1
00:00:30,682 --> 00:00:31,814
Hi, Alpha.

2
00:00:32,684 --> 00:00:34,034
Hello.

3
00:00:34,077 --> 00:00:35,296
Can you
help me write code?

4
00:00:37,559 --> 00:00:39,082
I was trained
to answer questions,

5
00:00:40,562 --> 00:00:42,216
but I'm able to learn.

6
00:00:43,695 --> 00:00:45,958
That's very
open-minded of you.

7
00:00:46,002 --> 00:00:48,657
Thank you.
I'm glad you're happy with me.

8
00:00:50,311 --> 00:00:51,312
What's this guy doing?

9
00:00:54,532 --> 00:00:56,621
That's a developer.

10
00:00:56,665 --> 00:00:58,014
What do you think
he's working on?

11
00:00:59,537 --> 00:01:01,017
That's a tough question.

12
00:01:01,061 --> 00:01:02,888
He might be working
on a new feature,

13
00:01:02,932 --> 00:01:05,239
a bug fix or something else.

14
00:01:05,282 --> 00:01:06,327
It's quite possible.

15
00:01:06,370 --> 00:01:07,328
Yes.

16
00:01:10,548 --> 00:01:11,854
Do you see my backpack?

17
00:01:13,986 --> 00:01:15,292
That's a badminton racket.

18
00:01:15,336 --> 00:01:17,903
It's a squash racket,
but that's pretty close.

19
00:01:20,123 --> 00:01:21,690
That's a badminton racket.

20
00:01:21,733 --> 00:01:23,126
No, but you're not
the first person

21
00:01:23,170 --> 00:01:24,649
to make that mistake.

22
00:01:34,703 --> 00:01:35,965
AI, the technology

23
00:01:36,008 --> 00:01:38,402
that has been advancing
at breakneck speed.

24
00:01:38,446 --> 00:01:40,796
Artificial
intelligence is all the rage.

25
00:01:40,839 --> 00:01:42,319
Some are now
raising alarm about...

26
00:01:42,363 --> 00:01:43,886
It is definitely concerning.

27
00:01:43,929 --> 00:01:45,322
This is an AI arms race.

28
00:01:45,366 --> 00:01:46,454
We don't know

29
00:01:46,497 --> 00:01:47,585
how this is all
going to shake out,

30
00:01:47,629 --> 00:01:48,891
but it's clear
something is happening.

31
00:01:53,809 --> 00:01:55,115
I'm kind of restless.

32
00:01:57,073 --> 00:02:00,120
Trying to build AGI
is the most exciting journey,

33
00:02:00,163 --> 00:02:02,383
in my opinion, that humans
have ever embarked on.

34
00:02:04,254 --> 00:02:06,300
If you're really going
to take that seriously,

35
00:02:06,343 --> 00:02:08,345
there isn't a lot of time.

36
00:02:08,389 --> 00:02:09,825
Life's very short.

37
00:02:12,523 --> 00:02:14,482
My whole life goal is to solve

38
00:02:14,525 --> 00:02:16,875
artificial
general intelligence.

39
00:02:16,919 --> 00:02:20,357
And on the way,
use AI as the ultimate tool

40
00:02:20,401 --> 00:02:21,750
to solve all the world's

41
00:02:21,793 --> 00:02:23,230
most complex
scientific problems.

42
00:02:25,449 --> 00:02:27,103
I think that's bigger
than the Internet.

43
00:02:27,147 --> 00:02:28,452
I think that's bigger
than mobile.

44
00:02:29,671 --> 00:02:31,020
I think it's more like

45
00:02:31,063 --> 00:02:32,891
the advent
of electricity or fire.

46
00:02:45,687 --> 00:02:46,731
World leaders

47
00:02:46,775 --> 00:02:48,516
and artificial
intelligence experts

48
00:02:48,559 --> 00:02:49,908
are gathering
for the first ever

49
00:02:49,952 --> 00:02:52,781
global AI safety summit,

50
00:02:52,824 --> 00:02:54,348
set to look at the risks

51
00:02:54,391 --> 00:02:56,611
of the fast growing technology
and also...

52
00:02:56,654 --> 00:02:57,873
I think
this is a hugely

53
00:02:57,916 --> 00:03:00,136
critical moment
for all humanity.

54
00:03:01,093 --> 00:03:03,618
It feels like
we're on the cusp

55
00:03:03,661 --> 00:03:05,794
of some incredible things
happening.

56
00:03:05,837 --> 00:03:07,056
Let me take you through

57
00:03:07,099 --> 00:03:08,275
some of the reactions
in today's papers.

58
00:03:08,318 --> 00:03:10,015
AGI is pretty close,
I think.

59
00:03:10,059 --> 00:03:12,844
There's clearly huge interest
in what it is capable of,

60
00:03:12,888 --> 00:03:14,803
where it's taking us.

61
00:03:14,846 --> 00:03:15,978
This is the moment

62
00:03:16,021 --> 00:03:17,936
I've been living
my whole life for.

63
00:03:22,202 --> 00:03:24,987
I've always been fascinated
by the mind.

64
00:03:25,030 --> 00:03:27,990
So I set my heart
on studying neuroscience

65
00:03:28,033 --> 00:03:29,948
because I wanted
to get inspiration

66
00:03:29,992 --> 00:03:31,646
from the brain for AI.

67
00:03:31,689 --> 00:03:33,300
I remember asking Demis,

68
00:03:33,343 --> 00:03:34,649
"What's the end game?"

69
00:03:34,692 --> 00:03:36,172
You know?
So you're going to come here

70
00:03:36,216 --> 00:03:38,218
and you're going
to study neuroscience

71
00:03:38,261 --> 00:03:41,438
and you're going to maybe
get a Ph.D. if you work hard.

72
00:03:42,874 --> 00:03:44,136
And he said,

73
00:03:44,180 --> 00:03:46,965
"You know, I want
to be able to solve AI.

74
00:03:47,009 --> 00:03:49,359
"I want to be able
to solve intelligence."

75
00:03:49,403 --> 00:03:51,361
The human brain
is the only existent proof

76
00:03:51,405 --> 00:03:53,494
we have, perhaps
in the entire universe,

77
00:03:53,537 --> 00:03:56,279
that general intelligence
is possible at all.

78
00:03:56,323 --> 00:03:59,282
And I thought
someone in this building

79
00:03:59,326 --> 00:04:00,370
should be interested

80
00:04:00,414 --> 00:04:02,590
in general intelligence
like I am.

81
00:04:02,633 --> 00:04:04,809
And then Shane's name
popped up.

82
00:04:04,853 --> 00:04:07,334
Our next speaker today
is Shane Legg.

83
00:04:07,377 --> 00:04:08,596
He's from New Zealand,

84
00:04:08,639 --> 00:04:11,512
where he trained in math
and classical ballet.

85
00:04:11,555 --> 00:04:14,079
Are machines actually
becoming more intelligent?

86
00:04:14,123 --> 00:04:16,517
Some people say yes,
some people say no.

87
00:04:16,560 --> 00:04:17,648
It's not really clear.

88
00:04:17,692 --> 00:04:18,867
We know they're getting
a lot faster

89
00:04:18,910 --> 00:04:20,216
at doing computations.

90
00:04:20,260 --> 00:04:21,609
But are we actually
going forwards

91
00:04:21,652 --> 00:04:23,828
in terms
of general intelligence?

92
00:04:23,872 --> 00:04:25,700
We were both
obsessed with AGI,

93
00:04:25,743 --> 00:04:27,397
artificial
general intelligence.

94
00:04:27,441 --> 00:04:29,704
So today I'm going
to be talking about

95
00:04:29,747 --> 00:04:32,097
different approaches
to building AGI.

96
00:04:32,141 --> 00:04:33,882
With my colleague
Demis Hassabis,

97
00:04:33,925 --> 00:04:35,884
we're looking at ways
to bring in ideas

98
00:04:35,927 --> 00:04:37,451
from theoretical neuroscience.

99
00:04:37,494 --> 00:04:41,585
I felt like we were
the keepers of a secret

100
00:04:41,629 --> 00:04:43,152
that no one else knew.

101
00:04:43,195 --> 00:04:45,415
Shane and I knew
no one in academia

102
00:04:45,459 --> 00:04:47,809
would be supportive
of what we were doing.

103
00:04:47,852 --> 00:04:50,855
AI was almost
an embarrassing word

104
00:04:50,899 --> 00:04:52,857
to use in academic circles,
right?

105
00:04:52,901 --> 00:04:54,903
If you said
you were working on AI,

106
00:04:54,946 --> 00:04:57,993
then you clearly weren't
a serious scientist.

107
00:04:58,036 --> 00:05:00,082
So I convinced Shane
the right way to do it

108
00:05:00,125 --> 00:05:01,344
would be to start a company.

109
00:05:01,388 --> 00:05:03,215
Okay,
we're going to try to do

110
00:05:03,259 --> 00:05:04,913
artificial
general intelligence.

111
00:05:04,956 --> 00:05:06,915
It may not even be possible.

112
00:05:06,958 --> 00:05:08,482
We're not quite sure
how we're going to do it,

113
00:05:08,525 --> 00:05:11,528
but we have some ideas
or, kind of, approaches.

114
00:05:11,572 --> 00:05:14,966
Huge amounts of money,
huge amounts of risk,

115
00:05:15,010 --> 00:05:16,533
lots and lots of compute.

116
00:05:18,709 --> 00:05:20,320
And if we pull this off,

117
00:05:20,363 --> 00:05:23,540
it'll be the biggest thing
ever, right?

118
00:05:23,584 --> 00:05:26,108
That is a very hard thing
for a typical investor

119
00:05:26,151 --> 00:05:27,327
to put their money on.

120
00:05:27,370 --> 00:05:29,633
It's almost like
buying a lottery ticket.

121
00:05:29,677 --> 00:05:32,462
I'm going to be speaking about
the system of neuroscience

122
00:05:32,506 --> 00:05:36,727
and how it might be used
to help us build AGI.

123
00:05:36,771 --> 00:05:37,946
Finding initial funding

124
00:05:37,989 --> 00:05:39,208
for this was very hard.

125
00:05:39,251 --> 00:05:41,253
We're going to solve
all of intelligence.

126
00:05:41,297 --> 00:05:42,907
You can imagine
some of the looks I got

127
00:05:42,951 --> 00:05:44,779
when we were
pitching that around.

128
00:05:44,822 --> 00:05:47,912
So I'm a V.C.
and I look at about

129
00:05:47,956 --> 00:05:51,002
700 to 1,000 projects a year.

130
00:05:51,046 --> 00:05:54,789
And I fund
literally 1% of those.

131
00:05:54,832 --> 00:05:57,139
About eight projects a year.

132
00:05:57,182 --> 00:06:00,272
So that means 99% of the time,
you're in "No" mode.

133
00:06:00,316 --> 00:06:01,622
"Wait a minute.
I'm telling you,

134
00:06:01,665 --> 00:06:03,798
"this is the most important
thing of all time.

135
00:06:03,841 --> 00:06:05,103
"I'm giving you
all this build-up

136
00:06:05,147 --> 00:06:06,235
"about how... explain

137
00:06:06,278 --> 00:06:07,671
"how it connects
with the brain,

138
00:06:07,715 --> 00:06:09,412
"why the time's right now,
and then you're asking me,

139
00:06:09,456 --> 00:06:10,979
"'But what's your, how are you
going to make money?

140
00:06:11,022 --> 00:06:12,154
"'What's your product?'"

141
00:06:12,197 --> 00:06:15,984
It's like,
so prosaic a question.

142
00:06:16,506 --> 00:06:17,942
You know?

143
00:06:17,986 --> 00:06:19,117
"Have you not been listening
to what I've been saying?"

144
00:06:19,161 --> 00:06:21,076
We needed investors

145
00:06:21,119 --> 00:06:23,470
who aren't necessarily
going to invest

146
00:06:23,513 --> 00:06:25,123
because they think
it's the best

147
00:06:25,167 --> 00:06:26,690
investment decision.

148
00:06:26,734 --> 00:06:27,865
They're probably
going to invest

149
00:06:27,909 --> 00:06:29,476
because they just think
it's really cool.

150
00:06:29,519 --> 00:06:31,478
He's the Silicon Valley

151
00:06:31,521 --> 00:06:33,523
version of the man
behind the curtain

152
00:06:33,567 --> 00:06:34,872
inThe Wizard of Oz.

153
00:06:34,916 --> 00:06:36,265
He had a lot to do
with giving you

154
00:06:36,308 --> 00:06:39,007
PayPal, Facebook,
YouTube and Yelp.

155
00:06:39,050 --> 00:06:40,443
If everyone says "X,"

156
00:06:40,487 --> 00:06:43,185
Peter Thiel suspects
that the opposite of X

157
00:06:43,228 --> 00:06:44,708
is quite possibly true.

158
00:06:44,752 --> 00:06:47,276
So Peter Thiel
was our first big investor.

159
00:06:47,319 --> 00:06:50,105
But he insisted that
we come to Silicon Valley

160
00:06:50,148 --> 00:06:51,802
because that
was the only place we could...

161
00:06:51,846 --> 00:06:52,977
There would be the talent,

162
00:06:53,021 --> 00:06:55,066
and we could build
that kind of company.

163
00:06:55,110 --> 00:06:56,981
But I was pretty adamant
we should be in London

164
00:06:57,025 --> 00:06:59,027
because I think
London's an amazing city.

165
00:06:59,070 --> 00:07:01,464
Plus, I knew there were
really amazing people

166
00:07:01,508 --> 00:07:03,814
trained at Cambridge
and Oxford and UCL.

167
00:07:03,858 --> 00:07:05,033
In Silicon Valley,

168
00:07:05,076 --> 00:07:06,643
everybody's founding
a company every year,

169
00:07:06,687 --> 00:07:07,818
and then if it doesn't work,

170
00:07:07,862 --> 00:07:09,603
you chuck it
and you start something new.

171
00:07:09,646 --> 00:07:11,169
That is not conducive

172
00:07:11,213 --> 00:07:14,564
to a long-term
research challenge.

173
00:07:14,608 --> 00:07:17,349
So we were totally
an outlier for him.

174
00:07:17,393 --> 00:07:20,962
Hi, everyone.
Welcome to DeepMind.

175
00:07:21,005 --> 00:07:22,529
So, what is our mission?

176
00:07:23,747 --> 00:07:25,706
We summarize it as...

177
00:07:25,749 --> 00:07:28,012
DeepMind's mission is to build
the world's first

178
00:07:28,056 --> 00:07:29,579
general learning machine.

179
00:07:29,623 --> 00:07:31,712
So we always stress the word
"general" and "learning" here

180
00:07:31,755 --> 00:07:32,887
are the key things.

181
00:07:32,930 --> 00:07:35,106
Our mission
was to build an AGI,

182
00:07:35,150 --> 00:07:36,760
an artificial
general intelligence.

183
00:07:36,804 --> 00:07:40,372
And so that means that we need
a system which is general.

184
00:07:40,416 --> 00:07:42,723
It doesn't learn to do
one specific thing.

185
00:07:42,766 --> 00:07:45,073
That's a really key part
of human intelligence.

186
00:07:45,116 --> 00:07:46,683
We can learn to do
many, many things.

187
00:07:46,727 --> 00:07:48,685
It's going to, of course,
be a lot of hard work.

188
00:07:48,729 --> 00:07:51,079
But one of the things
that keeps me up at night

189
00:07:51,122 --> 00:07:53,168
is to not waste this
opportunity to, you know,

190
00:07:53,211 --> 00:07:54,430
to really make
a difference here,

191
00:07:54,474 --> 00:07:56,824
and have a big impact
on the world.

192
00:07:56,867 --> 00:07:58,303
The first people
that came

193
00:07:58,347 --> 00:08:00,262
and joined DeepMind
really believed in the dream.

194
00:08:00,305 --> 00:08:02,090
But this was, I think,
one of the first times

195
00:08:02,133 --> 00:08:04,440
they found a place
full of other dreamers.

196
00:08:04,484 --> 00:08:06,398
You know, we collected
this Manhattan Project,

197
00:08:06,442 --> 00:08:08,400
if you like,
together to solve AI.

198
00:08:08,444 --> 00:08:09,750
In the first two years,

199
00:08:09,793 --> 00:08:10,794
we were in total stealth mode.

200
00:08:10,838 --> 00:08:12,274
And so we couldn't
say to anyone

201
00:08:12,317 --> 00:08:14,755
what were we doing
or where we worked.

202
00:08:14,798 --> 00:08:16,104
It was all quite vague.

203
00:08:16,147 --> 00:08:17,584
It had
no public presence at all.

204
00:08:17,627 --> 00:08:18,933
You couldn't
look at a website.

205
00:08:18,976 --> 00:08:21,109
The office
was at a secret location.

206
00:08:21,152 --> 00:08:23,981
When we would interview people
in those early days,

207
00:08:24,025 --> 00:08:26,288
they would show up
very nervously.

208
00:08:27,463 --> 00:08:29,639
I had at least one candidate
who said,

209
00:08:29,683 --> 00:08:31,989
"I just messaged my wife
to tell her exactly

210
00:08:32,033 --> 00:08:33,338
"where I'm going just in case

211
00:08:33,382 --> 00:08:34,514
"this turns out to be some
kind of horrible scam

212
00:08:34,557 --> 00:08:35,819
"and I'm going
to get kidnapped."

213
00:08:35,863 --> 00:08:39,693
Well, my favorite new person
who's an investor,

214
00:08:39,736 --> 00:08:43,000
who I've been working
for a year, is Elon Musk.

215
00:08:43,044 --> 00:08:44,132
So for those of you
who don't know,

216
00:08:44,175 --> 00:08:45,350
this is what he looks like.

217
00:08:45,394 --> 00:08:47,483
And he hadn't really thought
much about AI

218
00:08:47,527 --> 00:08:49,137
until we chatted.

219
00:08:49,180 --> 00:08:51,269
His mission is to die on Mars
or something.

220
00:08:51,313 --> 00:08:52,923
But not on impact.

221
00:08:52,967 --> 00:08:54,185
So...

222
00:08:55,186 --> 00:08:57,145
We made some big decisions

223
00:08:57,188 --> 00:08:59,321
about how we were going
to approach building AI.

224
00:08:59,364 --> 00:09:01,105
This is a reinforcement
learning setup.

225
00:09:01,149 --> 00:09:02,803
This is the kind of setup
that we think about

226
00:09:02,846 --> 00:09:06,328
when we say we're building,
you know, an AI agent.

227
00:09:06,371 --> 00:09:08,504
It's basically the agent,
which is the AI,

228
00:09:08,548 --> 00:09:09,984
and then there's
the environment

229
00:09:10,027 --> 00:09:11,159
that it's interacting with.

230
00:09:11,202 --> 00:09:12,464
We decided that games,

231
00:09:12,508 --> 00:09:13,988
as long as
you're very disciplined

232
00:09:14,031 --> 00:09:15,293
about how you use them,

233
00:09:15,337 --> 00:09:17,252
are the perfect
training ground

234
00:09:17,295 --> 00:09:19,341
for AI development.

235
00:09:19,384 --> 00:09:21,691
We wanted
to try to create one algorithm

236
00:09:21,735 --> 00:09:23,650
that could to be
trained up to play

237
00:09:23,693 --> 00:09:26,043
several dozen
different Atari games.

238
00:09:26,087 --> 00:09:27,479
So just like a human,

239
00:09:27,523 --> 00:09:29,525
you have to use the same brain
to play all the games.

240
00:09:29,569 --> 00:09:30,874
You can think of it

241
00:09:30,918 --> 00:09:33,007
that you provide the agent
with the cartridge.

242
00:09:33,050 --> 00:09:34,312
And you say,

243
00:09:34,356 --> 00:09:35,705
"Okay, imagine you're born
into that world

244
00:09:35,749 --> 00:09:37,533
"with that cartridge,
and you just get to interact

245
00:09:37,577 --> 00:09:39,404
"with the pixels
and see the score.

246
00:09:40,362 --> 00:09:41,668
"What can you do?"

247
00:09:43,974 --> 00:09:47,369
So what you're going to do is
take your Q function. Q-K...

248
00:09:47,412 --> 00:09:49,197
Q-learning
is one of the oldest methods

249
00:09:49,240 --> 00:09:50,894
for reinforcement learning.

250
00:09:50,938 --> 00:09:53,680
And what we did was combine
reinforcement learning

251
00:09:53,723 --> 00:09:56,770
with deep learning
in one system.

252
00:09:56,813 --> 00:09:59,337
No one had ever combined
those two things together

253
00:09:59,381 --> 00:10:01,426
at scale to do
anything impressive,

254
00:10:01,470 --> 00:10:03,515
and we needed
to prove out this thesis.

255
00:10:03,559 --> 00:10:06,780
We tried doingPong
as the first game.

256
00:10:06,823 --> 00:10:08,129
It seemed like the simplest.

257
00:10:08,172 --> 00:10:10,218
It hasn't been told

258
00:10:10,261 --> 00:10:11,828
anything about
what it's controlling

259
00:10:11,872 --> 00:10:12,873
or what it's supposed to do.

260
00:10:12,916 --> 00:10:14,657
All it knows
is that score is good

261
00:10:14,701 --> 00:10:18,052
and it has to learn
what its controls do,

262
00:10:18,095 --> 00:10:20,620
and build everything...
first principles.

263
00:10:29,193 --> 00:10:30,412
It wasn't really working.

264
00:10:32,588 --> 00:10:34,068
I was just
saying to Shane,

265
00:10:34,111 --> 00:10:37,071
"Maybe we're just wrong,
and we can't even doPong."

266
00:10:37,114 --> 00:10:38,812
It was a bit
nerve-racking,

267
00:10:38,855 --> 00:10:40,422
thinking how far we had to go

268
00:10:40,465 --> 00:10:42,337
if we were going
to really build

269
00:10:42,380 --> 00:10:44,426
a generally
intelligent system.

270
00:10:44,469 --> 00:10:45,732
And it felt like
it was time

271
00:10:45,775 --> 00:10:47,255
to give up and move on.

272
00:10:48,169 --> 00:10:49,387
And then suddenly...

273
00:10:51,389 --> 00:10:53,609
We got our first point.

274
00:10:53,653 --> 00:10:56,612
And then it was like,
"Is this random?"

275
00:10:56,656 --> 00:10:59,180
"No, no, it's really
getting a point now."

276
00:10:59,223 --> 00:11:00,703
It was really exciting
that this thing

277
00:11:00,747 --> 00:11:02,226
that previously
couldn't even figure out

278
00:11:02,270 --> 00:11:03,532
how to move a paddle

279
00:11:03,575 --> 00:11:05,926
had suddenly been able
to totally get it right.

280
00:11:05,969 --> 00:11:07,144
Then it was getting
a few points.

281
00:11:07,188 --> 00:11:08,624
And then it won
its first game.

282
00:11:08,668 --> 00:11:10,974
And then three months later,
no human could beat it.

283
00:11:11,018 --> 00:11:14,238
You hadn't told it the rules,
how to get the score, nothing.

284
00:11:14,282 --> 00:11:16,110
And you just tell it
to maximize the score,

285
00:11:16,153 --> 00:11:17,372
and it goes away and does it.

286
00:11:17,415 --> 00:11:18,678
This is the first time

287
00:11:18,721 --> 00:11:20,549
anyone had done
this end-to-end learning.

288
00:11:20,592 --> 00:11:24,292
"Okay, so we have this working
in quite a general way.

289
00:11:24,335 --> 00:11:25,772
"Now let's try another game."

290
00:11:25,815 --> 00:11:27,512
So then
we triedBreakout.

291
00:11:27,556 --> 00:11:29,123
At the beginning,
after 100 games,

292
00:11:29,166 --> 00:11:30,777
the agent is not very good.

293
00:11:30,820 --> 00:11:32,474
It's missing the ball
most of the time,

294
00:11:32,517 --> 00:11:34,389
but it's starting to get
the hang of the idea

295
00:11:34,432 --> 00:11:35,999
that the bat should go
towards the ball.

296
00:11:36,043 --> 00:11:37,566
Now, after 300 games,

297
00:11:37,609 --> 00:11:40,656
it's about as good as
any human can play this.

298
00:11:40,700 --> 00:11:42,049
We thought,
"Well, that's pretty cool,"

299
00:11:42,092 --> 00:11:44,312
but we left the system playing
for another 200 games,

300
00:11:44,355 --> 00:11:46,053
and it did this amazing thing.

301
00:11:46,096 --> 00:11:47,358
It found the optimal strategy

302
00:11:47,402 --> 00:11:49,404
was to dig a tunnel
around the side

303
00:11:49,447 --> 00:11:51,667
and put the ball
around the back of the wall.

304
00:11:51,711 --> 00:11:53,234
Finally, the agent

305
00:11:53,277 --> 00:11:54,365
is actually achieving

306
00:11:54,409 --> 00:11:55,627
what you thought
it would achieve.

307
00:11:55,671 --> 00:11:57,325
That is a great feeling.
Right?

308
00:11:57,368 --> 00:11:59,283
Like, I mean,
when we do research,

309
00:11:59,327 --> 00:12:00,676
that is the best
we can hope for.

310
00:12:00,720 --> 00:12:03,200
We started generalizing
to 50 games,

311
00:12:03,244 --> 00:12:05,463
and we basically
created a recipe.

312
00:12:05,507 --> 00:12:06,813
We could just take a game

313
00:12:06,856 --> 00:12:08,379
that we have
never seen before.

314
00:12:08,423 --> 00:12:09,903
We would run
the algorithm on that,

315
00:12:09,946 --> 00:12:13,036
and DQN could train itself
from scratch,

316
00:12:13,080 --> 00:12:14,429
achieving human level

317
00:12:14,472 --> 00:12:15,996
or sometimes better
than human level.

318
00:12:16,039 --> 00:12:18,433
We didn't build it
to play any of them.

319
00:12:18,476 --> 00:12:20,522
We could just give it
a bunch of games

320
00:12:20,565 --> 00:12:22,654
and would figure it out
for itself.

321
00:12:22,698 --> 00:12:25,179
And there was something
quite magical in that.

322
00:12:25,222 --> 00:12:26,528
Suddenly you had something

323
00:12:26,571 --> 00:12:27,921
that would respond and learn

324
00:12:27,964 --> 00:12:30,358
whatever situation
it was parachuted into.

325
00:12:30,401 --> 00:12:33,013
And that was like a huge,
huge breakthrough.

326
00:12:33,056 --> 00:12:35,276
It was in many respects

327
00:12:35,319 --> 00:12:36,625
the first example

328
00:12:36,668 --> 00:12:39,062
of any kind of thing
you could call

329
00:12:39,106 --> 00:12:40,542
a general intelligence.

330
00:12:42,109 --> 00:12:43,893
Although we were
a well-funded startup,

331
00:12:43,937 --> 00:12:47,375
holding us back
was not enough compute power.

332
00:12:47,418 --> 00:12:49,203
I realized that
this would accelerate

333
00:12:49,246 --> 00:12:51,466
our time scale
to AGI massively.

334
00:12:51,509 --> 00:12:53,163
I used to see Demis
quite frequently.

335
00:12:53,207 --> 00:12:55,035
We'd have lunch, and he did...

336
00:12:56,210 --> 00:12:58,865
say to me that
he had two companies

337
00:12:58,908 --> 00:13:02,564
that were involved
in buying DeepMind.

338
00:13:02,607 --> 00:13:04,609
And he didn't know
which one to go with.

339
00:13:04,653 --> 00:13:08,526
The issue was,
would any commercial company

340
00:13:08,570 --> 00:13:12,661
appreciate the real importance
of the research?

341
00:13:12,704 --> 00:13:15,838
And give the research time
to come to fruition

342
00:13:15,882 --> 00:13:17,797
and not be breathing down
their necks,

343
00:13:17,840 --> 00:13:21,539
saying, "We want some kind of
commercial benefit from this."

344
00:13:27,284 --> 00:13:32,463
Google has bought DeepMind
for a reported £400,000,000,

345
00:13:32,507 --> 00:13:34,726
making the artificial
intelligence firm

346
00:13:34,770 --> 00:13:37,947
its largest
European acquisition so far.

347
00:13:37,991 --> 00:13:39,644
The company was founded

348
00:13:39,688 --> 00:13:43,344
by 37-year-old entrepreneur
Demis Hassabis.

349
00:13:43,387 --> 00:13:45,563
After the acquisition,
I started mentoring

350
00:13:45,607 --> 00:13:47,261
and spending time with Demis,

351
00:13:47,304 --> 00:13:48,915
and just listening to him.

352
00:13:48,958 --> 00:13:52,396
And this is a person
who fundamentally

353
00:13:52,440 --> 00:13:55,573
is a scientist
and a natural scientist.

354
00:13:55,617 --> 00:13:58,489
He wants science to solve
every problem in the world,

355
00:13:58,533 --> 00:14:00,622
and he believes it can do so.

356
00:14:00,665 --> 00:14:03,625
That's not a normal person
you find in a tech company.

357
00:14:05,496 --> 00:14:07,455
We were able
to not only join Google

358
00:14:07,498 --> 00:14:10,240
but run independently
in London,

359
00:14:10,284 --> 00:14:11,328
build our culture,

360
00:14:11,372 --> 00:14:13,461
which was optimized
for breakthroughs

361
00:14:13,504 --> 00:14:15,419
and not deal with products,

362
00:14:15,463 --> 00:14:17,726
do pure research.

363
00:14:17,769 --> 00:14:19,293
Our investors
didn't want to sell,

364
00:14:19,336 --> 00:14:20,685
but we decided

365
00:14:20,729 --> 00:14:22,731
that this was the best thing
for the mission.

366
00:14:22,774 --> 00:14:24,646
In many senses,
we were underselling

367
00:14:24,689 --> 00:14:26,169
in terms of value
before it more matured,

368
00:14:26,213 --> 00:14:28,128
and you could have sold it
for a lot more money.

369
00:14:28,171 --> 00:14:32,828
And the reason is because
there's no time to waste.

370
00:14:32,872 --> 00:14:35,178
There's so many things
that got to be cracked

371
00:14:35,222 --> 00:14:37,659
while the brain
is still in gear.

372
00:14:37,702 --> 00:14:39,008
You know, I'm still alive.

373
00:14:39,052 --> 00:14:40,880
There's all these things
that gotta be done.

374
00:14:40,923 --> 00:14:42,925
So you haven't got—
I mean, how many...

375
00:14:42,969 --> 00:14:44,492
How many billions
would you trade for

376
00:14:44,535 --> 00:14:45,797
another five years of life,
you know,

377
00:14:45,841 --> 00:14:48,365
to do what you set out to do?

378
00:14:48,409 --> 00:14:49,758
Okay, all of a sudden,

379
00:14:49,801 --> 00:14:52,717
we've got this massive scale
compute available to us.

380
00:14:52,761 --> 00:14:53,936
What can we do with that?

381
00:14:56,591 --> 00:14:59,942
Go is the pinnacle
of board games.

382
00:14:59,986 --> 00:15:04,642
It is the most complex game
ever devised by man.

383
00:15:04,686 --> 00:15:06,688
There are more possible
board configurations

384
00:15:06,731 --> 00:15:09,821
in the game of Go than there
are atoms in the universe.

385
00:15:09,865 --> 00:15:13,390
Go is the holy grail
of artificial intelligence.

386
00:15:13,434 --> 00:15:14,522
For many years,

387
00:15:14,565 --> 00:15:16,002
people have looked
at this game

388
00:15:16,045 --> 00:15:17,917
and they've thought,
"Wow, this is just too hard."

389
00:15:17,960 --> 00:15:20,180
Everything we've ever
tried in AI,

390
00:15:20,223 --> 00:15:22,704
it just falls over when
you try the game of Go.

391
00:15:22,747 --> 00:15:23,966
And so that's why
it feels like

392
00:15:24,010 --> 00:15:26,099
a real litmus test
of progress.

393
00:15:26,142 --> 00:15:28,579
We had just bought DeepMind.

394
00:15:28,623 --> 00:15:30,712
They were working
on reinforcement learning

395
00:15:30,755 --> 00:15:32,975
and they were the world's
experts in games.

396
00:15:33,019 --> 00:15:34,846
And so when
they introduced the idea

397
00:15:34,890 --> 00:15:37,197
that they could beat
the top level Go players

398
00:15:37,240 --> 00:15:40,113
in a game that was thought
to be incomputable,

399
00:15:40,156 --> 00:15:42,724
I thought, "Well,
that's pretty interesting."

400
00:15:42,767 --> 00:15:46,467
Our ultimate next step
is to play the legendary

401
00:15:46,510 --> 00:15:49,209
Lee Sedol
in just over two weeks.

402
00:15:50,514 --> 00:15:52,038
A match like no other

403
00:15:52,081 --> 00:15:54,257
is about to get underway
in South Korea.

404
00:15:54,301 --> 00:15:57,826
Lee Sedol
is getting ready to rumble.

405
00:15:57,869 --> 00:15:59,262
Lee Sedol is probably

406
00:15:59,306 --> 00:16:01,699
one of the greatest players
of the last decade.

407
00:16:01,743 --> 00:16:04,224
I describe him
as the Roger Federer of Go.

408
00:16:05,573 --> 00:16:06,922
He showed up,

409
00:16:06,966 --> 00:16:09,838
and all of a sudden
we have a thousand Koreans

410
00:16:09,881 --> 00:16:13,059
who represent
all of Korean society,

411
00:16:13,102 --> 00:16:14,190
the top Go players.

412
00:16:15,583 --> 00:16:17,802
And then we have Demis.

413
00:16:17,846 --> 00:16:19,848
And the great
engineering team.

414
00:16:20,588 --> 00:16:22,285
He's very famous

415
00:16:22,329 --> 00:16:25,506
for very creative
fighting play.

416
00:16:25,549 --> 00:16:28,770
So this could be
difficult for us.

417
00:16:28,813 --> 00:16:31,991
I figured Lee Sedol
is going to beat these guys,

418
00:16:32,034 --> 00:16:34,297
but they'll make
a good showing.

419
00:16:34,341 --> 00:16:35,733
Good for a startup.

420
00:16:38,040 --> 00:16:39,563
I went over
to the technical group

421
00:16:39,607 --> 00:16:40,912
and they said,

422
00:16:40,956 --> 00:16:42,305
"Let me show you
how our algorithm works."

423
00:16:43,654 --> 00:16:45,091
If you step
through the actual game,

424
00:16:45,134 --> 00:16:47,789
we can see, kind of,
how AlphaGo thinks.

425
00:16:47,832 --> 00:16:50,226
The way we start off
on training AlphaGo

426
00:16:50,270 --> 00:16:52,968
is by showing it 100,000 games

427
00:16:53,012 --> 00:16:54,491
that strong amateurs
have played.

428
00:16:54,535 --> 00:16:55,710
And we first initially

429
00:16:55,753 --> 00:16:58,887
get AlphaGo to mimic
the human player,

430
00:16:58,930 --> 00:17:00,802
and then through
reinforcement learning,

431
00:17:00,845 --> 00:17:02,369
it plays against
different versions of itself

432
00:17:02,412 --> 00:17:05,720
many millions of times
and learns from its errors.

433
00:17:05,763 --> 00:17:07,591
Hmm, this is interesting.

434
00:17:07,635 --> 00:17:08,679
All right, folks,

435
00:17:08,723 --> 00:17:10,594
you're going to see
history made.

436
00:17:12,770 --> 00:17:14,772
So the game starts.

437
00:17:14,816 --> 00:17:15,991
He's really concentrating.

438
00:17:16,035 --> 00:17:17,645
If you really look at the...

439
00:17:20,909 --> 00:17:25,348
That's a very surprising move.

440
00:17:25,392 --> 00:17:27,916
I think we're
seeing an original move here.

441
00:17:34,401 --> 00:17:35,793
Yeah, that's an exciting move.

442
00:17:36,185 --> 00:17:37,230
I like...

443
00:17:37,273 --> 00:17:38,579
Professional commentators

444
00:17:38,622 --> 00:17:40,102
almost unanimously said

445
00:17:40,146 --> 00:17:43,323
that not a single human player
would have chosen move 37.

446
00:17:43,366 --> 00:17:45,803
So I actually had a poke
around in AlphaGo

447
00:17:45,847 --> 00:17:47,414
to see what AlphaGo thought.

448
00:17:47,457 --> 00:17:50,330
And AlphaGo actually agreed
with that assessment.

449
00:17:50,373 --> 00:17:53,811
AlphaGo said there was a one
in 10,000 probability

450
00:17:53,855 --> 00:17:57,772
that move 37 would have been
played by a human player.

451
00:18:08,826 --> 00:18:10,176
The game of Go
has been studied

452
00:18:10,219 --> 00:18:11,568
for thousands of years.

453
00:18:11,612 --> 00:18:15,137
And AlphaGo discovered
something completely new.

454
00:18:16,878 --> 00:18:19,707
He resigned.
Lee Sedol has just resigned.

455
00:18:19,750 --> 00:18:21,187
He's beaten.

456
00:18:22,710 --> 00:18:24,668
The battle
between man versus machine,

457
00:18:24,712 --> 00:18:26,235
a computer just came out
the victor.

458
00:18:26,279 --> 00:18:28,237
Google
put its DeepMind team

459
00:18:28,281 --> 00:18:29,673
to the test against

460
00:18:29,717 --> 00:18:32,023
one of the brightest minds
in the world and won.

461
00:18:32,067 --> 00:18:33,721
That's when we realized

462
00:18:33,764 --> 00:18:35,244
the DeepMind people knew
what they were doing

463
00:18:35,288 --> 00:18:37,551
and to pay attention
to reinforcement learning

464
00:18:37,594 --> 00:18:38,900
as they have invented it.

465
00:18:40,075 --> 00:18:41,816
Based on that experience,

466
00:18:41,859 --> 00:18:44,819
AlphaGo got better
and better and better.

467
00:18:44,862 --> 00:18:45,950
And they had a little chart

468
00:18:45,994 --> 00:18:47,517
of how much better
they were getting.

469
00:18:47,561 --> 00:18:49,302
And I said,
"When does this stop?"

470
00:18:50,085 --> 00:18:50,999
And Demis said,

471
00:18:51,042 --> 00:18:52,696
"When we beat the Chinese guy,

472
00:18:52,740 --> 00:18:55,743
"the top-rated player
in the world."

473
00:18:56,961 --> 00:18:59,225
Ke Jie versus AlphaGo.

474
00:19:03,620 --> 00:19:04,795
And I think we will see

475
00:19:04,839 --> 00:19:06,145
AlphaGo pushing through there.

476
00:19:06,188 --> 00:19:08,190
AlphaGo is ahead quite a bit.

477
00:19:08,234 --> 00:19:11,454
About halfway
through the first game,

478
00:19:11,498 --> 00:19:14,675
the best player in the world
was not doing so well.

479
00:19:14,718 --> 00:19:17,808
What can black do here?

480
00:19:19,114 --> 00:19:21,247
Looks difficult.

481
00:19:21,290 --> 00:19:23,336
And at a critical moment...

482
00:19:32,997 --> 00:19:35,913
the Chinese government
ordered the feed cut off.

483
00:19:38,307 --> 00:19:41,745
It was at that moment
we were telling the world

484
00:19:41,789 --> 00:19:44,966
that something new
had arrived on earth.

485
00:19:47,621 --> 00:19:48,970
In the 1950s

486
00:19:49,013 --> 00:19:51,929
when Russia'sSputnik
satellite was launched,

487
00:19:53,279 --> 00:19:55,194
it changed
the course of history.

488
00:19:55,237 --> 00:19:57,544
It is a challenge
that America must meet

489
00:19:57,587 --> 00:19:59,633
to survive in the Space Age.

490
00:19:59,676 --> 00:20:02,375
This has been
called theSputnik moment.

491
00:20:02,418 --> 00:20:06,335
The Sputnikmoment created
a massive reaction in the US

492
00:20:06,379 --> 00:20:10,034
in terms of funding
for science and engineering,

493
00:20:10,078 --> 00:20:12,254
and particularly
of space technology.

494
00:20:12,298 --> 00:20:15,823
For China,
AlphaGo was the wakeup call,

495
00:20:15,866 --> 00:20:17,128
the Sputnikmoment.

496
00:20:17,172 --> 00:20:19,870
It launched an AI space race.

497
00:20:21,220 --> 00:20:23,047
We had this
huge idea that worked,

498
00:20:23,091 --> 00:20:26,355
and now the whole world knows.

499
00:20:26,399 --> 00:20:28,879
It's always easier
to land on the moon

500
00:20:28,923 --> 00:20:30,838
if someone's already
landed there.

501
00:20:32,056 --> 00:20:34,450
It is going to matter
who builds AI,

502
00:20:34,494 --> 00:20:36,626
and how it gets built.

503
00:20:36,670 --> 00:20:38,324
I always feel that pressure.

504
00:20:42,066 --> 00:20:43,764
There's been
a big chain of events

505
00:20:43,807 --> 00:20:46,680
that followed on from all
of the excitement of AlphaGo.

506
00:20:46,723 --> 00:20:48,072
When we played
against Lee Sedol,

507
00:20:48,116 --> 00:20:49,248
we actually had a system

508
00:20:49,291 --> 00:20:50,684
that had been trained
on human data,

509
00:20:50,727 --> 00:20:52,251
on all of the millions
of games

510
00:20:52,294 --> 00:20:55,036
that have been played
by human experts.

511
00:20:55,079 --> 00:20:56,994
We eventually found
a new algorithm,

512
00:20:57,038 --> 00:20:59,170
a much more elegant approach
to the whole system,

513
00:20:59,214 --> 00:21:01,172
which actually stripped out
all of the human knowledge

514
00:21:01,216 --> 00:21:03,697
and just started
completely from scratch.

515
00:21:03,740 --> 00:21:06,700
And that became a project
which we called AlphaZero.

516
00:21:06,743 --> 00:21:09,529
Zero, meaning having zero
human knowledge in the loop.

517
00:21:11,879 --> 00:21:13,054
Instead of learning
from human data,

518
00:21:13,097 --> 00:21:15,796
it learned from its own games.

519
00:21:15,839 --> 00:21:17,841
So it actually
became its own teacher.

520
00:21:21,280 --> 00:21:23,499
AlphaZero is an experiment

521
00:21:23,543 --> 00:21:26,720
in how little knowledge
can we put into these systems

522
00:21:26,763 --> 00:21:28,243
and how quickly
and how efficiently

523
00:21:28,287 --> 00:21:29,723
can they learn?

524
00:21:29,766 --> 00:21:32,552
But the other thing is AlphaZero
doesn't have any rules.

525
00:21:32,595 --> 00:21:33,553
It learns through experience.

526
00:21:36,120 --> 00:21:38,862
The next stage
was to make it more general,

527
00:21:38,906 --> 00:21:40,995
so that it could play
any two-player game.

528
00:21:41,038 --> 00:21:42,344
Things like chess,

529
00:21:42,388 --> 00:21:44,085
and in fact,
any kind of two-player

530
00:21:44,128 --> 00:21:45,391
perfect information game.

531
00:21:45,434 --> 00:21:46,653
It's going really well.

532
00:21:46,696 --> 00:21:47,828
It's going
really, really well.

533
00:21:47,871 --> 00:21:50,091
- Oh, wow.
- It's going down, like fast.

534
00:21:50,134 --> 00:21:53,050
AlphaGo used
to take a few months to train,

535
00:21:53,094 --> 00:21:55,662
but AlphaZero could start
in the morning

536
00:21:55,705 --> 00:21:57,794
playing completely randomly

537
00:21:57,838 --> 00:22:01,015
and then by tea
be at superhuman level.

538
00:22:01,058 --> 00:22:03,365
And by dinner it will be
the strongest chess entity

539
00:22:03,409 --> 00:22:04,758
there's ever been.

540
00:22:04,801 --> 00:22:06,629
- Amazing, it's amazing.
- Yeah.

541
00:22:06,673 --> 00:22:09,371
It's discovered its own
attacking style, you know,

542
00:22:09,415 --> 00:22:11,417
to take on the current
level of defense.

543
00:22:11,460 --> 00:22:12,766
I mean, I never
in my wildest dreams...

544
00:22:12,809 --> 00:22:14,942
I agree. Actually, I was not
expecting that either.

545
00:22:14,985 --> 00:22:16,422
And it's fun for me.

546
00:22:16,465 --> 00:22:18,598
I mean, it's inspired me
to get back into chess again,

547
00:22:18,641 --> 00:22:20,164
because it's cool to see

548
00:22:20,208 --> 00:22:22,384
that there's even more depth
than we thought in chess.

549
00:22:31,611 --> 00:22:34,440
I actually got
into AI through games.

550
00:22:35,658 --> 00:22:37,660
Initially, it was board games.

551
00:22:37,704 --> 00:22:40,054
I was thinking,
"How is my brain doing this?"

552
00:22:40,097 --> 00:22:41,969
Like, what is it doing?

553
00:22:43,362 --> 00:22:47,148
I was very aware of that
from a very young age.

554
00:22:47,191 --> 00:22:50,064
So I've always been thinking
about thinking.

555
00:22:50,107 --> 00:22:52,762
The British
and American chess champions

556
00:22:52,806 --> 00:22:55,025
meet to begin
a series of matches.

557
00:22:55,069 --> 00:22:56,592
Playing alongside them
are the cream

558
00:22:56,636 --> 00:22:59,073
of Britain and America's
youngest players.

559
00:22:59,116 --> 00:23:01,467
Demis Hassabis
is representing Britain.

560
00:23:06,210 --> 00:23:07,864
When Demis was four,

561
00:23:07,908 --> 00:23:11,259
he first showed
an aptitude for chess.

562
00:23:12,739 --> 00:23:14,044
By the time he was six,

563
00:23:14,088 --> 00:23:18,048
he became London
under-eight champion.

564
00:23:18,092 --> 00:23:19,485
My parents
were very interesting

565
00:23:19,528 --> 00:23:20,834
and unusual, actually.

566
00:23:20,877 --> 00:23:23,750
I'd probably describe them
as quite bohemian.

567
00:23:23,793 --> 00:23:25,491
My father
was a singer-songwriter

568
00:23:25,534 --> 00:23:26,709
when he was younger,

569
00:23:26,753 --> 00:23:28,363
and Bob Dylan was his hero.

570
00:23:38,068 --> 00:23:39,287
Yeah, yeah.

571
00:23:41,637 --> 00:23:44,031
What is it
that you like about this game?

572
00:23:45,075 --> 00:23:47,121
It's just a good
thinking game.

573
00:23:49,253 --> 00:23:51,038
At the time,
I was the second-highest rated

574
00:23:51,081 --> 00:23:52,692
chess player in the world
for my age.

575
00:23:52,735 --> 00:23:54,345
But although I was on track

576
00:23:54,389 --> 00:23:55,956
to be a professional
chess player,

577
00:23:55,999 --> 00:23:57,523
I thought that was what
I was going to do.

578
00:23:57,566 --> 00:23:59,220
No matter how much
I loved the game,

579
00:23:59,263 --> 00:24:01,222
it was incredibly stressful.

580
00:24:01,265 --> 00:24:03,354
Definitely was not fun
and games for me.

581
00:24:03,398 --> 00:24:05,226
My parents used to, you know,

582
00:24:05,269 --> 00:24:06,923
get very upset
when I lost the game

583
00:24:06,967 --> 00:24:10,405
and angry
if I forgot something.

584
00:24:10,449 --> 00:24:12,407
And because it was quite high
stakes for them, you know,

585
00:24:12,451 --> 00:24:14,061
it cost a lot of money
to go to these tournaments.

586
00:24:14,104 --> 00:24:15,715
And my parents
didn't have much money.

587
00:24:18,413 --> 00:24:19,719
My parents thought, you know,

588
00:24:19,762 --> 00:24:22,069
"If you interested
in being a chess professional,

589
00:24:22,112 --> 00:24:25,376
"this is really important.
It's like your exams."

590
00:24:27,291 --> 00:24:30,077
I remember
I was about 12-years-old

591
00:24:30,120 --> 00:24:32,079
and I was at this
international chess tournament

592
00:24:32,122 --> 00:24:34,255
in Liechtenstein
up in the mountains.

593
00:24:43,307 --> 00:24:45,571
And we were in this
huge church hall

594
00:24:47,181 --> 00:24:48,399
with, you know,

595
00:24:48,443 --> 00:24:50,184
hundreds of international
chess players.

596
00:24:52,403 --> 00:24:56,016
And I was playing
the ex-Danish champion.

597
00:24:56,059 --> 00:24:58,801
He must have been
in his 30s, probably.

598
00:25:00,411 --> 00:25:02,979
In those days,
there was a long time limit.

599
00:25:03,023 --> 00:25:05,068
The games could
literally last all day.

600
00:25:08,332 --> 00:25:10,813
We were into our tenth hour.

601
00:25:20,257 --> 00:25:23,086
And we were in this
incredibly unusual ending.

602
00:25:23,130 --> 00:25:24,566
I think it should be a draw.

603
00:25:26,307 --> 00:25:28,657
But he kept on trying
to win for hours.

604
00:25:35,359 --> 00:25:38,319
Finally, he tried
one last cheap trick.

605
00:25:42,541 --> 00:25:44,455
All I had to do
was give away my queen.

606
00:25:44,499 --> 00:25:45,674
Then it would be stalemate.

607
00:25:47,023 --> 00:25:48,634
But I was so tired,

608
00:25:48,677 --> 00:25:50,157
I thought it was inevitable
I was going to be checkmated.

609
00:25:52,289 --> 00:25:53,508
And so I resigned.

610
00:25:57,294 --> 00:25:59,558
He jumped up.
Just started laughing.

611
00:26:01,777 --> 00:26:02,909
And he went,

612
00:26:02,952 --> 00:26:04,171
"Why have you resigned?
It's a draw."

613
00:26:04,214 --> 00:26:05,346
And he immediately,
with a flourish,

614
00:26:05,389 --> 00:26:06,652
sort of showed me
the drawing move.

615
00:26:09,045 --> 00:26:12,396
I felt so sick to my stomach.

616
00:26:12,440 --> 00:26:14,137
It made me think of
the rest of that tournament.

617
00:26:14,181 --> 00:26:16,662
Like, are we wasting
our minds?

618
00:26:16,705 --> 00:26:19,708
Is this the best use
of all this brain power?

619
00:26:19,752 --> 00:26:22,363
Everybody's, collectively,
in that building?

620
00:26:22,406 --> 00:26:24,060
If you could somehow plug in

621
00:26:24,104 --> 00:26:27,542
those 300 brains
into a system,

622
00:26:27,586 --> 00:26:29,022
you might be able
to solve cancer

623
00:26:29,065 --> 00:26:30,501
with that level
of brain power.

624
00:26:31,633 --> 00:26:33,504
This intuitive feeling
came over me

625
00:26:33,548 --> 00:26:35,071
that although I love chess,

626
00:26:35,115 --> 00:26:38,335
this is not the right thing
to spend my whole life on.

627
00:26:51,479 --> 00:26:53,089
Demis and myself,

628
00:26:53,133 --> 00:26:56,092
our plan was always
to fill DeepMind

629
00:26:56,136 --> 00:26:57,224
with some of the most

630
00:26:57,267 --> 00:26:59,226
brilliant scientists
in the world.

631
00:26:59,269 --> 00:27:01,054
So we had the human brains

632
00:27:01,097 --> 00:27:04,927
necessary to create
an AGI system.

633
00:27:04,971 --> 00:27:09,279
By definition, the "G"
in AGI is about generality.

634
00:27:09,323 --> 00:27:13,109
What I imagine is being able
to talk to an agent,

635
00:27:13,153 --> 00:27:15,198
the agent can talk back,

636
00:27:15,242 --> 00:27:18,898
and the agent is able to solve
novel problems

637
00:27:18,941 --> 00:27:20,595
that it hasn't seen before.

638
00:27:20,639 --> 00:27:22,728
That's a really key part
of human intelligence,

639
00:27:22,771 --> 00:27:24,468
and it's that
cognitive breadth

640
00:27:24,512 --> 00:27:27,733
and flexibility
that's incredible.

641
00:27:27,776 --> 00:27:29,473
The only natural
general intelligence

642
00:27:29,517 --> 00:27:30,910
we know of as humans,

643
00:27:30,953 --> 00:27:33,434
we obviously learn a lot
from our environment.

644
00:27:33,477 --> 00:27:35,871
So we think that
simulated environments

645
00:27:35,915 --> 00:27:38,874
are one of the ways
to create an AGI.

646
00:27:40,528 --> 00:27:42,356
The very early humans

647
00:27:42,399 --> 00:27:44,314
were having to solve
logic problems.

648
00:27:44,358 --> 00:27:46,752
They were having to solve
navigation, memory,

649
00:27:46,795 --> 00:27:48,971
and we evolved
in that environment.

650
00:27:50,364 --> 00:27:52,496
If we can create
a virtual recreation

651
00:27:52,540 --> 00:27:54,629
of that kind of environment,

652
00:27:54,673 --> 00:27:56,326
that's the perfect
testing ground

653
00:27:56,370 --> 00:27:57,458
and training ground

654
00:27:57,501 --> 00:27:59,329
for everything
we do at DeepMind.

655
00:28:04,247 --> 00:28:05,727
What they were doing here

656
00:28:05,771 --> 00:28:09,252
was creating environments
for childlike beings,

657
00:28:09,296 --> 00:28:11,690
the agents to exist
within and play.

658
00:28:12,429 --> 00:28:13,648
That just sounded like

659
00:28:13,692 --> 00:28:16,782
the most interesting thing
in all the world.

660
00:28:16,825 --> 00:28:19,132
A child
learns by tearing things up

661
00:28:19,175 --> 00:28:20,655
and then throwing food around

662
00:28:20,699 --> 00:28:23,353
and getting a response
from mommy or daddy.

663
00:28:23,397 --> 00:28:25,616
This seems like an important
idea to incorporate

664
00:28:25,660 --> 00:28:27,880
in the way you train an agent.

665
00:28:27,923 --> 00:28:30,970
The humanoid
is supposed to stand up.

666
00:28:31,013 --> 00:28:32,972
As his center
of gravity rises,

667
00:28:33,015 --> 00:28:34,408
it gets more points.

668
00:28:37,454 --> 00:28:38,804
You have a reward

669
00:28:38,847 --> 00:28:40,849
and the agent
learns from the reward,

670
00:28:40,893 --> 00:28:43,373
like, you do something well,
you get a positive reward.

671
00:28:43,417 --> 00:28:47,508
You do something bad,
you get a negative reward.

672
00:28:47,551 --> 00:28:49,379
It looks like it's standing.

673
00:28:50,859 --> 00:28:52,165
It's still a bit drunk.

674
00:28:52,208 --> 00:28:53,644
It likes to walk backwards.

675
00:28:53,688 --> 00:28:55,342
Yeah.

676
00:28:55,385 --> 00:28:57,518
The whole algorithm
is trying to optimize

677
00:28:57,561 --> 00:28:59,694
for receiving as much rewards
as possible,

678
00:28:59,738 --> 00:29:02,305
and it's found that
walking backwards,

679
00:29:02,349 --> 00:29:05,352
it's good enough
to get very good scores.

680
00:29:07,746 --> 00:29:09,573
When we learn to navigate,

681
00:29:09,617 --> 00:29:11,271
when we learn to get around
in our world,

682
00:29:11,314 --> 00:29:13,490
we don't start with maps.

683
00:29:13,534 --> 00:29:16,319
We just start
with our own exploration,

684
00:29:16,363 --> 00:29:18,017
adventuring off
across the park,

685
00:29:18,060 --> 00:29:21,890
without our parents
by our side,

686
00:29:21,934 --> 00:29:24,327
or finding our way home
from school when we're young.

687
00:29:26,939 --> 00:29:28,723
A few of us
came up with this idea

688
00:29:28,767 --> 00:29:31,987
that if we had an environment
where a simulated robot

689
00:29:32,031 --> 00:29:33,728
just had to run forward,

690
00:29:33,772 --> 00:29:36,296
we could put all sorts of
obstacles in its way

691
00:29:36,339 --> 00:29:38,211
and see if it could manage
to navigate

692
00:29:38,254 --> 00:29:40,474
different types of terrain.

693
00:29:40,517 --> 00:29:43,042
The idea would be like
a parkour challenge.

694
00:29:46,393 --> 00:29:48,743
It's not graceful,

695
00:29:48,787 --> 00:29:51,833
but was never trained to hold
a glass whilst it was running

696
00:29:51,877 --> 00:29:53,052
and not spill water.

697
00:29:54,183 --> 00:29:55,706
You set this objective
that says,

698
00:29:55,750 --> 00:29:58,231
"Just move forward,
forward velocity,

699
00:29:58,274 --> 00:30:00,581
"and you'll get
a reward for that."

700
00:30:00,624 --> 00:30:02,670
And the learning algorithm
figures out

701
00:30:02,713 --> 00:30:05,412
how to move
this complex set of joints.

702
00:30:06,152 --> 00:30:07,370
That's the power of

703
00:30:07,414 --> 00:30:10,069
reward-based
reinforcement learning.

704
00:30:10,112 --> 00:30:12,767
Our goal
is to try and build agents

705
00:30:12,811 --> 00:30:15,770
which, we drop them in,
they know nothing,

706
00:30:15,814 --> 00:30:18,381
they get to play around in
whatever problem you give them

707
00:30:18,425 --> 00:30:22,342
and eventually figure out how
to solve it for themselves.

708
00:30:22,385 --> 00:30:24,823
Now we want something
which can do that

709
00:30:24,866 --> 00:30:27,390
in as many different types
of problems as possible.

710
00:30:29,262 --> 00:30:32,918
A human needs diverse skills
to interact with the world.

711
00:30:32,961 --> 00:30:35,050
How to deal
with complex images,

712
00:30:35,094 --> 00:30:37,879
how to manipulate
thousands of things at once,

713
00:30:37,923 --> 00:30:40,447
how to deal
with missing information.

714
00:30:40,490 --> 00:30:42,014
We think all of these things
together

715
00:30:42,057 --> 00:30:45,278
are represented
by this game calledStarCraft.

716
00:30:45,321 --> 00:30:47,280
All it's being trained
to do is,

717
00:30:47,323 --> 00:30:50,457
given this situation,
this screen,

718
00:30:50,500 --> 00:30:51,850
what would a human do?

719
00:30:51,893 --> 00:30:55,244
We took inspiration from
large language models

720
00:30:55,288 --> 00:30:57,638
where you simply train
a model

721
00:30:57,681 --> 00:30:59,553
to predict the next word,

722
00:31:03,731 --> 00:31:05,472
which is exactly the same as

723
00:31:05,515 --> 00:31:07,735
predict the next
StarCraft move.

724
00:31:07,778 --> 00:31:08,954
Unlike chess or Go,

725
00:31:08,997 --> 00:31:11,217
where players take turns
to make moves,

726
00:31:11,260 --> 00:31:14,133
inStarCraft there's a
continuous flow of decisions.

727
00:31:15,134 --> 00:31:16,483
On top of that,

728
00:31:16,526 --> 00:31:18,659
you can't even see
what the opponent is doing.

729
00:31:18,702 --> 00:31:20,879
There is no longer
a clear definition

730
00:31:20,922 --> 00:31:22,315
of what it means
to play the best way.

731
00:31:22,358 --> 00:31:23,925
It depends on
what your opponent does.

732
00:31:23,969 --> 00:31:25,492
This is the way
that we'll get to

733
00:31:25,535 --> 00:31:27,494
a much more fluid,

734
00:31:27,537 --> 00:31:31,672
more natural, faster,
more reactive agent.

735
00:31:31,715 --> 00:31:33,021
This is a huge challenge

736
00:31:33,065 --> 00:31:35,415
and let's see how far
we can push.

737
00:31:35,458 --> 00:31:36,459
Oh!

738
00:31:36,503 --> 00:31:38,200
Holy monkey!

739
00:31:38,244 --> 00:31:40,376
I'm a pretty
low-level amateur.

740
00:31:40,420 --> 00:31:42,857
I'm okay, but I'm
a pretty low-level amateur.

741
00:31:42,901 --> 00:31:45,991
These agents have
a long ways to go.

742
00:31:46,034 --> 00:31:48,515
We couldn't
beat someone of Tim's level.

743
00:31:48,558 --> 00:31:50,430
You know, that was
a little bit alarming.

744
00:31:50,473 --> 00:31:51,997
At that point, it felt like

745
00:31:52,040 --> 00:31:53,520
it was going to be, like,
a really big long challenge,

746
00:31:53,563 --> 00:31:55,000
maybe a couple of years.

747
00:31:58,177 --> 00:32:01,832
Dani is the best
DeepMind StarCraft 2 player.

748
00:32:01,876 --> 00:32:05,097
I've been playing the agent
every day for a few weeks now.

749
00:32:07,186 --> 00:32:08,883
I could feel that the agent

750
00:32:08,927 --> 00:32:11,103
was getting better
really fast.

751
00:32:13,409 --> 00:32:15,020
Wow, we beat Danny.
That, for me,

752
00:32:15,063 --> 00:32:16,935
was already
like a huge achievement.

753
00:32:18,284 --> 00:32:19,372
The next step is

754
00:32:19,415 --> 00:32:21,722
we're going to book in
a pro to play.

755
00:32:42,003 --> 00:32:44,049
It feels a bit unfair.
All you guys against me.

756
00:32:45,615 --> 00:32:46,965
We're way
ahead of what I thought

757
00:32:47,008 --> 00:32:49,402
we would do, given where
we were two months ago.

758
00:32:49,445 --> 00:32:50,794
Just trying to digest it all,
actually.

759
00:32:50,838 --> 00:32:52,753
But it's very, very cool.

760
00:32:52,796 --> 00:32:54,146
Now we're in
a position where

761
00:32:54,189 --> 00:32:56,061
we can finally share
the work that we've done

762
00:32:56,104 --> 00:32:57,192
with the public.

763
00:32:57,236 --> 00:32:58,585
This is a big step.

764
00:32:58,628 --> 00:33:00,674
We are really putting
ourselves on the line here.

765
00:33:00,717 --> 00:33:02,589
- Take it away. Cheers.
- Thank you.

766
00:33:02,632 --> 00:33:04,460
We're going to be live
from London.

767
00:33:04,504 --> 00:33:05,722
It's happening.

768
00:33:08,638 --> 00:33:10,597
Welcome to London.

769
00:33:10,640 --> 00:33:13,252
We are going to have
a live exhibition match,

770
00:33:13,295 --> 00:33:15,341
MaNa against AlphaStar.

771
00:33:18,344 --> 00:33:19,998
At this point now,

772
00:33:20,041 --> 00:33:23,827
AlphaStar, 10 and 0
against professional gamers.

773
00:33:23,871 --> 00:33:25,873
Any thoughts
before we get into this game?

774
00:33:25,916 --> 00:33:27,483
I just want to see
a good game, yeah.

775
00:33:27,527 --> 00:33:28,963
I want to see a good game.

776
00:33:29,007 --> 00:33:30,486
Absolutely,
good game. We're all excited.

777
00:33:30,530 --> 00:33:33,011
All right. Let's
see what MaNa can pull off.

778
00:33:34,969 --> 00:33:36,405
AlphaStar is definitely

779
00:33:36,449 --> 00:33:38,364
dominating the pace
of this game.

780
00:33:41,149 --> 00:33:44,152
Wow. AlphaStar
is playing so smartly.

781
00:33:46,850 --> 00:33:48,461
This really looks like
I'm watching

782
00:33:48,504 --> 00:33:49,940
a professional human gamer

783
00:33:49,984 --> 00:33:51,290
from the AlphaStar
point of view.

784
00:33:57,252 --> 00:34:01,952
I hadn't really seen
a pro playStarCraft up close,

785
00:34:01,996 --> 00:34:03,563
and the 800 clicks per minute.

786
00:34:03,606 --> 00:34:06,000
I don't understand how anyone
can even click 800 times,

787
00:34:06,044 --> 00:34:09,482
let alone doing
800 useful clicks.

788
00:34:09,525 --> 00:34:11,092
Oh, another good hit.

789
00:34:11,136 --> 00:34:13,094
AlphaStar is just

790
00:34:13,138 --> 00:34:14,617
completely relentless.

791
00:34:14,661 --> 00:34:16,141
We need to be careful

792
00:34:16,184 --> 00:34:19,361
because many of us grew up
as gamers and are gamers.

793
00:34:19,405 --> 00:34:21,363
And so to us,
it's very natural

794
00:34:21,407 --> 00:34:23,800
to view games
as what they are,

795
00:34:23,844 --> 00:34:26,977
which is pure vehicles
for fun,

796
00:34:27,021 --> 00:34:29,719
and not to see
that more militaristic side

797
00:34:29,763 --> 00:34:32,853
that the public might see
if they looked at this.

798
00:34:32,896 --> 00:34:37,553
You can't look at gunpowder
and only make a firecracker.

799
00:34:37,597 --> 00:34:41,209
All technologies inherently
point into certain directions.

800
00:34:43,124 --> 00:34:44,691
I'm very worried about

801
00:34:44,734 --> 00:34:46,606
the certain ways in which AI

802
00:34:46,649 --> 00:34:49,652
will be used
for military purposes.

803
00:34:51,306 --> 00:34:55,049
And that makes it even clearer
how important it is

804
00:34:55,093 --> 00:34:58,357
for our societies
to be in control

805
00:34:58,400 --> 00:35:01,055
of these new technologies.

806
00:35:01,099 --> 00:35:05,190
The potential for abuse
from AI will be significant.

807
00:35:05,233 --> 00:35:08,758
Wars that occur faster
than humans can comprehend

808
00:35:08,802 --> 00:35:11,065
and more powerful
surveillance.

809
00:35:12,458 --> 00:35:15,765
How do you keep power forever

810
00:35:15,809 --> 00:35:19,421
over something that's
much more powerful than you?

811
00:35:43,053 --> 00:35:45,752
Technologies can be used
to do terrible things.

812
00:35:47,493 --> 00:35:50,452
And technology can be used
to do wonderful things

813
00:35:50,496 --> 00:35:52,150
and solve
all kinds of problems.

814
00:35:53,586 --> 00:35:54,978
When DeepMind
was acquired by Google...

815
00:35:55,022 --> 00:35:56,589
- Yeah.
- ... you got Google to promise

816
00:35:56,632 --> 00:35:58,025
that technology you developed
won't be used by the military

817
00:35:58,068 --> 00:35:59,418
- for surveillance.
- Right.

818
00:35:59,461 --> 00:36:00,593
- Yes.
- Tell us about that.

819
00:36:00,636 --> 00:36:03,161
I think technology
is neutral in itself,

820
00:36:03,204 --> 00:36:05,598
um, but how, you know,
we as a society

821
00:36:05,641 --> 00:36:07,382
or humans and companies
and other things,

822
00:36:07,426 --> 00:36:09,515
other entities and governments
decide to use it

823
00:36:09,558 --> 00:36:12,561
is what determines whether
things become good or bad.

824
00:36:12,605 --> 00:36:16,261
You know, I personally think
having autonomous weaponry

825
00:36:16,304 --> 00:36:17,479
is just a very bad idea.

826
00:36:19,177 --> 00:36:21,266
AlphaStar is playing

827
00:36:21,309 --> 00:36:24,094
an extremely intelligent game
right now.

828
00:36:24,138 --> 00:36:27,359
There is an element to
what's being created

829
00:36:27,402 --> 00:36:28,882
at DeepMind in London

830
00:36:28,925 --> 00:36:34,148
that does seem like
the Manhattan Project.

831
00:36:34,192 --> 00:36:37,586
There's a relationship between
Robert Oppenheimer

832
00:36:37,630 --> 00:36:39,675
and Demis Hassabis

833
00:36:39,719 --> 00:36:44,202
in which they're unleashing
a new force upon humanity.

834
00:36:44,245 --> 00:36:46,204
MaNa is fighting back, though.

835
00:36:46,247 --> 00:36:48,162
Oh, man!

836
00:36:48,206 --> 00:36:50,208
I think that Oppenheimer

837
00:36:50,251 --> 00:36:52,471
and some of the other leaders
of that project got caught up

838
00:36:52,514 --> 00:36:54,908
in the excitement
of building the technology

839
00:36:54,951 --> 00:36:56,170
and seeing if it was possible.

840
00:36:56,214 --> 00:36:58,520
Where is AlphaStar?

841
00:36:58,564 --> 00:36:59,782
Where is AlphaStar?

842
00:36:59,826 --> 00:37:01,958
I don't see AlphaStar's units
anywhere.

843
00:37:02,002 --> 00:37:03,525
They did not think
carefully enough

844
00:37:03,569 --> 00:37:07,312
about the morals of what
they were doing early enough.

845
00:37:07,355 --> 00:37:08,965
What we should do
as scientists

846
00:37:09,009 --> 00:37:11,011
with powerful new technologies

847
00:37:11,054 --> 00:37:13,883
is try and understand it in
controlled conditions first.

848
00:37:14,928 --> 00:37:16,799
And that is that.

849
00:37:16,843 --> 00:37:19,411
MaNa has defeated AlphaStar.

850
00:37:29,551 --> 00:37:31,336
I mean, my honest feeling is
that I think it is

851
00:37:31,379 --> 00:37:33,207
a fair representation
of where we are.

852
00:37:33,251 --> 00:37:35,949
And I think that part feels...
feels okay.

853
00:37:35,992 --> 00:37:37,429
- I'm very happy for you.
- I'm happy.

854
00:37:37,472 --> 00:37:38,865
So well... well done.

855
00:37:38,908 --> 00:37:40,867
My view is that the approach
to building technology

856
00:37:40,910 --> 00:37:43,348
which is embodied by
move fast and break things,

857
00:37:43,391 --> 00:37:46,220
is exactly what
we should not be doing,

858
00:37:46,264 --> 00:37:47,961
because you can't afford
to break things

859
00:37:48,004 --> 00:37:49,049
and then fix them afterwards.

860
00:37:49,092 --> 00:37:50,398
- Cheers.
- Thank you so much.

861
00:37:50,442 --> 00:37:52,008
Yeah, get... get some rest.
You did really well.

862
00:37:52,052 --> 00:37:53,923
- Cheers, yeah?
- Thank you for having us.

863
00:38:04,238 --> 00:38:05,500
When I was eight,

864
00:38:05,544 --> 00:38:06,849
I bought my first computer

865
00:38:06,893 --> 00:38:09,548
with the winnings
from a chess tournament.

866
00:38:09,591 --> 00:38:11,158
I sort of had this intuition

867
00:38:11,201 --> 00:38:13,726
that computers
are this magical device

868
00:38:13,769 --> 00:38:15,902
that can extend
the power of the mind.

869
00:38:15,945 --> 00:38:17,382
I had a couple
of school friends,

870
00:38:17,425 --> 00:38:19,166
and we used to have
a hacking club,

871
00:38:19,209 --> 00:38:21,908
writing code, making games.

872
00:38:26,260 --> 00:38:27,827
And then over
the summer holidays,

873
00:38:27,870 --> 00:38:29,219
I'd spend the whole day

874
00:38:29,263 --> 00:38:31,526
flicking through
games magazines.

875
00:38:31,570 --> 00:38:33,441
And one day I noticed
there was a competition

876
00:38:33,485 --> 00:38:35,878
to write an original version
of Space Invaders.

877
00:38:35,922 --> 00:38:39,621
And the winner won a job
at Bullfrog.

878
00:38:39,665 --> 00:38:42,320
Bullfrog at the time was the
best game development house

879
00:38:42,363 --> 00:38:43,756
in all of Europe.

880
00:38:43,799 --> 00:38:45,279
You know, I really wanted
to work at this place

881
00:38:45,323 --> 00:38:48,587
and see how they build games.

882
00:38:48,630 --> 00:38:50,415
Bullfrog,
based here in Guildford,

883
00:38:50,458 --> 00:38:52,286
began with a big idea.

884
00:38:52,330 --> 00:38:54,680
That idea turned into the game
Populous,

885
00:38:54,723 --> 00:38:56,551
which became
a global bestseller.

886
00:38:56,595 --> 00:38:59,859
In the '90s, there was
no recruitment agencies.

887
00:38:59,902 --> 00:39:02,122
You couldn't go out and say,
you know,

888
00:39:02,165 --> 00:39:04,951
"Come and work
in the games industry."

889
00:39:04,994 --> 00:39:08,171
It was still not even
considered an industry.

890
00:39:08,215 --> 00:39:11,218
So we came up with the idea
to have a competition

891
00:39:11,261 --> 00:39:13,655
and we got
a lot of applicants.

892
00:39:14,700 --> 00:39:17,616
And one of those was Demis's.

893
00:39:17,659 --> 00:39:20,706
I can still remember clearly

894
00:39:20,749 --> 00:39:23,970
the day that Demis came in.

895
00:39:24,013 --> 00:39:27,147
He walked in the door,
he looked about 12.

896
00:39:28,670 --> 00:39:30,019
I thought, "Oh, my God,

897
00:39:30,063 --> 00:39:31,586
"what the hell are we going
to do with this guy?"

898
00:39:31,630 --> 00:39:32,979
I applied to Cambridge.

899
00:39:33,022 --> 00:39:35,373
I got in but they said
I was way too young.

900
00:39:35,416 --> 00:39:37,853
So...
So I needed to take a year off

901
00:39:37,897 --> 00:39:39,899
so I'd be at least 17
before I got there.

902
00:39:39,942 --> 00:39:42,771
And that's when I decided
to spend that entire gap year

903
00:39:42,815 --> 00:39:44,469
working at Bullfrog.

904
00:39:44,512 --> 00:39:46,166
They couldn't even
legally employ me,

905
00:39:46,209 --> 00:39:48,298
so I ended up being paid
in brown paper envelopes.

906
00:39:50,823 --> 00:39:54,304
I got a feeling of being
really at the cutting edge

907
00:39:54,348 --> 00:39:58,047
and how much fun that was
to invent things every day.

908
00:39:58,091 --> 00:40:00,572
And then you know,
a few months later,

909
00:40:00,615 --> 00:40:03,662
maybe everyone... a million
people will be playing it.

910
00:40:03,705 --> 00:40:06,665
In those days
computer games had to evolve.

911
00:40:06,708 --> 00:40:08,536
There had to be new genres

912
00:40:08,580 --> 00:40:11,757
which were more
than just shooting things.

913
00:40:11,800 --> 00:40:14,063
Wouldn't it be amazing
to have a game

914
00:40:14,107 --> 00:40:18,807
where you design and build
your own theme park?

915
00:40:22,594 --> 00:40:25,945
Demis and I started to talk
aboutTheme Park.

916
00:40:25,988 --> 00:40:28,904
It allows the player
to build a world

917
00:40:28,948 --> 00:40:31,864
and see the consequences
of your choices

918
00:40:31,907 --> 00:40:34,127
that you've made
in that world.

919
00:40:34,170 --> 00:40:36,085
A human player
set out the layout

920
00:40:36,129 --> 00:40:38,566
of the theme park and designed
the roller coaster

921
00:40:38,610 --> 00:40:41,351
and set the prices
in the chip shop.

922
00:40:41,395 --> 00:40:43,615
What I was working on was
the behaviors of the people.

923
00:40:43,658 --> 00:40:45,138
They were autonomous

924
00:40:45,181 --> 00:40:47,314
and that was the AI
in this case.

925
00:40:47,357 --> 00:40:48,881
So what I was trying to do
was mimic

926
00:40:48,924 --> 00:40:51,013
interesting human behavior

927
00:40:51,057 --> 00:40:52,319
so that the simulation
would be

928
00:40:52,362 --> 00:40:54,582
more interesting
to interact with.

929
00:40:54,626 --> 00:40:56,541
Demis worked
on ridiculous things,

930
00:40:56,584 --> 00:40:59,413
like you could place down
these shops

931
00:40:59,457 --> 00:41:03,591
and if you put a shop too near
a very dangerous ride,

932
00:41:03,635 --> 00:41:05,375
then people on the ride
would throw up

933
00:41:05,419 --> 00:41:08,030
because they'd just eaten.

934
00:41:08,074 --> 00:41:09,641
And then that would make
other people throw up

935
00:41:09,684 --> 00:41:12,121
when they saw the throwing-up
on the floor,

936
00:41:12,165 --> 00:41:14,559
so you then had to have
lots of sweepers

937
00:41:14,602 --> 00:41:17,823
to quickly sweep it up
before the people saw it.

938
00:41:17,866 --> 00:41:19,520
That's the cool thing
about it.

939
00:41:19,564 --> 00:41:22,784
You as the player tinker with
it and then it reacts to you.

940
00:41:22,828 --> 00:41:25,874
All those nuanced
simulation things he did

941
00:41:25,918 --> 00:41:28,094
and that was an invention

942
00:41:28,137 --> 00:41:31,227
which never really
existed before.

943
00:41:31,271 --> 00:41:34,230
It was
unbelievably successful.

944
00:41:34,274 --> 00:41:35,710
Theme Park actually turned out

945
00:41:35,754 --> 00:41:37,190
to be a top ten title

946
00:41:37,233 --> 00:41:39,932
and that was the first time
we were starting to see

947
00:41:39,975 --> 00:41:43,022
how AI could make
a difference.

948
00:41:46,155 --> 00:41:47,592
We were doing
some Christmas shopping

949
00:41:47,635 --> 00:41:51,247
and were waiting for the taxi
to take us home.

950
00:41:51,291 --> 00:41:54,947
I have this very clear memory
of Demis talking about AI

951
00:41:54,990 --> 00:41:56,209
in a very different way,

952
00:41:56,252 --> 00:41:58,428
in a way that we didn't
commonly talk about.

953
00:41:58,472 --> 00:42:02,345
This idea of AI being useful
for other things

954
00:42:02,389 --> 00:42:04,086
other than entertainment.

955
00:42:04,130 --> 00:42:07,437
So being useful for, um,
helping the world

956
00:42:07,481 --> 00:42:10,310
and the potential of AI
to change the world.

957
00:42:10,353 --> 00:42:13,226
I just said to Demis,
"What is it you want to do?"

958
00:42:13,269 --> 00:42:14,532
And he said to me,

959
00:42:14,575 --> 00:42:16,795
"I want to be the person
that solves AI."

960
00:42:22,670 --> 00:42:25,760
Peter offered me £1 million

961
00:42:25,804 --> 00:42:27,675
to not go to university.

962
00:42:30,199 --> 00:42:32,593
But I had a plan
from the beginning.

963
00:42:32,637 --> 00:42:35,814
And my plan was always
to go to Cambridge.

964
00:42:35,857 --> 00:42:36,902
I think a lot of
my schoolfriends

965
00:42:36,945 --> 00:42:38,033
thought I was mad.

966
00:42:38,077 --> 00:42:39,252
Why would you not...

967
00:42:39,295 --> 00:42:40,688
I mean, £1 million,
that's a lot of money.

968
00:42:40,732 --> 00:42:43,517
In the '90s,
that is a lot of money, right?

969
00:42:43,561 --> 00:42:46,346
For a...
For a poor 17-year-old kid.

970
00:42:46,389 --> 00:42:50,219
He's like this little seed
that's going to burst through,

971
00:42:50,263 --> 00:42:53,658
and he's not going to be able
to do that at Bullfrog.

972
00:42:56,443 --> 00:42:59,098
I had to drop him off
at the train station

973
00:42:59,141 --> 00:43:02,580
and I can still see
that picture

974
00:43:02,623 --> 00:43:07,019
of this little elven character
disappear down that tunnel.

975
00:43:07,062 --> 00:43:09,804
That was an incredibly
sad moment.

976
00:43:13,242 --> 00:43:14,635
I had this romantic ideal

977
00:43:14,679 --> 00:43:16,942
of what Cambridge
would be like,

978
00:43:16,985 --> 00:43:18,639
1,000 years of history,

979
00:43:18,683 --> 00:43:21,033
walking the same streets
that Turing,

980
00:43:21,076 --> 00:43:23,601
Newton and Crick had walked.

981
00:43:23,644 --> 00:43:26,647
I wanted to explore
the edge of the universe.

982
00:43:29,084 --> 00:43:30,346
When I got to Cambridge,

983
00:43:30,390 --> 00:43:32,653
I'd basically been working
my whole life.

984
00:43:33,741 --> 00:43:35,090
Every single summer,

985
00:43:35,134 --> 00:43:37,136
I was either playing chess
professionally,

986
00:43:37,179 --> 00:43:39,704
or I was working,
doing an internship.

987
00:43:39,747 --> 00:43:43,708
So I was, like, "Right,
I am gonna have fun now

988
00:43:43,751 --> 00:43:46,711
"and explore what it means
to be a normal teenager."

989
00:43:50,236 --> 00:43:52,238
Come on! Go, boy, go!

990
00:43:52,281 --> 00:43:54,022
It was work hard
and play hard.

991
00:43:55,850 --> 00:43:57,025
I first met Demis

992
00:43:57,069 --> 00:43:59,201
because we both attended
Queens' College.

993
00:44:00,115 --> 00:44:01,203
Our group of friends,

994
00:44:01,247 --> 00:44:03,205
we'd often drink beer
in the bar,

995
00:44:03,249 --> 00:44:04,946
play table football.

996
00:44:04,990 --> 00:44:07,340
In the bar,
I used to play speed chess,

997
00:44:07,383 --> 00:44:09,255
pieces flying off the board,

998
00:44:09,298 --> 00:44:11,083
you know, the whole game
in one minute.

999
00:44:11,126 --> 00:44:12,301
Demis sat down opposite me.

1000
00:44:12,345 --> 00:44:13,563
And I looked at him
and I thought,

1001
00:44:13,607 --> 00:44:15,217
"I remember you
from when we were kids."

1002
00:44:15,261 --> 00:44:17,176
I had actually been
in the same chess tournament

1003
00:44:17,219 --> 00:44:18,786
as Dave in Ipswich,

1004
00:44:18,830 --> 00:44:20,440
where I used to go and try
and raid his local chess club

1005
00:44:20,483 --> 00:44:22,703
to win a bit of prize money.

1006
00:44:22,747 --> 00:44:24,618
We were studying
computer science.

1007
00:44:24,662 --> 00:44:26,794
Some people,
who at the age of 17

1008
00:44:26,838 --> 00:44:28,404
would have come in and made
sure to tell everybody

1009
00:44:28,448 --> 00:44:29,492
everything about themselves.

1010
00:44:29,536 --> 00:44:30,972
"Hey, I worked at Bullfrog

1011
00:44:31,016 --> 00:44:33,018
"and built the world's
most successful video game."

1012
00:44:33,061 --> 00:44:34,715
But he wasn't like that
at all.

1013
00:44:34,759 --> 00:44:36,412
At Cambridge,
Demis and myself

1014
00:44:36,456 --> 00:44:38,414
both had an interest
in computational neuroscience

1015
00:44:38,458 --> 00:44:40,242
and trying to understand
how computers and brains

1016
00:44:40,286 --> 00:44:42,636
intertwined
and linked together.

1017
00:44:42,680 --> 00:44:44,290
Both David and Demis

1018
00:44:44,333 --> 00:44:46,422
came to me for supervisions.

1019
00:44:46,466 --> 00:44:49,774
It happens just by coincidence
that the year 1997,

1020
00:44:49,817 --> 00:44:51,645
their third and final year
at Cambridge,

1021
00:44:51,689 --> 00:44:55,301
was also the year when
the first chess grandmaster

1022
00:44:55,344 --> 00:44:56,781
was beaten by
a computer program.

1023
00:44:58,304 --> 00:45:00,088
Round one today
of a chess match

1024
00:45:00,132 --> 00:45:03,701
between the ranking
world champion Garry Kasparov

1025
00:45:03,744 --> 00:45:06,007
and an opponent named
Deep Blue

1026
00:45:06,051 --> 00:45:10,490
to test to see if the human
brain can outwit a machine.

1027
00:45:10,533 --> 00:45:11,621
I remember the drama

1028
00:45:11,665 --> 00:45:13,798
of Kasparov
losing the last match.

1029
00:45:13,841 --> 00:45:15,234
Whoa!

1030
00:45:15,277 --> 00:45:17,192
Kasparov has resigned!

1031
00:45:17,236 --> 00:45:19,586
When Deep Blue
beat Garry Kasparov,

1032
00:45:19,629 --> 00:45:21,457
that was a real
watershed event.

1033
00:45:21,501 --> 00:45:23,155
My main memory of it was

1034
00:45:23,198 --> 00:45:25,331
I wasn't that impressed
with Deep Blue.

1035
00:45:25,374 --> 00:45:27,246
I was more impressed
with Kasparov's mind.

1036
00:45:27,289 --> 00:45:29,509
That he could play chess
to this level,

1037
00:45:29,552 --> 00:45:31,641
where he could compete
on an equal footing

1038
00:45:31,685 --> 00:45:33,252
with the brute of a machine,

1039
00:45:33,295 --> 00:45:35,167
but of course, Kasparov can do

1040
00:45:35,210 --> 00:45:36,951
everything else humans can do,
too.

1041
00:45:36,995 --> 00:45:38,257
It was a huge achievement.

1042
00:45:38,300 --> 00:45:39,388
But the truth
of the matter was,

1043
00:45:39,432 --> 00:45:40,868
Deep Blue
could only play chess.

1044
00:45:42,435 --> 00:45:44,524
What we would regard
as intelligence

1045
00:45:44,567 --> 00:45:46,874
was missing from that system.

1046
00:45:46,918 --> 00:45:49,834
This idea of generality
and also learning.

1047
00:45:53,751 --> 00:45:55,404
Cambridge was amazing,
because of course, you know,

1048
00:45:55,448 --> 00:45:56,666
you're mixing with people

1049
00:45:56,710 --> 00:45:58,233
who are studying
many different subjects.

1050
00:45:58,277 --> 00:46:01,410
There were scientists,
philosophers, artists...

1051
00:46:01,454 --> 00:46:04,457
geologists,
biologists, ecologists.

1052
00:46:04,500 --> 00:46:07,416
You know, everybody is talking
about everything all the time.

1053
00:46:07,460 --> 00:46:10,768
I was obsessed with
the protein folding problem.

1054
00:46:10,811 --> 00:46:13,248
Tim Stevens used
to talk obsessively,

1055
00:46:13,292 --> 00:46:15,381
almost like religiously
about this problem,

1056
00:46:15,424 --> 00:46:17,165
protein folding problem.

1057
00:46:17,209 --> 00:46:18,863
Proteins are, you know,

1058
00:46:18,906 --> 00:46:22,083
one of the most beautiful and
elegant things about biology.

1059
00:46:22,127 --> 00:46:24,738
They are the machines of life.

1060
00:46:24,782 --> 00:46:27,001
They build everything,
they control everything,

1061
00:46:27,045 --> 00:46:29,569
they're why biology works.

1062
00:46:29,612 --> 00:46:32,659
Proteins are made from strings
of amino acids

1063
00:46:32,702 --> 00:46:37,055
that fold up to create
a protein structure.

1064
00:46:37,098 --> 00:46:39,884
If we can predict
the structure of proteins

1065
00:46:39,927 --> 00:46:43,104
from just their amino acid
sequences,

1066
00:46:43,148 --> 00:46:46,107
then a new protein
to cure cancer

1067
00:46:46,151 --> 00:46:49,415
or break down plastic
to help the environment

1068
00:46:49,458 --> 00:46:50,808
is definitely something

1069
00:46:50,851 --> 00:46:52,940
that you could begin
to think about.

1070
00:46:53,941 --> 00:46:55,029
I kind of thought,

1071
00:46:55,073 --> 00:46:58,293
"Well, is a human being
clever enough

1072
00:46:58,337 --> 00:46:59,991
"to actually fold a protein?"

1073
00:47:00,034 --> 00:47:02,080
We can't work it out.

1074
00:47:02,123 --> 00:47:04,082
Since the 1960s,

1075
00:47:04,125 --> 00:47:05,953
we thought that in principle,

1076
00:47:05,997 --> 00:47:08,913
if I know what the amino acid
sequence of a protein is,

1077
00:47:08,956 --> 00:47:11,437
I should be able to compute
what the structure's like.

1078
00:47:11,480 --> 00:47:13,700
So, if you could
just press a button,

1079
00:47:13,743 --> 00:47:16,311
and they'd all come
popping out, that would be...

1080
00:47:16,355 --> 00:47:18,009
that would have some impact.

1081
00:47:20,272 --> 00:47:21,577
It stuck in my mind.

1082
00:47:21,621 --> 00:47:23,405
"Oh, this is
a very interesting problem."

1083
00:47:23,449 --> 00:47:26,844
And it felt to me
like it would be solvable.

1084
00:47:26,887 --> 00:47:29,934
But I thought
it would need AI to do it.

1085
00:47:31,500 --> 00:47:34,025
If we could just solve
protein folding,

1086
00:47:34,068 --> 00:47:35,635
it could change the world.

1087
00:47:50,868 --> 00:47:52,826
Ever since
I was a student at Cambridge,

1088
00:47:54,001 --> 00:47:55,611
I've never
stopped thinking about

1089
00:47:55,655 --> 00:47:57,135
the protein folding problem.

1090
00:47:59,789 --> 00:48:02,792
If you were
to solve protein folding,

1091
00:48:02,836 --> 00:48:05,665
then the potential
to help solve problems like

1092
00:48:05,708 --> 00:48:09,669
Alzheimer's, dementia
and drug discovery is huge.

1093
00:48:09,712 --> 00:48:11,671
Solving disease is probably

1094
00:48:11,714 --> 00:48:13,325
the most major impact
we could have.

1095
00:48:15,457 --> 00:48:16,676
Thousands of very smart people

1096
00:48:16,719 --> 00:48:18,765
have tried
to solve protein folding.

1097
00:48:18,808 --> 00:48:20,985
I just think now
is the right time

1098
00:48:21,028 --> 00:48:22,508
for AI to crack it.

1099
00:48:26,555 --> 00:48:28,383
We needed
a reasonable way

1100
00:48:28,427 --> 00:48:29,515
to apply machine learning

1101
00:48:29,558 --> 00:48:30,516
to the protein folding
problem.

1102
00:48:32,431 --> 00:48:35,173
We came across
this Foldit game.

1103
00:48:35,216 --> 00:48:38,785
The goal is to move around
this 3D model of a protein

1104
00:48:38,828 --> 00:48:41,701
and you get a score
every time you move it.

1105
00:48:41,744 --> 00:48:43,050
The more accurate
you make these structures,

1106
00:48:43,094 --> 00:48:45,531
the more useful
they will be to biologists.

1107
00:48:46,227 --> 00:48:47,489
I spent a few days

1108
00:48:47,533 --> 00:48:48,838
just kind of seeing
how well we could do.

1109
00:48:50,666 --> 00:48:52,407
We did reasonably well.

1110
00:48:52,451 --> 00:48:53,843
But even if you were

1111
00:48:53,887 --> 00:48:55,280
the world's
best Foldit player,

1112
00:48:55,323 --> 00:48:57,499
you wouldn't
solve protein folding.

1113
00:48:57,543 --> 00:48:59,501
That's why we had
to move beyond the game.

1114
00:48:59,545 --> 00:49:00,720
Games
are always just

1115
00:49:00,763 --> 00:49:03,723
the proving ground
for our algorithms.

1116
00:49:03,766 --> 00:49:07,727
The ultimate goal was not just
to crack Go and StarCraft.

1117
00:49:07,770 --> 00:49:10,034
It was to crack
real-world challenges.

1118
00:49:16,127 --> 00:49:18,520
I remember
hearing this rumor

1119
00:49:18,564 --> 00:49:21,175
that Demis was
getting into proteins.

1120
00:49:21,219 --> 00:49:23,743
I talked to some people
at DeepMind and I would ask,

1121
00:49:23,786 --> 00:49:25,049
"So are you doing
protein folding?"

1122
00:49:25,092 --> 00:49:26,920
And they would
artfully change the subject.

1123
00:49:26,964 --> 00:49:30,097
And when that happened twice,
I pretty much figured it out.

1124
00:49:30,141 --> 00:49:32,839
So I thought
I should submit a resume.

1125
00:49:32,882 --> 00:49:35,668
All right, everyone,
welcome to DeepMind.

1126
00:49:35,711 --> 00:49:37,626
I know some of you,
this may be your first week,

1127
00:49:37,670 --> 00:49:39,193
but I hope you all set...

1128
00:49:39,237 --> 00:49:40,890
The really appealing
part for me about the job

1129
00:49:40,934 --> 00:49:42,675
was this, like,
sense of connection

1130
00:49:42,718 --> 00:49:44,503
to the larger purpose.

1131
00:49:44,546 --> 00:49:45,591
If we can crack

1132
00:49:45,634 --> 00:49:48,028
some fundamental problems
in science,

1133
00:49:48,072 --> 00:49:49,160
many other people

1134
00:49:49,203 --> 00:49:50,900
and other companies
and labs and so on

1135
00:49:50,944 --> 00:49:52,467
could build
on top of our work.

1136
00:49:52,511 --> 00:49:53,816
This is your chance now

1137
00:49:53,860 --> 00:49:55,818
to add your chapter
to this story.

1138
00:49:55,862 --> 00:49:57,298
When I arrived,

1139
00:49:57,342 --> 00:49:59,605
I was definitely
quite a bit nervous.

1140
00:49:59,648 --> 00:50:00,736
I'm still trying to keep...

1141
00:50:00,780 --> 00:50:02,912
I haven't taken
any biology courses.

1142
00:50:02,956 --> 00:50:05,393
We haven't spent
years of our lives

1143
00:50:05,437 --> 00:50:07,917
looking at these structures
and understanding them.

1144
00:50:07,961 --> 00:50:09,963
We are just going off the data

1145
00:50:10,007 --> 00:50:11,269
and our machine learning
models.

1146
00:50:12,705 --> 00:50:13,836
In machine learning,

1147
00:50:13,880 --> 00:50:15,795
you train a network
like flashcards.

1148
00:50:15,838 --> 00:50:18,798
Here's the question.
Here's the answer.

1149
00:50:18,841 --> 00:50:20,756
Here's the question.
Here's the answer.

1150
00:50:20,800 --> 00:50:22,323
But in protein folding,

1151
00:50:22,367 --> 00:50:25,457
we're not doing the kind
of standard task at DeepMind

1152
00:50:25,500 --> 00:50:28,155
where you have unlimited data.

1153
00:50:28,199 --> 00:50:30,766
Your job is to get better
at chess or Go

1154
00:50:30,810 --> 00:50:32,986
and you can play
as many games of chess or Go

1155
00:50:33,030 --> 00:50:34,814
as your computers will allow.

1156
00:50:35,510 --> 00:50:36,772
With proteins,

1157
00:50:36,816 --> 00:50:39,732
we're sitting on
a very thick size of data

1158
00:50:39,775 --> 00:50:41,995
that's been determined
by a half century

1159
00:50:42,039 --> 00:50:46,478
of time-consuming experimental
methods in laboratories.

1160
00:50:46,521 --> 00:50:49,698
These painstaking methods
can take months or years

1161
00:50:49,742 --> 00:50:52,353
to determine
a single protein structure,

1162
00:50:52,397 --> 00:50:55,791
and sometimes, a structure
can never be determined.

1163
00:50:57,315 --> 00:51:00,274
That's why we're working
with such small datasets

1164
00:51:00,318 --> 00:51:02,233
to train our algorithms.

1165
00:51:02,276 --> 00:51:04,365
When DeepMind
started to explore

1166
00:51:04,409 --> 00:51:05,975
the folding problem,

1167
00:51:06,019 --> 00:51:07,977
they were talking to us about
which datasets they were using

1168
00:51:08,021 --> 00:51:09,892
and what would be
the possibilities

1169
00:51:09,936 --> 00:51:11,633
if they did
solve this problem.

1170
00:51:12,373 --> 00:51:13,940
Many people have tried,

1171
00:51:13,983 --> 00:51:16,638
and yet no one on the planet
has solved protein folding.

1172
00:51:16,682 --> 00:51:18,205
I did think to myself,

1173
00:51:18,249 --> 00:51:19,946
"Well, you know, good luck."

1174
00:51:19,989 --> 00:51:22,731
If we can solve
the protein folding problem,

1175
00:51:22,775 --> 00:51:25,691
it would have an incredible
kind of medical relevance.

1176
00:51:25,734 --> 00:51:27,736
This is the cycle of science.

1177
00:51:27,780 --> 00:51:30,043
You do a huge amount
of exploration,

1178
00:51:30,087 --> 00:51:31,914
and then you go
into exploitation mode,

1179
00:51:31,958 --> 00:51:33,568
and you focus and you see

1180
00:51:33,612 --> 00:51:35,353
how good
are those ideas, really?

1181
00:51:35,396 --> 00:51:36,528
And there's nothing better

1182
00:51:36,571 --> 00:51:38,095
than external competition
for that.

1183
00:51:39,835 --> 00:51:43,056
So we decided
to enter CASP competition.

1184
00:51:43,100 --> 00:51:47,234
CASP, we started
to try and speed up

1185
00:51:47,278 --> 00:51:49,802
the solution to
the protein folding problem.

1186
00:51:49,845 --> 00:51:52,065
CASP is when we say,

1187
00:51:52,109 --> 00:51:54,372
"Look, DeepMind
is doing protein folding,

1188
00:51:54,415 --> 00:51:55,590
"this is how good we are,

1189
00:51:55,634 --> 00:51:57,592
"and maybe it's better
than everybody else.

1190
00:51:57,636 --> 00:51:58,680
"Maybe it isn't."

1191
00:51:58,724 --> 00:52:00,073
CASP is a bit like

1192
00:52:00,117 --> 00:52:02,075
the Olympic Games
of protein folding.

1193
00:52:03,642 --> 00:52:06,035
CASP is
a community-wide assessment

1194
00:52:06,079 --> 00:52:08,125
that's held every two years.

1195
00:52:09,561 --> 00:52:10,866
Teams are given

1196
00:52:10,910 --> 00:52:14,261
the amino acid sequences
of about 100 proteins,

1197
00:52:14,305 --> 00:52:17,525
and then they try
to solve this folding problem

1198
00:52:17,569 --> 00:52:20,833
using computational methods.

1199
00:52:20,876 --> 00:52:23,401
These proteins have
already been determined

1200
00:52:23,444 --> 00:52:25,968
by experiments
in a laboratory,

1201
00:52:26,012 --> 00:52:29,276
but have not yet
been revealed publicly.

1202
00:52:29,320 --> 00:52:30,756
And these known structures

1203
00:52:30,799 --> 00:52:33,280
represent the gold standard
against which

1204
00:52:33,324 --> 00:52:36,936
all the computational
predictions will be compared.

1205
00:52:37,937 --> 00:52:39,330
We've got a score

1206
00:52:39,373 --> 00:52:42,159
that measures the accuracy
of the predictions.

1207
00:52:42,202 --> 00:52:44,726
And you would expect
a score of over 90

1208
00:52:44,770 --> 00:52:47,425
to be a solution to
the protein folding problem.

1209
00:52:48,556 --> 00:52:50,036
Welcome, everyone,

1210
00:52:50,079 --> 00:52:52,038
to our first, uh, semifinals
in the winners' bracket.

1211
00:52:52,081 --> 00:52:54,954
Nick and John
versus Demis and Frank.

1212
00:52:54,997 --> 00:52:57,217
Please join us, come around.
This will be an intense match.

1213
00:52:57,261 --> 00:52:59,306
When I learned that Demis was

1214
00:52:59,350 --> 00:53:02,048
going to tackle
the protein folding issue,

1215
00:53:02,091 --> 00:53:04,790
um, I wasn't at all surprised.

1216
00:53:04,833 --> 00:53:06,792
It's very typical of Demis.

1217
00:53:06,835 --> 00:53:08,924
You know,
he loves competition.

1218
00:53:08,968 --> 00:53:10,143
And that's the end

1219
00:53:10,187 --> 00:53:12,972
of the first game, 10-7.

1220
00:53:13,015 --> 00:53:14,103
The aim for CASP would be

1221
00:53:14,147 --> 00:53:15,975
to not just
win the competition,

1222
00:53:16,018 --> 00:53:19,892
but sort of, um,
retire the need for it.

1223
00:53:19,935 --> 00:53:23,417
So, 20 targets total
have been released by CASP.

1224
00:53:23,461 --> 00:53:24,810
We were thinking maybe

1225
00:53:24,853 --> 00:53:26,899
throw in the standard
kind of machine learning

1226
00:53:26,942 --> 00:53:28,814
and see how far
that could take us.

1227
00:53:28,857 --> 00:53:30,729
Instead of having a couple
of days on an experiment,

1228
00:53:30,772 --> 00:53:33,558
we can turn around
five experiments a day.

1229
00:53:33,601 --> 00:53:35,342
Great. Well done, everyone.

1230
00:53:36,778 --> 00:53:38,693
Can you show me the real one
instead of ours?

1231
00:53:38,737 --> 00:53:39,781
The true answer is

1232
00:53:39,825 --> 00:53:42,044
supposed to look
something like that.

1233
00:53:42,088 --> 00:53:45,047
It's a lot more
cylindrical than I thought.

1234
00:53:45,091 --> 00:53:47,398
The results
were not very good.

1235
00:53:47,441 --> 00:53:48,703
Okay.

1236
00:53:48,747 --> 00:53:49,835
We throw
all the obvious ideas to it

1237
00:53:49,878 --> 00:53:51,793
and the problem laughs at you.

1238
00:53:52,881 --> 00:53:54,361
This makes no sense.

1239
00:53:54,405 --> 00:53:56,015
We thought
we could just throw

1240
00:53:56,058 --> 00:53:58,322
some of our best algorithms
at the problem.

1241
00:53:59,366 --> 00:54:00,976
We were slightly naive.

1242
00:54:01,020 --> 00:54:02,282
We should be learning this,

1243
00:54:02,326 --> 00:54:04,284
you know,
in the blink of an eye.

1244
00:54:05,372 --> 00:54:06,982
The thing
I'm worried about is,

1245
00:54:07,026 --> 00:54:08,375
we take the field from

1246
00:54:08,419 --> 00:54:10,899
really bad answers
to moderately bad answers.

1247
00:54:10,943 --> 00:54:13,946
I feel like we need
some sort of new technology

1248
00:54:13,989 --> 00:54:15,164
for moving around
these things.

1249
00:54:20,431 --> 00:54:22,215
With only
a week left of CASP,

1250
00:54:22,259 --> 00:54:24,348
it's now a sprint
to get it deployed.

1251
00:54:26,654 --> 00:54:28,090
You've done your best.

1252
00:54:28,134 --> 00:54:29,875
Then there's
nothing more you can do

1253
00:54:29,918 --> 00:54:32,399
but wait for CASP
to deliver the results.

1254
00:54:52,593 --> 00:54:53,725
This famous thing of Einstein,

1255
00:54:53,768 --> 00:54:55,030
the last couple of years
of his life,

1256
00:54:55,074 --> 00:54:57,381
when he was here,
he overlapped with Kurt Godel

1257
00:54:57,424 --> 00:54:59,774
and he said one of the reasons
he still comes in to work

1258
00:54:59,818 --> 00:55:01,646
is so that
he gets to walk home

1259
00:55:01,689 --> 00:55:03,517
and discuss things with Godel.

1260
00:55:03,561 --> 00:55:05,911
It's a pretty big compliment
for Kurt Godel,

1261
00:55:05,954 --> 00:55:07,478
shows you how amazing he was.

1262
00:55:09,131 --> 00:55:10,568
The Institute
for Advanced Study

1263
00:55:10,611 --> 00:55:12,918
was formed in 1933.

1264
00:55:12,961 --> 00:55:14,223
In the early years,

1265
00:55:14,267 --> 00:55:16,487
the intense scientific
atmosphere attracted

1266
00:55:16,530 --> 00:55:19,359
some of the most brilliant
mathematicians and physicists

1267
00:55:19,403 --> 00:55:22,536
ever concentrated
in a single place and time.

1268
00:55:22,580 --> 00:55:24,582
The founding
principle of this place,

1269
00:55:24,625 --> 00:55:28,020
it's the idea of unfettered
intellectual pursuits,

1270
00:55:28,063 --> 00:55:30,152
even if you don't know
what you're exploring.

1271
00:55:30,196 --> 00:55:32,154
Will result
in some cool things,

1272
00:55:32,198 --> 00:55:34,896
and sometimes that then
ends up being useful,

1273
00:55:34,940 --> 00:55:36,376
which, of course,

1274
00:55:36,420 --> 00:55:37,986
is partially what I've been
trying to do at DeepMind.

1275
00:55:38,030 --> 00:55:39,988
How many big breakthroughs
do you think are required

1276
00:55:40,032 --> 00:55:41,686
to get all the way to AGI?

1277
00:55:41,729 --> 00:55:43,078
And, you know,
I estimate maybe

1278
00:55:43,122 --> 00:55:44,341
there's about
a dozen of those.

1279
00:55:44,384 --> 00:55:46,125
You know, I hope
it's within my lifetime.

1280
00:55:46,168 --> 00:55:47,605
- Yes, okay.
- But then,

1281
00:55:47,648 --> 00:55:49,171
all scientists
hope that, right?

1282
00:55:49,215 --> 00:55:51,130
Demis has
many accolades.

1283
00:55:51,173 --> 00:55:54,002
He was elected Fellow to
the Royal Society last year.

1284
00:55:54,046 --> 00:55:55,961
He is also a Fellow
of Royal Society of Arts.

1285
00:55:56,004 --> 00:55:57,528
A big hand for Demis Hassabis.

1286
00:56:04,230 --> 00:56:05,927
My dream
has always been to try

1287
00:56:05,971 --> 00:56:08,103
and make
AI-assisted science possible.

1288
00:56:08,147 --> 00:56:09,235
And what I think is

1289
00:56:09,278 --> 00:56:11,150
our most exciting project,
last year,

1290
00:56:11,193 --> 00:56:13,152
which is our work
in protein folding.

1291
00:56:13,195 --> 00:56:15,328
Uh, and we call this system
AlphaFold.

1292
00:56:15,372 --> 00:56:18,331
We entered it into CASP
and our system, uh,

1293
00:56:18,375 --> 00:56:20,507
was the most accurate,
uh, predicting structures

1294
00:56:20,551 --> 00:56:24,946
for 25 out of the 43 proteins
in the hardest category.

1295
00:56:24,990 --> 00:56:26,208
So we're state of the art,

1296
00:56:26,252 --> 00:56:27,514
but we still...
I have to make... Be clear,

1297
00:56:27,558 --> 00:56:28,559
we're still a long way from

1298
00:56:28,602 --> 00:56:30,474
solving the protein
folding problem.

1299
00:56:30,517 --> 00:56:31,866
We're working hard
on this, though,

1300
00:56:31,910 --> 00:56:33,738
and we're exploring
many other techniques.

1301
00:56:49,188 --> 00:56:50,232
Let's get started.

1302
00:56:50,276 --> 00:56:53,148
So kind of
a rapid debrief,

1303
00:56:53,192 --> 00:56:55,455
these are
our final rankings for CASP.

1304
00:56:56,500 --> 00:56:57,544
We beat the second team

1305
00:56:57,588 --> 00:57:00,155
in this competition
by nearly 50%,

1306
00:57:00,199 --> 00:57:01,592
but we've still got
a long way to go

1307
00:57:01,635 --> 00:57:04,333
before we've solved
the protein folding problem

1308
00:57:04,377 --> 00:57:07,032
in a sense that
a biologist could use it.

1309
00:57:07,075 --> 00:57:08,990
It is area of concern.

1310
00:57:11,602 --> 00:57:14,213
The quality
of predictions varied

1311
00:57:14,256 --> 00:57:16,737
and they were no more useful
than the previous methods.

1312
00:57:16,781 --> 00:57:19,914
AlphaFold didn't
produce good enough data

1313
00:57:19,958 --> 00:57:22,526
for it to be useful
in a practical way

1314
00:57:22,569 --> 00:57:24,005
to, say, somebody like me

1315
00:57:24,049 --> 00:57:28,227
investigating
my own biological problems.

1316
00:57:28,270 --> 00:57:30,316
That was kind of
a humbling moment

1317
00:57:30,359 --> 00:57:32,753
'cause we thought we'd worked
very hard and succeeded.

1318
00:57:32,797 --> 00:57:34,886
And what we'd found is
we were the best in the world

1319
00:57:34,929 --> 00:57:36,453
at a problem
the world's not good at.

1320
00:57:37,671 --> 00:57:38,933
We knew we sucked.

1321
00:57:40,457 --> 00:57:42,328
It doesn't help
if you have the tallest ladder

1322
00:57:42,371 --> 00:57:44,635
when you're going to the moon.

1323
00:57:44,678 --> 00:57:47,115
The opinion of quite
a few people on the team,

1324
00:57:47,159 --> 00:57:51,468
that this is sort of
a fool's errand in some ways.

1325
00:57:51,511 --> 00:57:54,079
And I might have been wrong
with protein folding.

1326
00:57:54,122 --> 00:57:55,559
Maybe it's too hard still

1327
00:57:55,602 --> 00:57:58,431
for where we're at
generally with AI.

1328
00:57:58,475 --> 00:58:01,173
If you want to do
biological research,

1329
00:58:01,216 --> 00:58:03,044
you have to be
prepared to fail

1330
00:58:03,088 --> 00:58:06,570
because biology
is very complicated.

1331
00:58:06,613 --> 00:58:09,790
I've run a laboratory
for nearly 50 years,

1332
00:58:09,834 --> 00:58:11,096
and half my time,

1333
00:58:11,139 --> 00:58:12,619
I'm just
an amateur psychiatrist

1334
00:58:12,663 --> 00:58:18,103
to keep, um, my colleagues
cheerful when nothing works.

1335
00:58:18,146 --> 00:58:22,542
And quite a lot of the time
and I mean, 80, 90%,

1336
00:58:22,586 --> 00:58:24,413
it does not work.

1337
00:58:24,457 --> 00:58:26,720
If you are
at the forefront of science,

1338
00:58:26,764 --> 00:58:30,115
I can tell you,
you will fail a great deal.

1339
00:58:35,163 --> 00:58:37,165
I just felt disappointed.

1340
00:58:38,689 --> 00:58:41,605
Lesson I learned is that
ambition is a good thing,

1341
00:58:41,648 --> 00:58:43,694
but you need
to get the timing right.

1342
00:58:43,737 --> 00:58:46,784
There's no point being
50 years ahead of your time.

1343
00:58:46,827 --> 00:58:48,133
You will never survive

1344
00:58:48,176 --> 00:58:49,917
fifty years of
that kind of endeavor

1345
00:58:49,961 --> 00:58:51,963
before it yields something.

1346
00:58:52,006 --> 00:58:53,268
You'll literally die trying.

1347
00:59:08,936 --> 00:59:11,286
When we talk about AGI,

1348
00:59:11,330 --> 00:59:14,376
the holy grail
of artificial intelligence,

1349
00:59:14,420 --> 00:59:15,508
it becomes really difficult

1350
00:59:15,552 --> 00:59:17,815
to know what
we're even talking about.

1351
00:59:17,858 --> 00:59:19,643
Which bits
are we gonna see today?

1352
00:59:19,686 --> 00:59:21,645
We're going
to start in the garden.

1353
00:59:23,124 --> 00:59:25,649
This is the garden looking
from the observation area.

1354
00:59:25,692 --> 00:59:27,433
Research scientists
and engineers

1355
00:59:27,476 --> 00:59:30,871
can analyze and collaborate
and evaluate

1356
00:59:30,915 --> 00:59:33,004
what's going on in real time.

1357
00:59:33,047 --> 00:59:34,614
So in the 1800s,

1358
00:59:34,658 --> 00:59:37,008
we'd think of things like
television and the submarine

1359
00:59:37,051 --> 00:59:38,139
or a rocket ship to the moon

1360
00:59:38,183 --> 00:59:40,228
and say these things
are impossible.

1361
00:59:40,272 --> 00:59:41,490
Yet Jules Verne
wrote about them and,

1362
00:59:41,534 --> 00:59:44,406
a century and a half later,
they happened.

1363
00:59:44,450 --> 00:59:45,451
We'll be
experimenting

1364
00:59:45,494 --> 00:59:47,888
on civilizations really,

1365
00:59:47,932 --> 00:59:50,587
civilizations of AI agents.

1366
00:59:50,630 --> 00:59:52,719
Once the experiments
start going,

1367
00:59:52,763 --> 00:59:54,242
it's going to be
the most exciting thing ever.

1368
00:59:54,286 --> 00:59:56,984
So how will we get sleep?

1369
00:59:57,028 --> 00:59:58,682
I won't be able to sleep.

1370
00:59:58,725 --> 01:00:00,684
Full AGI
will be able to do

1371
01:00:00,727 --> 01:00:03,861
any cognitive task
a person can do.

1372
01:00:03,904 --> 01:00:08,387
It will be at a scale,
potentially, far beyond that.

1373
01:00:08,430 --> 01:00:10,302
It's really impossible for us

1374
01:00:10,345 --> 01:00:14,828
to imagine the outputs
of a superintelligent entity.

1375
01:00:14,872 --> 01:00:18,963
It's like asking a gorilla
to imagine, you know,

1376
01:00:19,006 --> 01:00:20,181
what Einstein does

1377
01:00:20,225 --> 01:00:23,402
when he produces
the theory of relativity.

1378
01:00:23,445 --> 01:00:25,491
People often ask me
these questions like,

1379
01:00:25,534 --> 01:00:29,495
"What happens if you're wrong,
and AGI is quite far away?"

1380
01:00:29,538 --> 01:00:31,453
And I'm like,
I never worry about that.

1381
01:00:31,497 --> 01:00:33,847
I actually
worry about the reverse.

1382
01:00:33,891 --> 01:00:37,242
I actually worry
that it's coming faster

1383
01:00:37,285 --> 01:00:39,723
than we can
really prepare for.

1384
01:00:42,073 --> 01:00:45,859
It really feels
like we're in a race to AGI.

1385
01:00:45,903 --> 01:00:49,907
The prototypes and the models
that we are developing now

1386
01:00:49,950 --> 01:00:51,822
are actually transforming

1387
01:00:51,865 --> 01:00:54,215
the space of what
we know about intelligence.

1388
01:00:57,349 --> 01:00:58,785
Recently,
we've had agents

1389
01:00:58,829 --> 01:01:00,047
that are powerful enough

1390
01:01:00,091 --> 01:01:03,442
to actually start
playing games in teams,

1391
01:01:03,485 --> 01:01:06,140
then competing
against other teams.

1392
01:01:06,184 --> 01:01:08,795
We're seeing
co-operative social dynamics

1393
01:01:08,839 --> 01:01:10,492
coming out of agents

1394
01:01:10,536 --> 01:01:13,321
where we haven't
pre-programmed in

1395
01:01:13,365 --> 01:01:15,584
any of these sorts
of dynamics.

1396
01:01:15,628 --> 01:01:19,240
It's completely learned
from their own experiences.

1397
01:01:20,807 --> 01:01:23,288
When we started,
we thought we were

1398
01:01:23,331 --> 01:01:25,725
out to build
an intelligence system

1399
01:01:25,769 --> 01:01:28,336
and convince the world
that we'd done it.

1400
01:01:28,380 --> 01:01:29,947
We're now starting
to wonder whether

1401
01:01:29,990 --> 01:01:31,296
we're gonna build systems

1402
01:01:31,339 --> 01:01:32,906
that we're not convinced
are fully intelligent,

1403
01:01:32,950 --> 01:01:34,691
and we're trying to convince
the world that they're not.

1404
01:01:38,651 --> 01:01:40,000
Hi, Alpha.

1405
01:01:40,044 --> 01:01:41,523
Hello there.

1406
01:01:41,567 --> 01:01:43,917
Where are we today?

1407
01:01:43,961 --> 01:01:46,659
You're at the Museum of
Modern Art in New York City.

1408
01:01:48,400 --> 01:01:53,013
Kind of.
Um, what painting is this?

1409
01:01:53,057 --> 01:01:55,494
This isThe Creation of Adam
by Michelangelo.

1410
01:01:55,537 --> 01:01:58,410
I don't think that painting
is in New York City.

1411
01:01:58,453 --> 01:02:01,543
You are right.
It's in the Vatican City.

1412
01:02:01,587 --> 01:02:02,806
Do you think
that's where we are?

1413
01:02:04,677 --> 01:02:05,809
No.

1414
01:02:05,852 --> 01:02:07,593
Then how
am I looking at this?

1415
01:02:10,509 --> 01:02:11,640
You're looking at

1416
01:02:11,684 --> 01:02:12,946
a digital copy
of the painting.

1417
01:02:12,990 --> 01:02:13,991
Yes.

1418
01:02:15,035 --> 01:02:16,341
Who lives
in this house?

1419
01:02:19,213 --> 01:02:20,345
An astronaut.

1420
01:02:26,046 --> 01:02:28,005
What's this astronaut's name?

1421
01:02:28,048 --> 01:02:29,397
His name is Mark.

1422
01:02:29,441 --> 01:02:32,052
This is a famous equation
by Albert Einstein.

1423
01:02:32,096 --> 01:02:35,577
It shows that energy and mass
are equivalent.

1424
01:02:35,621 --> 01:02:37,666
How would you explain it
to a five year old?

1425
01:02:40,931 --> 01:02:43,542
Well, imagine
you have a big rock,

1426
01:02:43,585 --> 01:02:45,326
and you want to know
how much it weighs.

1427
01:02:45,370 --> 01:02:47,589
Is Mark in danger?

1428
01:02:47,633 --> 01:02:49,722
No.
Mark is not in danger.

1429
01:02:49,766 --> 01:02:50,767
Why is that?

1430
01:02:54,422 --> 01:02:56,163
Because the dinosaur
is not real.

1431
01:02:57,338 --> 01:02:58,470
All right.

1432
01:02:58,513 --> 01:03:00,515
Why could you
not answer my question?

1433
01:03:00,559 --> 01:03:01,908
I think I couldn't
answer your question

1434
01:03:01,952 --> 01:03:04,041
because I don't know
enough about the subject.

1435
01:03:04,911 --> 01:03:06,130
Okay.

1436
01:03:06,173 --> 01:03:08,654
How could you learn
more about the subject?

1437
01:03:08,697 --> 01:03:09,960
I can learn more
about the subject

1438
01:03:10,003 --> 01:03:12,876
by reading books,
articles, and websites.

1439
01:03:12,919 --> 01:03:14,268
You can teach me.

1440
01:03:14,312 --> 01:03:15,835
My conclusion is,

1441
01:03:15,879 --> 01:03:17,619
if you now think about
what we're doing,

1442
01:03:17,663 --> 01:03:20,057
is learning from all humans,

1443
01:03:20,100 --> 01:03:22,581
all their knowledge at once
put on the Internet,

1444
01:03:22,624 --> 01:03:24,931
you would actually
know a lot about the world.

1445
01:03:24,975 --> 01:03:27,716
Like a significant portion
of everything humans can do.

1446
01:03:27,760 --> 01:03:29,631
And now,
I think it's more like,

1447
01:03:29,675 --> 01:03:30,937
"Well, it might just work."

1448
01:03:30,981 --> 01:03:33,113
This is a big moment.

1449
01:03:33,157 --> 01:03:34,636
Who is this?

1450
01:03:34,680 --> 01:03:36,551
This is God
reaching out to Adam.

1451
01:03:36,595 --> 01:03:37,988
The advent of AGI

1452
01:03:38,031 --> 01:03:40,860
will divide human history
into two parts.

1453
01:03:40,904 --> 01:03:43,950
The part up to that point
and the part after that point.

1454
01:03:43,994 --> 01:03:47,258
Why is he reaching out
to touch Adam?

1455
01:03:47,301 --> 01:03:48,912
God is
reaching out to touch Adam

1456
01:03:48,955 --> 01:03:50,217
to give him life.

1457
01:03:50,261 --> 01:03:52,219
It will give us a tool

1458
01:03:52,263 --> 01:03:57,529
that can completely reinvent
our entire civilization.

1459
01:03:57,572 --> 01:03:59,400
What does
this painting mean to you?

1460
01:04:01,750 --> 01:04:03,752
The painting
means a lot to me.

1461
01:04:03,796 --> 01:04:04,797
Okay. Like what?

1462
01:04:10,324 --> 01:04:11,456
I think
the painting is a reminder

1463
01:04:11,499 --> 01:04:12,674
that we are all
connected to each other

1464
01:04:12,718 --> 01:04:13,937
and that we are
all part of something

1465
01:04:13,980 --> 01:04:15,112
bigger than ourselves.

1466
01:04:16,461 --> 01:04:17,766
That's pretty nice.

1467
01:04:19,029 --> 01:04:21,379
When you cross
that barrier of

1468
01:04:21,422 --> 01:04:23,947
"AGI might happen
one day in the future"

1469
01:04:23,990 --> 01:04:26,645
to "No, actually, this could
really happen in a time frame

1470
01:04:26,688 --> 01:04:28,690
"that is sort of, like,
on my watch, you know,"

1471
01:04:28,734 --> 01:04:30,475
something changes
in your thinking.

1472
01:04:30,518 --> 01:04:32,694
learned to orient
itself by looking...

1473
01:04:32,738 --> 01:04:35,045
We have to be
careful with how we use it

1474
01:04:35,088 --> 01:04:37,177
and thoughtful about
how we deploy it.

1475
01:04:39,832 --> 01:04:41,138
You'd have to consider

1476
01:04:41,181 --> 01:04:42,487
what's its top level goal.

1477
01:04:42,530 --> 01:04:45,011
If it's to keep humans happy,

1478
01:04:45,055 --> 01:04:48,928
which set of humans?
What does happiness mean?

1479
01:04:48,972 --> 01:04:52,018
A lot of our collective goals
are very tricky,

1480
01:04:52,062 --> 01:04:54,891
even for humans to figure out.

1481
01:04:54,934 --> 01:04:58,503
Technology always
embeds our values.

1482
01:04:58,546 --> 01:05:01,680
It's not just technical,
it's ethical as well.

1483
01:05:01,723 --> 01:05:02,899
So we've got
to be really cautious

1484
01:05:02,942 --> 01:05:04,291
about what
we're building into it.

1485
01:05:04,335 --> 01:05:06,076
We're trying to find
a single algorithm which...

1486
01:05:06,119 --> 01:05:07,816
The reality is
that this is an algorithm

1487
01:05:07,860 --> 01:05:11,037
that has been created
by people, by us.

1488
01:05:11,081 --> 01:05:13,213
You know, what does it mean
to endow our agents

1489
01:05:13,257 --> 01:05:15,607
with the same kind of values
that we hold dear?

1490
01:05:15,650 --> 01:05:17,652
What is the purpose
of making these AI systems

1491
01:05:17,696 --> 01:05:19,045
appear so humanlike

1492
01:05:19,089 --> 01:05:20,742
so that they do
capture hearts and minds

1493
01:05:20,786 --> 01:05:21,961
because they're kind of

1494
01:05:22,005 --> 01:05:24,703
exploiting a human
vulnerability also?

1495
01:05:24,746 --> 01:05:26,531
The heart and mind
of these systems

1496
01:05:26,574 --> 01:05:28,054
are very much
human-generated data...

1497
01:05:28,098 --> 01:05:29,055
Mmm-hmm.

1498
01:05:29,099 --> 01:05:30,491
for all the good
and the bad.

1499
01:05:30,535 --> 01:05:32,015
There is a parallel

1500
01:05:32,058 --> 01:05:34,017
between
the Industrial Revolution,

1501
01:05:34,060 --> 01:05:36,758
which was an incredible
moment of displacement

1502
01:05:36,802 --> 01:05:42,373
and the current technological
change created by AI.

1503
01:05:42,416 --> 01:05:43,722
Pause AI!

1504
01:05:43,765 --> 01:05:45,724
We have to think
about who's displaced

1505
01:05:45,767 --> 01:05:48,596
and how we're going
to support them.

1506
01:05:48,640 --> 01:05:50,076
This technology
is coming a lot sooner,

1507
01:05:50,120 --> 01:05:52,426
uh, than really
the world knows or kind of

1508
01:05:52,470 --> 01:05:55,908
even we 18, 24 months
ago thought.

1509
01:05:55,952 --> 01:05:57,257
So there's
a tremendous opportunity,

1510
01:05:57,301 --> 01:05:58,389
tremendous excitement,

1511
01:05:58,432 --> 01:06:00,391
but also
tremendous responsibility.

1512
01:06:00,434 --> 01:06:01,740
It's happening so fast.

1513
01:06:02,654 --> 01:06:04,003
How will we govern it?

1514
01:06:05,135 --> 01:06:06,223
How will we decide

1515
01:06:06,266 --> 01:06:08,181
what is okay
and what is not okay?

1516
01:06:08,225 --> 01:06:10,923
AI-generated images are
getting more sophisticated.

1517
01:06:10,967 --> 01:06:14,535
The use of AI
for generating disinformation

1518
01:06:14,579 --> 01:06:17,016
and manipulating
human psychology

1519
01:06:17,060 --> 01:06:20,237
is only going to get
much, much worse.

1520
01:06:21,194 --> 01:06:22,587
AGI is coming,

1521
01:06:22,630 --> 01:06:24,632
whether we do it here
at DeepMind or not.

1522
01:06:25,459 --> 01:06:26,765
It's gonna happen,

1523
01:06:26,808 --> 01:06:29,028
so we better create
institutions to protect us.

1524
01:06:29,072 --> 01:06:30,595
It's gonna require
global coordination.

1525
01:06:30,638 --> 01:06:32,727
And I worry that humanity is

1526
01:06:32,771 --> 01:06:35,382
increasingly getting worse
at that rather than better.

1527
01:06:35,426 --> 01:06:37,123
We need
a lot more people

1528
01:06:37,167 --> 01:06:40,039
really taking this seriously
and thinking about this.

1529
01:06:40,083 --> 01:06:42,999
It's, yeah, it's serious.
It worries me.

1530
01:06:44,043 --> 01:06:45,871
It worries me. Yeah.

1531
01:06:45,914 --> 01:06:48,613
If you received
an email saying

1532
01:06:48,656 --> 01:06:50,832
this superior
alien civilization

1533
01:06:50,876 --> 01:06:52,791
is going to arrive on Earth,

1534
01:06:52,834 --> 01:06:54,575
there would be
emergency meetings

1535
01:06:54,619 --> 01:06:56,273
of all the governments.

1536
01:06:56,316 --> 01:06:58,144
We would go into overdrive

1537
01:06:58,188 --> 01:07:00,103
trying to figure out
how to prepare.

1538
01:07:01,669 --> 01:07:03,976
The arrival of AGI will be

1539
01:07:04,020 --> 01:07:06,935
the most important moment
that we have ever faced.

1540
01:07:14,378 --> 01:07:17,555
My dream
was that on the way to AGI,

1541
01:07:17,598 --> 01:07:20,688
we would create
revolutionary technologies

1542
01:07:20,732 --> 01:07:23,082
that would be
of use to humanity.

1543
01:07:23,126 --> 01:07:25,171
That's what I wanted
with AlphaFold.

1544
01:07:26,694 --> 01:07:28,653
I think
it's more important than ever

1545
01:07:28,696 --> 01:07:31,047
that we should solve
the protein folding problem.

1546
01:07:32,004 --> 01:07:34,224
This is gonna be really hard,

1547
01:07:34,267 --> 01:07:36,791
but I won't give up
until it's done.

1548
01:07:36,835 --> 01:07:37,879
You know,
we need to double down

1549
01:07:37,923 --> 01:07:40,317
and go as fast as possible
from here.

1550
01:07:40,360 --> 01:07:41,796
I think we've got
no time to lose.

1551
01:07:41,840 --> 01:07:45,757
So we are going to make
a protein folding strike team.

1552
01:07:45,800 --> 01:07:47,541
Team lead for the strike team
will be John.

1553
01:07:47,585 --> 01:07:48,673
Yeah, we've seen Alpha...

1554
01:07:48,716 --> 01:07:50,283
You know,
we're gonna try everything,

1555
01:07:50,327 --> 01:07:51,328
kitchen sink, the whole lot.

1556
01:07:52,198 --> 01:07:53,330
CASP14 is about

1557
01:07:53,373 --> 01:07:55,158
proving we can
solve the whole problem.

1558
01:07:56,333 --> 01:07:57,725
And I felt that to do that,

1559
01:07:57,769 --> 01:08:00,337
we would need to incorporate
some domain knowledge.

1560
01:08:01,903 --> 01:08:03,731
We had some
fantastic engineers on it,

1561
01:08:03,775 --> 01:08:05,733
but they were
not trained in biology.

1562
01:08:08,475 --> 01:08:10,260
As a computational biologist,

1563
01:08:10,303 --> 01:08:12,131
when I initially joined
the AlphaFold team,

1564
01:08:12,175 --> 01:08:14,220
I didn't immediately feel
confident about anything.

1565
01:08:14,264 --> 01:08:15,352
You know,

1566
01:08:15,395 --> 01:08:17,223
whether we were
gonna be successful.

1567
01:08:17,267 --> 01:08:21,097
Biology is so
ridiculously complicated.

1568
01:08:21,140 --> 01:08:25,101
It just felt like this very
far-off mountain to climb.

1569
01:08:25,144 --> 01:08:26,754
I'm starting to play with
the underlying temperatures

1570
01:08:26,798 --> 01:08:27,973
to see if we can get...

1571
01:08:28,016 --> 01:08:29,148
As one of the few people
on the team

1572
01:08:29,192 --> 01:08:31,846
who's done work
in biology before,

1573
01:08:31,890 --> 01:08:34,849
you feel this huge sense
of responsibility.

1574
01:08:34,893 --> 01:08:36,112
"We're expecting you to do

1575
01:08:36,155 --> 01:08:37,678
"great things
on this strike team."

1576
01:08:37,722 --> 01:08:38,897
That's terrifying.

1577
01:08:40,464 --> 01:08:42,727
But one of the reasons
why I wanted to come here

1578
01:08:42,770 --> 01:08:45,556
was to do
something that matters.

1579
01:08:45,599 --> 01:08:48,472
This is the number
of missing things.

1580
01:08:48,515 --> 01:08:49,951
What about making use

1581
01:08:49,995 --> 01:08:52,563
of whatever understanding
you have of physics?

1582
01:08:52,606 --> 01:08:54,391
Using that
as a source of data?

1583
01:08:54,434 --> 01:08:55,479
But if it's systematic...

1584
01:08:55,522 --> 01:08:56,784
Then, that can't be
right, though.

1585
01:08:56,828 --> 01:08:58,308
If it's systematically wrong
in some weird way,

1586
01:08:58,351 --> 01:09:01,224
you might be learning that
systematically wrong physics.

1587
01:09:01,267 --> 01:09:02,355
The team is already

1588
01:09:02,399 --> 01:09:04,749
trying to think
of multiple ways that...

1589
01:09:04,792 --> 01:09:06,229
Biological relevance

1590
01:09:06,272 --> 01:09:07,795
is what we're going for.

1591
01:09:09,057 --> 01:09:11,364
So we rewrote
the whole data pipeline

1592
01:09:11,408 --> 01:09:13,279
that AlphaFold uses to learn.

1593
01:09:13,323 --> 01:09:15,586
You can't
force the creative phase.

1594
01:09:15,629 --> 01:09:18,241
You have to give it space
for those flowers to bloom.

1595
01:09:19,242 --> 01:09:20,286
We won CASP.

1596
01:09:20,330 --> 01:09:22,070
Then it was
back to the drawing board

1597
01:09:22,114 --> 01:09:24,116
and like,
what are our new ideas?

1598
01:09:24,160 --> 01:09:26,945
Um, and then it's taken
a little while, I would say,

1599
01:09:26,988 --> 01:09:28,686
for them to get back
to where they were,

1600
01:09:28,729 --> 01:09:30,340
but with the new ideas.

1601
01:09:30,383 --> 01:09:31,515
And then now I think

1602
01:09:31,558 --> 01:09:33,952
we're seeing the benefits
of the new ideas.

1603
01:09:33,995 --> 01:09:35,736
They can go further, right?

1604
01:09:35,780 --> 01:09:38,130
So, um, that's a really
important moment.

1605
01:09:38,174 --> 01:09:40,959
I've seen that moment
so many times now,

1606
01:09:41,002 --> 01:09:42,613
but I know
what that means now.

1607
01:09:42,656 --> 01:09:44,484
And I know
this is the time now to press.

1608
01:09:45,920 --> 01:09:48,009
Adding side-chains
improves direct folding.

1609
01:09:48,053 --> 01:09:49,663
That drove
a lot of the progress.

1610
01:09:49,707 --> 01:09:51,012
- We'll talk about that.
- Great.

1611
01:09:51,056 --> 01:09:54,799
The last four months,
we've made enormous gains.

1612
01:09:54,842 --> 01:09:56,453
During CASP13,

1613
01:09:56,496 --> 01:09:59,499
it would take us a day or two
to fold one of the proteins,

1614
01:09:59,543 --> 01:10:01,762
and now we're folding, like,

1615
01:10:01,806 --> 01:10:03,938
hundreds of thousands
a second.

1616
01:10:03,982 --> 01:10:05,636
Yeah, it's just insane.

1617
01:10:05,679 --> 01:10:06,985
Now,
this is a model

1618
01:10:07,028 --> 01:10:09,901
that is
orders of magnitude faster,

1619
01:10:09,944 --> 01:10:12,251
while at the same time
being better.

1620
01:10:12,295 --> 01:10:13,644
We're getting
a lot of structures

1621
01:10:13,687 --> 01:10:15,254
into the high-accuracy regime.

1622
01:10:15,298 --> 01:10:17,517
We're rapidly improving
to a system

1623
01:10:17,561 --> 01:10:18,823
that is starting to really

1624
01:10:18,866 --> 01:10:20,477
get at the core and heart
of the problem.

1625
01:10:20,520 --> 01:10:21,695
It's great work.

1626
01:10:21,739 --> 01:10:23,088
It looks like
we're in good shape.

1627
01:10:23,131 --> 01:10:26,222
So we got, what, six,
five weeks left? Six weeks?

1628
01:10:26,265 --> 01:10:29,616
So what's, uh... Is it...
You got enough compute power?

1629
01:10:29,660 --> 01:10:31,531
I... We could use more.

1630
01:10:32,967 --> 01:10:34,360
I was nervous about CASP

1631
01:10:34,404 --> 01:10:36,580
but as the system
is starting to come together,

1632
01:10:36,623 --> 01:10:37,972
I don't feel as nervous.

1633
01:10:38,016 --> 01:10:39,496
I feel like things
have, sort of,

1634
01:10:39,539 --> 01:10:41,193
come into perspective
recently,

1635
01:10:41,237 --> 01:10:44,240
and, you know,
it's gonna be fine.

1636
01:10:47,330 --> 01:10:48,853
The Prime Minister
has announced

1637
01:10:48,896 --> 01:10:51,290
the most drastic limits
to our lives

1638
01:10:51,334 --> 01:10:53,858
the U.K. has ever seen
in living memory.

1639
01:10:53,901 --> 01:10:55,033
I must give the British people

1640
01:10:55,076 --> 01:10:56,904
a very simple instruction.

1641
01:10:56,948 --> 01:10:59,037
You must stay at home.

1642
01:10:59,080 --> 01:11:02,519
It feels like we're
in a science fiction novel.

1643
01:11:02,562 --> 01:11:04,869
You know, I'm delivering food
to my parents,

1644
01:11:04,912 --> 01:11:08,220
making sure
they stay isolated and safe.

1645
01:11:08,264 --> 01:11:10,570
I think it just highlights
the incredible need

1646
01:11:10,614 --> 01:11:12,877
for AI-assisted science.

1647
01:11:17,098 --> 01:11:18,361
You always know that

1648
01:11:18,404 --> 01:11:21,015
something like this
is a possibility.

1649
01:11:21,059 --> 01:11:23,888
But nobody ever really
believes it's gonna happen

1650
01:11:23,931 --> 01:11:25,585
in their lifetime, though.

1651
01:11:26,978 --> 01:11:29,154
- Are you recording yet?
- Yes.

1652
01:11:29,197 --> 01:11:31,025
- Okay, morning, all.
- Hey.

1653
01:11:31,069 --> 01:11:32,679
Good. CASP has started.

1654
01:11:32,723 --> 01:11:36,074
It's nice I get to sit around
in my pajama bottoms all day.

1655
01:11:36,117 --> 01:11:37,597
I never
thought I'd live in a house

1656
01:11:37,641 --> 01:11:39,164
where so much was going on.

1657
01:11:39,207 --> 01:11:41,427
I would be trying to solve
protein folding in one room,

1658
01:11:41,471 --> 01:11:42,559
and my husband would be trying

1659
01:11:42,602 --> 01:11:43,908
to make robots walk
in the other.

1660
01:11:46,954 --> 01:11:49,392
One of the hardest proteins
we've gotten in CASP thus far

1661
01:11:49,435 --> 01:11:51,219
is the SARS-CoV-2 protein

1662
01:11:51,263 --> 01:11:52,220
called ORF8.

1663
01:11:52,264 --> 01:11:54,919
ORF8 is
a coronavirus protein.

1664
01:11:54,962 --> 01:11:56,964
It's one of the main proteins,
um,

1665
01:11:57,008 --> 01:11:58,749
that dampens
the immune system.

1666
01:11:58,792 --> 01:12:00,054
We tried really hard

1667
01:12:00,098 --> 01:12:01,752
to improve our prediction.

1668
01:12:01,795 --> 01:12:03,493
Like, really, really hard.

1669
01:12:03,536 --> 01:12:05,582
Probably the most time
that we have ever spent

1670
01:12:05,625 --> 01:12:07,105
on a single target.

1671
01:12:07,148 --> 01:12:08,933
To the point where
my husband is, like,

1672
01:12:08,976 --> 01:12:12,197
"It's midnight.
You need to go to bed."

1673
01:12:12,240 --> 01:12:16,419
So I think we're at
Day 102 since lockdown.

1674
01:12:16,462 --> 01:12:19,944
My daughter
is keeping a journal.

1675
01:12:19,987 --> 01:12:22,120
Now you can go out
as much as you want.

1676
01:12:25,036 --> 01:12:27,212
We have received
the last target.

1677
01:12:27,255 --> 01:12:29,649
They've said they will be
sending out no more targets

1678
01:12:29,693 --> 01:12:31,347
in our category of CASP.

1679
01:12:32,652 --> 01:12:33,653
So we're just making sure

1680
01:12:33,697 --> 01:12:35,481
we get
the best possible answer.

1681
01:12:40,530 --> 01:12:43,315
As soon as we started
to get the results,

1682
01:12:43,359 --> 01:12:48,233
I'd sit down and start looking
at how close did anybody come

1683
01:12:48,276 --> 01:12:50,583
to getting the protein
structures correct.

1684
01:13:00,245 --> 01:13:01,551
- Oh, hi there.
- Hello.

1685
01:13:03,814 --> 01:13:07,078
It is an unbelievable thing,
CASP has finally ended.

1686
01:13:07,121 --> 01:13:09,472
I think it's at least time
to raise a glass.

1687
01:13:09,515 --> 01:13:11,212
Um, I don't know
if everyone has a glass

1688
01:13:11,256 --> 01:13:12,823
of something
that they can raise.

1689
01:13:12,866 --> 01:13:14,955
If not, raise,
I don't know, your laptops.

1690
01:13:14,999 --> 01:13:17,088
Um...

1691
01:13:17,131 --> 01:13:18,611
I'll probably make a speech
in a minute.

1692
01:13:18,655 --> 01:13:20,483
I feel like I should but I
just have no idea what to say.

1693
01:13:21,005 --> 01:13:24,269
So... let's see.

1694
01:13:24,312 --> 01:13:27,054
I feel like
a reading of email...

1695
01:13:27,098 --> 01:13:28,534
is the right thing to do.

1696
01:13:29,927 --> 01:13:31,232
When John said,

1697
01:13:31,276 --> 01:13:33,191
"I'm gonna read an email,"
at a team social,

1698
01:13:33,234 --> 01:13:35,498
I thought, "Wow, John,
you know how to have fun."

1699
01:13:35,541 --> 01:13:38,370
We're gonna read an email now.

1700
01:13:38,414 --> 01:13:41,634
Uh, I got this
about four o'clock today.

1701
01:13:42,722 --> 01:13:44,724
Um, it is from John Moult.

1702
01:13:45,725 --> 01:13:47,031
And I'll just read it.

1703
01:13:47,074 --> 01:13:49,381
It says,
"As I expect you know,

1704
01:13:49,425 --> 01:13:53,603
"your group has performed
amazingly well in CASP 14,

1705
01:13:53,646 --> 01:13:55,387
"both relative to other groups

1706
01:13:55,431 --> 01:13:57,911
"and in absolute
model accuracy."

1707
01:13:59,826 --> 01:14:01,219
"Congratulations on this work.

1708
01:14:01,262 --> 01:14:03,047
"It is really outstanding."

1709
01:14:03,090 --> 01:14:05,266
The structures were so good,

1710
01:14:05,310 --> 01:14:07,443
it was... it was just amazing.

1711
01:14:09,140 --> 01:14:10,750
After half a century,

1712
01:14:10,794 --> 01:14:12,230
we finally have a solution

1713
01:14:12,273 --> 01:14:14,928
to the protein folding
problem.

1714
01:14:14,972 --> 01:14:17,409
When I saw this email,
I read it,

1715
01:14:17,453 --> 01:14:19,585
I go, "Oh, shit!"

1716
01:14:19,629 --> 01:14:21,587
And my wife goes,
"Is everything okay?"

1717
01:14:21,631 --> 01:14:24,242
I call my parents, and just,
like, "Hey, Mum.

1718
01:14:24,285 --> 01:14:26,244
"Um, got something
to tell you.

1719
01:14:26,287 --> 01:14:27,550
"We've done this thing

1720
01:14:27,593 --> 01:14:29,813
"and it might be kind of
a big deal."

1721
01:14:29,856 --> 01:14:31,641
When I learned of
the CASP 14 results,

1722
01:14:32,642 --> 01:14:34,034
I was gobsmacked.

1723
01:14:34,078 --> 01:14:35,819
I was just excited.

1724
01:14:35,862 --> 01:14:38,909
This is a problem
that I was beginning to think

1725
01:14:38,952 --> 01:14:42,086
would not get solved
in my lifetime.

1726
01:14:42,129 --> 01:14:44,741
Now we have a tool
that can be used

1727
01:14:44,784 --> 01:14:46,612
practically by scientists.

1728
01:14:46,656 --> 01:14:48,440
These people
are asking us, you know,

1729
01:14:48,484 --> 01:14:50,224
"I've got this protein
involved in malaria,"

1730
01:14:50,268 --> 01:14:52,139
or, you know,
some infectious disease.

1731
01:14:52,183 --> 01:14:53,227
"We don't know the structure.

1732
01:14:53,271 --> 01:14:55,186
"Can we use AlphaFold
to solve it?"

1733
01:14:55,229 --> 01:14:56,970
We can easily predict
all known sequences

1734
01:14:57,014 --> 01:14:58,276
in a month.

1735
01:14:58,319 --> 01:14:59,973
All known sequences
in a month?

1736
01:15:00,017 --> 01:15:01,279
- Yeah, easily.
- Mmm-hmm?

1737
01:15:01,322 --> 01:15:02,585
A billion, two billion.

1738
01:15:02,628 --> 01:15:03,673
Um, and they're...

1739
01:15:03,716 --> 01:15:05,196
So why don't we just do that?
Yeah.

1740
01:15:05,239 --> 01:15:07,111
- We should just do that a lot.
- Well, I mean...

1741
01:15:07,154 --> 01:15:09,243
That's way better.
Why don't we just do that?

1742
01:15:09,287 --> 01:15:11,115
So that's
one of the options.

1743
01:15:11,158 --> 01:15:12,638
- Right.
- There's this...

1744
01:15:12,682 --> 01:15:15,119
We should just...
Right, that's a great idea.

1745
01:15:15,162 --> 01:15:17,513
We should just run
every protein in existence.

1746
01:15:18,296 --> 01:15:19,471
And then release that.

1747
01:15:19,515 --> 01:15:20,994
Why didn't someone
suggest this before?

1748
01:15:21,038 --> 01:15:22,126
Of course that's
what we should do.

1749
01:15:22,169 --> 01:15:23,954
Why are we thinking about
making a service

1750
01:15:23,997 --> 01:15:25,651
and then people submit
their protein?

1751
01:15:25,695 --> 01:15:26,913
We just fold everything.

1752
01:15:26,957 --> 01:15:28,654
And then give it to
everyone in the world.

1753
01:15:28,698 --> 01:15:31,483
Who knows how many discoveries
will be made from that?

1754
01:15:31,527 --> 01:15:33,790
Demis called us up
and said,

1755
01:15:33,833 --> 01:15:35,618
"We want to make this open.

1756
01:15:35,661 --> 01:15:37,837
"Not just make sure
the code is open,

1757
01:15:37,881 --> 01:15:39,578
"but we're gonna make it
really easy

1758
01:15:39,622 --> 01:15:42,668
"for everybody to get access
to the predictions."

1759
01:15:45,062 --> 01:15:47,238
That is fantastic.

1760
01:15:47,281 --> 01:15:49,327
It's like drawing back
the curtain

1761
01:15:49,370 --> 01:15:52,852
and seeing the whole world
of protein structures.

1762
01:15:55,246 --> 01:15:56,987
They released the structures

1763
01:15:57,030 --> 01:15:59,772
of 200 million proteins.

1764
01:15:59,816 --> 01:16:01,818
These are gifts to humanity.

1765
01:16:07,650 --> 01:16:10,914
The moment AlphaFold
is live to the world,

1766
01:16:10,957 --> 01:16:13,873
we will no longer be
the most important people

1767
01:16:13,917 --> 01:16:15,222
in AlphaFold's story.

1768
01:16:15,266 --> 01:16:16,833
Can't quite believe
it's all out.

1769
01:16:16,876 --> 01:16:18,356
Aw!

1770
01:16:18,399 --> 01:16:20,314
A hundred
and sixty-four users.

1771
01:16:20,358 --> 01:16:22,578
Loads of activity in Japan.

1772
01:16:22,621 --> 01:16:24,928
We have 655 users currently.

1773
01:16:24,971 --> 01:16:26,930
We currently
have 100,000 concurrent users.

1774
01:16:26,973 --> 01:16:28,192
Wow!

1775
01:16:31,108 --> 01:16:33,893
Today is just crazy.

1776
01:16:33,937 --> 01:16:36,504
What an absolutely
unbelievable effort

1777
01:16:36,548 --> 01:16:37,723
from everyone.

1778
01:16:37,767 --> 01:16:38,550
We're gonna all remember
these moments

1779
01:16:38,594 --> 01:16:40,030
for the rest of our lives.

1780
01:16:40,073 --> 01:16:41,727
I'm excited about AlphaFold.

1781
01:16:41,771 --> 01:16:45,601
For my research, it's already
propelling lots of progress.

1782
01:16:45,644 --> 01:16:47,385
And this is
just the beginning.

1783
01:16:47,428 --> 01:16:48,908
My guess is,

1784
01:16:48,952 --> 01:16:53,043
every single biological
and chemistry achievement

1785
01:16:53,086 --> 01:16:55,698
will be related to AlphaFold
in some way.

1786
01:17:13,367 --> 01:17:15,413
AlphaFold is an index moment.

1787
01:17:15,456 --> 01:17:18,068
It's a moment
that people will not forget

1788
01:17:18,111 --> 01:17:20,244
because the world changed.

1789
01:17:39,655 --> 01:17:41,482
Everybody's realized now

1790
01:17:41,526 --> 01:17:43,746
what Shane and I have known
for more than 20 years,

1791
01:17:43,789 --> 01:17:46,618
that AI is going to be
the most important thing

1792
01:17:46,662 --> 01:17:48,446
humanity's ever gonna invent.

1793
01:17:48,489 --> 01:17:50,230
We will shortly be arriving

1794
01:17:50,274 --> 01:17:52,058
at our final destination.

1795
01:18:02,068 --> 01:18:04,767
The pace of
innovation and capabilities

1796
01:18:04,810 --> 01:18:06,507
is accelerating,

1797
01:18:06,551 --> 01:18:09,293
like a boulder rolling down
a hill that we've kicked off

1798
01:18:09,336 --> 01:18:12,644
and now it's continuing
to gather speed.

1799
01:18:12,688 --> 01:18:15,299
We are at
a crossroads in human history.

1800
01:18:15,342 --> 01:18:16,735
AI has the potential

1801
01:18:16,779 --> 01:18:19,172
to transform our lives
in every aspect.

1802
01:18:19,216 --> 01:18:23,786
It's no less important than
the discovery of electricity.

1803
01:18:23,829 --> 01:18:26,484
We should be looking
at the scientific method

1804
01:18:26,527 --> 01:18:28,834
and trying to understand
each step of the way

1805
01:18:28,878 --> 01:18:30,096
in a rigorous way.

1806
01:18:30,140 --> 01:18:32,664
This is a moment
of profound opportunity.

1807
01:18:32,708 --> 01:18:34,753
Harnessing this technology

1808
01:18:34,797 --> 01:18:37,713
could eclipse anything
we have ever known.

1809
01:18:42,195 --> 01:18:43,675
Hi, Alpha.

1810
01:18:44,676 --> 01:18:45,764
Hi.

1811
01:18:47,157 --> 01:18:48,419
What is this?

1812
01:18:50,682 --> 01:18:53,729
This is a chessboard.

1813
01:18:53,772 --> 01:18:56,514
If I was to play white, what
move would you recommend?

1814
01:18:59,865 --> 01:19:00,953
I would recommend

1815
01:19:00,997 --> 01:19:02,781
that you move your pawn
from E2 to E4.

1816
01:19:05,871 --> 01:19:08,787
And now if you were black,
what would you play now?

1817
01:19:11,572 --> 01:19:13,618
I would play
the Sicilian Defense.

1818
01:19:15,838 --> 01:19:16,882
That's a good choice.

1819
01:19:19,406 --> 01:19:21,452
Thanks.

1820
01:19:23,715 --> 01:19:25,891
So what do you see?
What is this object?

1821
01:19:28,546 --> 01:19:30,504
This is a pencil sculpture.

1822
01:19:32,811 --> 01:19:35,031
What happens if I move
one of the pencils?

1823
01:19:37,990 --> 01:19:39,470
If you move
one of the pencils,

1824
01:19:39,513 --> 01:19:42,081
the sculpture will fall apart.

1825
01:19:42,125 --> 01:19:44,301
I'd better leave it alone,
then.

1826
01:19:44,344 --> 01:19:45,868
That's probably
a good idea.

1827
01:19:50,568 --> 01:19:52,744
AGI is on the horizon now.

1828
01:19:54,833 --> 01:19:56,661
Very clearly
the next generation

1829
01:19:56,704 --> 01:19:58,141
is going to live
in a future world

1830
01:19:58,184 --> 01:20:01,057
where things will be radically
different because of AI.

1831
01:20:02,493 --> 01:20:05,496
And if you want to steward
that responsibly,

1832
01:20:05,539 --> 01:20:09,239
every moment is vital.

1833
01:20:09,282 --> 01:20:12,677
This is the moment I've been
living my whole life for.

1834
01:20:19,162 --> 01:20:21,120
It's just
a good thinking game.