:17:15,677
material needs met by uh technology.

2475
02:17:15,760 --> 02:17:18,317
>> Great. Uh Salem

2476
02:17:18,400 --> 02:17:22,877
>> I will take number uh seven. If we

2477
02:17:22,960 --> 02:17:26,877
remove 10x the cars from the street what
happens to the insurance industry? We

2478
02:17:26,960 --> 02:17:29,197
won't need driver's insurance. Mark and

2479
02:17:29,280 --> 02:17:33,277
that's from Robert Zerby JB10H. Um so

2480
02:17:33,360 --> 02:17:35,597
you know we talked about how um

2481
02:17:35,680 --> 02:17:40,477
liability lawyers won't be needed and I
got a huge flame from a bunch of folks

2482
02:17:40,560 --> 02:17:45,437
saying hey we really protect
the citizenry and I there are just so I'm

2483
02:17:45,520 --> 02:17:48,477
just apologizing because there's
a spectrum of people and some people are

2484
02:17:48,560 --> 02:17:52,717
full ambulance chasers and other people
do things. There's a whole segment of

2485
02:17:52,800 --> 02:17:56,797
this called ethical lawsuits where
people get together and try and uh sue

2486
02:17:56,880 --> 02:18:00,237
big companies for doing the right
ethical thing. And so that's an

2487
02:18:00,320 --> 02:18:04,157
important uh segment of that. So I just
want to acknowledge that side of it. But

2488
02:18:04,240 --> 02:18:06,557
just to answer the question, um you know

2489
02:18:06,640 --> 02:18:09,117
the the thing is industries don't uh

2490
02:18:09,200 --> 02:18:11,917
disappear when the risk changes. You

2491
02:18:12,000 --> 02:18:14,077
change the risk, right? So driver

2492
02:18:14,160 --> 02:18:16,637
liability may fail, software liability

2493
02:18:16,720 --> 02:18:19,357
rises, right? You've you what's your uh

2494
02:18:19,440 --> 02:18:21,357
cyber risk? what's your manufacturer

2495
02:18:21,440 --> 02:18:24,237
liability risk? Um, so the in the

2496
02:18:24,320 --> 02:18:26,877
insurance transitions from did did

2497
02:18:26,960 --> 02:18:29,757
Seline crash to which layer of the

2498
02:18:29,840 --> 02:18:33,277
autonomous stack failed and and we we've

2499
02:18:33,360 --> 02:18:37,037
had this before in product liability
is different layers. So we'll end up with

2500
02:18:37,120 --> 02:18:41,117
the same type of model. You shift
the insurance risk to a different level.

2501
02:18:41,200 --> 02:18:44,077
>> And there's going to be all kinds of
new insurance markets for humanoid robots,

2502
02:18:44,160 --> 02:18:46,637
for flying cars, for drones,
for all kinds of things.

2503
02:18:46,720 --> 02:18:49,917
>> Well, just a data point on that.
You know, a single big data center like

2504
02:18:50,000 --> 02:18:54,477
Abalene, Texas is half a trillion dollars.
All the cars combined are 4

2505
02:18:54,560 --> 02:18:58,557
trillion. One data center is half
a trillion. And it's in a tornado alley.

2506
02:18:58,640 --> 02:19:01,437
You think you probably want to ensure
that. So the things number of things

2507
02:19:01,520 --> 02:19:05,997
that need insurance is going up 10x just
with the economy going up 10x. You just

2508
02:19:06,080 --> 02:19:07,917
need to move. >> If you
really if you really want to be

2509
02:19:08,000 --> 02:19:10,797
brave, you can't get home
insurance in Florida anymore.

2510
02:19:10,880 --> 02:19:13,837
>> So go create an insurance
company for that.

2511
02:19:13,920 --> 02:19:17,037
>> Nice. Uh Dave, pick your question.

2512
02:19:17,120 --> 02:19:20,397
I like number eight. Uh,
what is the lowest possible job in an AI

2513
02:19:20,480 --> 02:19:22,077
civilization? What a what an interesting

2514
02:19:22,160 --> 02:19:26,557
question. I that's from LBN ODK. Um,

2515
02:19:26,640 --> 02:19:29,117
yeah, you know, I I saw this pile of a

2516
02:19:29,200 --> 02:19:34,397
million valves for a liquid cooled data
center and a million freaking valves.

2517
02:19:34,480 --> 02:19:38,557
I'm like, how do those actually end up
in pipes? The humanoid robots that can

2518
02:19:38,640 --> 02:19:40,237
install those things are pretty far out.

2519
02:19:40,320 --> 02:19:45,117
It's a very subtle process to to install
those. So that job will be around for a

2520
02:19:45,200 --> 02:19:49,037
long time and then they're paying a lot
for it. But that's not the lowest I'm

2521
02:19:49,120 --> 02:19:52,877
it's it's hard to think like what is the
lowest surviving job. Do you guys have

2522
02:19:52,960 --> 02:19:54,317
any thoughts?

2523
02:19:54,400 --> 02:19:57,277
>> So many thoughts.
If if I may, it I want

2524
02:19:57,360 --> 02:19:59,917
to conrue the question as in a pure AI

2525
02:20:00,000 --> 02:20:02,397
civilization, in which case arguably the

2526
02:20:02,480 --> 02:20:05,117
way that you measure low is the the job

2527
02:20:05,200 --> 02:20:09,837
that requires the least compute, in which
case the the jobs that require the

2528
02:20:09,920 --> 02:20:11,757
least compute as a results are the least

2529
02:20:11,840 --> 02:20:14,157
economically valuable if if their inputs

2530
02:20:14,240 --> 02:20:16,877
are the lowest might ironically look

2531
02:20:16,960 --> 02:20:19,677
like the most valuable jobs in a pre-I

2532
02:20:19,760 --> 02:20:22,317
civilization. More of a paradox style.

2533
02:20:22,400 --> 02:20:25,357
So the the the great writers, business

2534
02:20:25,440 --> 02:20:28,157
people, uh the the the creative actors

2535
02:20:28,240 --> 02:20:31,517
that Einrand may fetishize would

2536
02:20:31,600 --> 02:20:34,317
ironically in a post AAI civilization be

2537
02:20:34,400 --> 02:20:38,637
the lowest possible jobs because Moravec
paradox paradox style those were the

2538
02:20:38,720 --> 02:20:42,237
first ones to be automated.
>> Interesting. All right.

2539
02:20:42,320 --> 02:20:43,837
>> I think there's two similarities here.

2540
02:20:43,920 --> 02:20:46,237
One is very high judgment per what you

2541
02:20:46,320 --> 02:20:47,997
were just saying, Alex, and second is

2542
02:20:48,080 --> 02:20:51,197
very physical, highly contextual work.

2543
02:20:51,280 --> 02:20:53,357
I would I would argue maybe also low is

2544
02:20:53,440 --> 02:20:56,077
just in terms of how valuable uh and and

2545
02:20:56,160 --> 02:20:58,957
and moral society deems those kinds of

2546
02:20:59,040 --> 02:21:01,997
jobs. And I do think actually that uh

2547
02:21:02,080 --> 02:21:06,477
the other question about the cars on
the street um I don't think we removed uh

2548
02:21:06,560 --> 02:21:10,877
10x the cars when we have self-driving
but self-driving might be making things

2549
02:21:10,960 --> 02:21:13,357
so much safer that indeed uh people

2550
02:21:13,440 --> 02:21:18,157
don't need as much accident insurance
and no not as many ER people not as many

2551
02:21:18,240 --> 02:21:21,997
ambulances and so in a weird way you
make objectively the world better by

2552
02:21:22,080 --> 02:21:26,157
reducing traffic deaths but it does
actually have potentially a mildly

2553
02:21:26,240 --> 02:21:30,797
negative effect on parts of the economy
and we should all be rooting for that uh

2554
02:21:30,880 --> 02:21:35,357
in this case, right? And so I think
the lowest jobs in terms of moral standing

2555
02:21:35,440 --> 02:21:39,517
are things where you don't really
progress uh humanity forward. And my

2556
02:21:39,600 --> 02:21:41,997
hunch is there the types of jobs in

2557
02:21:42,080 --> 02:21:44,237
entertainment um that that people will

2558
02:21:44,320 --> 02:21:47,437
value a lot and and fame and attention

2559
02:21:47,520 --> 02:21:51,277
uh will become more of a currency
in that world. And so if your job doesn't

2560
02:21:51,360 --> 02:21:56,077
get you any of that, it might be
considered uh lower in in that future. I

2561
02:21:56,160 --> 02:21:59,277
think the lowest job that just will
never go away will be something in

2562
02:21:59,360 --> 02:22:03,837
politics where it's completely
irrelevant already, but but it's just

2563
02:22:03,920 --> 02:22:06,477
there and it'll stay there
and no one's going to change it.

2564
02:22:06,560 --> 02:22:10,157
>> Nice. All right. >> I got
to throw in ID's perspective. He

2565
02:22:10,240 --> 02:22:13,757
always thinks it's the BART train
driver because it's unionized to hell.

2566
02:22:13,840 --> 02:22:16,717
>> Yeah. Yeah. There you go.
>> Alex, number six, close us out.

2567
02:22:16,800 --> 02:22:20,957
>> Number six. If we cure illnesses
that are often due to bad behavior, what's

2568
02:22:21,040 --> 02:22:25,117
going to take care of the cause? Joe
Wilder. I assume the cause is reference

2569
02:22:25,200 --> 02:22:27,757
to people choosing to behave quote

2570
02:22:27,840 --> 02:22:32,317
unquote badly. And the answer I this
is another case of my not buying the

2571
02:22:32,400 --> 02:22:35,677
premise. If you look at some of the
really spectacular results that have

2572
02:22:35,760 --> 02:22:38,237
been coming out of GLP1 class studies,

2573
02:22:38,320 --> 02:22:40,797
including third and soon presumably

2574
02:22:40,880 --> 02:22:43,997
fourth generation GLP1 RAS, they're

2575
02:22:44,080 --> 02:22:48,397
actually addressing the cause. like
addictive behaviors are being mitigated

2576
02:22:48,480 --> 02:22:50,957
or at least partially treated by the

2577
02:22:51,040 --> 02:22:53,197
same drugs that are curing in or at

2578
02:22:53,280 --> 02:22:55,997
least treating have to to caveat that

2579
02:22:56,080 --> 02:23:00,317
inflammation and blood sugar, diabetes,

2580
02:23:00,400 --> 02:23:02,477
all of these other conditions the the

2581
02:23:02,560 --> 02:23:04,717
root causes which is or human behavior

2582
02:23:04,800 --> 02:23:07,597
are themselves the the human behavior is

2583
02:23:07,680 --> 02:23:10,557
being affected by the same drugs. So I'm

2584
02:23:10,640 --> 02:23:14,717
so optimistic about what Alex is saying
because I really fully believe AI this

2585
02:23:14,800 --> 02:23:18,557
is going to be one of the highest
callings of AI very very soon is to make

2586
02:23:18,640 --> 02:23:22,157
you feel really good about doing good
things and happy as you're doing it

2587
02:23:22,240 --> 02:23:24,157
>> love >> and not want
not want to do bad things

2588
02:23:24,240 --> 02:23:27,517
for yourself. It turns out we have the
capability now. We figured out how to do

2589
02:23:27,600 --> 02:23:30,717
that at least in part and we're
going to figure out a lot more.

2590
02:23:30,800 --> 02:23:33,037
>> All right, as always a call out to our

2591
02:23:33,120 --> 02:23:35,757
amazing community. If you've got a music

2592
02:23:35,840 --> 02:23:40,077
video that you'd like to show as outro,
please send it to the team at

2593
02:23:40,160 --> 02:23:42,317
mediadmandis.com.

2594
02:23:42,400 --> 02:23:45,117
And speaking about amazing outros, here

2595
02:23:45,200 --> 02:23:48,797
is Abundance by Steven Gross. Uh my dear

2596
02:23:48,880 --> 02:23:55,160
Moonshot mates, this is the real you.
So check it out.

2597
02:23:55,920 --> 02:24:01,720
>> Got it. Peter a wgave.

2598
02:24:03,520 --> 02:24:08,520
This one's for you. >> Bun dance.

2599
02:24:09,920 --> 02:24:13,880
Everybody do the bun dance.

2600
02:24:16,800 --> 02:24:20,157
Everybody do the bun dance. Thinking

2601
02:24:20,240 --> 02:24:24,237
moonshot dreaming. A moving future

2602
02:24:24,320 --> 02:24:27,837
screaming. Build it faster. Think much

2603
02:24:27,920 --> 02:24:30,717
bigger. Exponential.

2604
02:24:30,800 --> 02:24:33,117
Pull the trigger.

2605
02:24:33,200 --> 02:24:37,757
Drink, Alec. Drink >> drink water

2606
02:24:37,840 --> 02:24:41,277
martini glass. >> Oh,
that was fun. That was Moonshots

2607
02:24:41,360 --> 02:24:45,757
Live 2040 on the moon. >> Nice. Uh,
so >> I don't think we should have to wait

2608
02:24:45,840 --> 02:24:49,437
that long. >> Yeah. Uh,
Richard, your new book, The

2609
02:24:49,520 --> 02:24:53,277
Eureka Machine, wherever you purchase
your books. Do you have the Audible out?

2610
02:24:53,360 --> 02:24:55,757
>> Uh, should come out very soon. Yeah.

2611
02:24:55,840 --> 02:24:59,117
>> All right. I'm an Audible reader,
but I have skimmed the book here. Uh,

2612
02:24:59,200 --> 02:25:02,957
congratulations. Say, Richard, there's
I got to say something. What you're doing

2613
02:25:03,040 --> 02:25:05,037
with the recursive looks like such an

2614
02:25:05,120 --> 02:25:07,277
incredible opportunity to move humanity

2615
02:25:07,360 --> 02:25:10,477
forward. So, congrats on taking

2616
02:25:10,560 --> 02:25:14,797
it a good one for humanity.
>> And and Richard, please resist the urge

2617
02:25:14,880 --> 02:25:18,957
to have you.com acquire your own
Frontier Lab like everyone else seems to

2618
02:25:19,040 --> 02:25:23,277
be using, spinning off their own
Frontier Labs, not to name names, uh, as

2619
02:25:23,360 --> 02:25:27,117
a a financial engineering exercise to
maximize their equity and their original

2620
02:25:27,200 --> 02:25:30,157
startup. Please resist the urge
to to follow that trend.

2621
02:25:30,240 --> 02:25:31,597
>> All right.

2622
02:25:31,680 --> 02:25:33,837
>> All right. I love you guys.

2623
02:25:33,920 --> 02:25:38,120
>> Be well. Until next time.