1
00:00:00,160 --> 00:00:02,240
Speaker 1: Latest interview of Elon Musk.

2
00:00:02,120 --> 00:00:04,639
Speaker 2: Bill Gates said, there is no one in our time

3
00:00:04,679 --> 00:00:07,280
who has done more to push the bounds of science.

4
00:00:06,960 --> 00:00:07,879
Speaker 3: Innovation than you.

5
00:00:08,119 --> 00:00:09,599
Speaker 1: What's kind of going to say, well.

6
00:00:09,480 --> 00:00:11,400
Speaker 2: That's a nice thing to have anyone say about you.

7
00:00:11,519 --> 00:00:13,960
Nice coming from Bill Gates. But oddly enough, when it

8
00:00:13,960 --> 00:00:17,839
comes to AI, actually, for around a decade, you've almost

9
00:00:17,839 --> 00:00:20,320
been doing the opposite and saying, hang on, you need

10
00:00:20,359 --> 00:00:22,719
to think about what we're doing and what we're pushing here,

11
00:00:22,760 --> 00:00:23,600
and what do we do to.

12
00:00:23,640 --> 00:00:24,199
Speaker 3: Make this safe?

13
00:00:24,280 --> 00:00:26,519
Speaker 2: And actually maybe we shouldn't be pushing as faster or

14
00:00:26,879 --> 00:00:29,000
as hard as we are. Like, I mean, you've been

15
00:00:29,000 --> 00:00:30,760
doing it for a decade, Like what was it that

16
00:00:31,079 --> 00:00:33,359
caused you to think about it that way? And you know,

17
00:00:33,359 --> 00:00:34,799
why do we need to be worried?

18
00:00:34,880 --> 00:00:36,799
Speaker 1: Yeah? I've been somewhat of a Cassandra for quite a

19
00:00:36,880 --> 00:00:39,119
while where people would I tell you, like, we should

20
00:00:39,119 --> 00:00:40,399
really be concerned about AI, and that'd be like, what

21
00:00:40,399 --> 00:00:42,119
are you talking about? Like I've never really had any

22
00:00:42,119 --> 00:00:46,079
experience with it since I was immersed in technology. I

23
00:00:46,079 --> 00:00:48,320
have been immersed in technology for a long time. I

24
00:00:48,320 --> 00:00:50,840
could see it coming, so, but I think this year

25
00:00:51,039 --> 00:00:53,280
was that there's been a number of breakthroughs. I mean,

26
00:00:53,719 --> 00:00:55,439
you know, the point in which someone can see a

27
00:00:55,600 --> 00:00:58,840
dynamically created video of themselves, you know, like it, so

28
00:00:58,960 --> 00:01:01,359
you can make a video of you anything in real time,

29
00:01:01,560 --> 00:01:03,719
or me. And so there's sort of the deep pic

30
00:01:03,840 --> 00:01:06,040
videos which are really incredibly good, in fact, sometimes more

31
00:01:06,040 --> 00:01:09,120
convincing than real ones and deep real and then and

32
00:01:09,159 --> 00:01:12,599
then obviously things like chat GPT were quite remarkable, and

33
00:01:12,599 --> 00:01:15,480
now I saw a GPT one, GP two, GPT three,

34
00:01:15,560 --> 00:01:17,680
GPD four that you know, the whole sort of lead

35
00:01:17,760 --> 00:01:20,159
up to that, so it was easy for me to

36
00:01:20,239 --> 00:01:21,920
kind of see where it's going. If you just sort

37
00:01:21,959 --> 00:01:24,480
of extrapolate the points on a curve and as seeing

38
00:01:24,480 --> 00:01:28,359
that tread will continue, then we will have profound artificial intelligence,

39
00:01:28,719 --> 00:01:31,319
and obviously at the level that far exceeds human intelligence.

40
00:01:31,560 --> 00:01:33,439
So but I'm glad to see at this point that

41
00:01:33,680 --> 00:01:36,400
people are taking safety seriously, and I like, I say

42
00:01:36,599 --> 00:01:38,879
thank you for holding this AI safety conference. I think

43
00:01:38,920 --> 00:01:41,120
actually it will go down in history as being very important.

44
00:01:41,159 --> 00:01:43,959
I think it's really quite profound, and I do think

45
00:01:44,040 --> 00:01:47,280
overall that the potential is there for artificial intelligence AI

46
00:01:47,439 --> 00:01:51,239
to have most likely a positive effect and to create

47
00:01:51,280 --> 00:01:53,920
a future of abundance where there is no scarcity of

48
00:01:53,920 --> 00:01:56,920
goods and services. But it is somewhat of the magic

49
00:01:57,000 --> 00:01:59,439
gene problem where if you have a magic gene that

50
00:01:59,599 --> 00:02:02,959
can grab all the wishes, usually those stories don't end well.

51
00:02:03,120 --> 00:02:05,280
Too careful what you wish for, including wishes.

52
00:02:05,400 --> 00:02:07,879
Speaker 2: So you talked a little bit about the summit and

53
00:02:08,199 --> 00:02:10,280
thank you for being engaged in it, which has been

54
00:02:10,280 --> 00:02:13,800
great and people enjoyed having you there pulling in this dialogue. Now,

55
00:02:13,800 --> 00:02:16,240
one of the things that we achieved today in the

56
00:02:16,280 --> 00:02:20,960
meetings between the companies and the leaders was an agreement

57
00:02:21,080 --> 00:02:25,840
that externally, ideally, governments should be doing safety testing of

58
00:02:26,000 --> 00:02:28,280
models before their relief. I think this is something that

59
00:02:28,319 --> 00:02:30,000
you've spoken about a little bit. It was something we

60
00:02:30,039 --> 00:02:32,639
worked really hard on because you know, my job in

61
00:02:32,759 --> 00:02:35,719
government is to say, hang on, there is a potential risk,

62
00:02:35,759 --> 00:02:38,080
and not a definite risk, but a potential risk of

63
00:02:38,159 --> 00:02:39,560
something that could be bad.

64
00:02:40,120 --> 00:02:41,599
Speaker 3: My job is to protect.

65
00:02:41,240 --> 00:02:43,680
Speaker 2: The country and we can only do that if we

66
00:02:43,800 --> 00:02:46,280
develop the capability we need in our safety institute and

67
00:02:46,280 --> 00:02:47,840
then go in and make sure we can test the.

68
00:02:47,759 --> 00:02:49,599
Speaker 3: Models before they are released.

69
00:02:49,639 --> 00:02:51,759
Speaker 2: Delight of that happened today, But you know, what's your

70
00:02:51,840 --> 00:02:53,759
view on what we should be doing. Right, You've talked

71
00:02:53,759 --> 00:02:56,000
about the potential risk. Right again, we don't know, but

72
00:02:56,319 --> 00:02:58,159
you know, what are the types of things governments like

73
00:02:58,199 --> 00:03:01,319
ours should be doing to manage and degate against those risks.

74
00:03:01,479 --> 00:03:03,639
Speaker 1: Well, I generally think that it is good for government

75
00:03:03,680 --> 00:03:07,319
to play a role when the public safety is at risk. So,

76
00:03:07,719 --> 00:03:10,080
you know, really, for the vast majority of software, the

77
00:03:10,080 --> 00:03:12,840
public safety is not at risk. If the app crashes

78
00:03:12,840 --> 00:03:15,719
on your phone, your laptop, it's not a massive catastrophe.

79
00:03:15,879 --> 00:03:20,319
But when talking about digital superintelligence, I think, which does

80
00:03:20,400 --> 00:03:22,960
pose a risk to the public, then there is a

81
00:03:23,039 --> 00:03:25,560
role for government to play to safeguard the interest of

82
00:03:25,599 --> 00:03:28,800
the public. And this is of course true in many fields,

83
00:03:29,080 --> 00:03:32,159
you know, aviation, cars, you know, I deal with regulators

84
00:03:32,360 --> 00:03:37,080
throughout the world because of stalling being communications, rockets being aerospace,

85
00:03:37,319 --> 00:03:40,159
and cars, you know, being vehicle transport. So I'm very

86
00:03:40,159 --> 00:03:42,599
familiar with dealing with with regulators, and I actually agree

87
00:03:42,599 --> 00:03:44,319
with the vast majority of regulations. There's a few that

88
00:03:44,360 --> 00:03:46,000
I disagree with from time to time, but zero point

89
00:03:46,080 --> 00:03:49,199
one percent, probably less than one percent of regulations I

90
00:03:49,240 --> 00:03:52,120
disagree with. So there is some concern from people in

91
00:03:52,120 --> 00:03:55,240
Silicon Valley who who have never dealt with regulators before,

92
00:03:55,280 --> 00:03:57,680
and they think that this is going to just crush

93
00:03:57,680 --> 00:04:00,719
innovation and slow them down and be annoying. But and

94
00:04:00,879 --> 00:04:03,159
it will be annoying, It's true. They're not wrong about that.

95
00:04:03,400 --> 00:04:06,159
But I think there's we've learnt over the years that

96
00:04:06,639 --> 00:04:09,680
having a referee is a good thing. And if you

97
00:04:09,719 --> 00:04:12,599
look at any sports game, there's always a referee, and

98
00:04:12,680 --> 00:04:15,159
nobody's suggesting, I think to have a sports game without one.

99
00:04:15,240 --> 00:04:16,959
And I think that's the right way to think about this,

100
00:04:17,120 --> 00:04:19,519
is for for governed to be a referee, to make

101
00:04:19,560 --> 00:04:23,879
sure the sportsmanlike conduct and that the public safety is

102
00:04:24,160 --> 00:04:26,279
you know, is addressed. That we care about the public

103
00:04:26,279 --> 00:04:28,360
safety because I think there might be at times too

104
00:04:28,439 --> 00:04:30,600
much optimism about technology. And I speak I say that

105
00:04:30,639 --> 00:04:32,600
as a technologist. I mean, so I ought to know

106
00:04:32,800 --> 00:04:35,360
and like I said that, on balance, I think that

107
00:04:35,439 --> 00:04:39,560
the AI will be a forceful good most likely, but

108
00:04:39,600 --> 00:04:42,439
the probability of it going bad it's not zero percent. Yeah,

109
00:04:42,680 --> 00:04:45,079
so we just need to mitigate the downside potential.

110
00:04:45,279 --> 00:04:47,639
Speaker 3: And then how you talk about referee and that's.

111
00:04:47,560 --> 00:04:48,680
Speaker 1: What we're demonstrate there.

112
00:04:48,759 --> 00:04:49,759
Speaker 3: Yeah, well, there we go.

113
00:04:49,879 --> 00:04:51,199
Speaker 2: I mean you know, and we talked about this in

114
00:04:51,240 --> 00:04:53,560
demos and I discussed this a long time ago, like

115
00:04:53,600 --> 00:04:56,959
literally facing right and actually Demoster his credit and the

116
00:04:57,000 --> 00:04:58,480
credit of people in the industry.

117
00:04:58,079 --> 00:04:58,800
Speaker 3: Did say that to us.

118
00:04:59,160 --> 00:05:01,439
Speaker 2: You say it's not right. Yeah, that Demis and his

119
00:05:01,519 --> 00:05:02,959
colleagues are marking their own homework.

120
00:05:03,040 --> 00:05:05,160
Speaker 3: Right. There needs to be someone independent, and.

121
00:05:05,120 --> 00:05:08,519
Speaker 2: That's why we've developed the Safety Institute here. Do you

122
00:05:08,600 --> 00:05:11,279
think governments can develop the expertise. One of the things

123
00:05:11,360 --> 00:05:13,199
we need to do is they hang on you know,

124
00:05:13,560 --> 00:05:15,839
Denis and Sam all the others. I've got a lot

125
00:05:15,839 --> 00:05:18,879
of very smart people doing this. Governments need to quickly

126
00:05:19,000 --> 00:05:22,399
tool up capability wise, personnel wise, which is what we're doing.

127
00:05:22,759 --> 00:05:24,240
I mean, do you think it is possible for governments

128
00:05:24,279 --> 00:05:26,800
to do that fast enough, given how quickly the technology

129
00:05:27,079 --> 00:05:28,439
is developing, or what do we need to do to

130
00:05:28,439 --> 00:05:29,680
make sure we do it quick enough?

131
00:05:29,800 --> 00:05:31,680
Speaker 1: No, I think it's a great point you're making. The

132
00:05:31,720 --> 00:05:35,000
pace of AI is faster than any technology I've seen

133
00:05:35,040 --> 00:05:37,399
in history by far, and it seems to be growing

134
00:05:37,399 --> 00:05:41,240
in capability by at least fivefold peraps tenfold per year.

135
00:05:41,360 --> 00:05:44,360
It'll certainly grow by an automatitude next year. Yeah, so,

136
00:05:45,199 --> 00:05:48,000
and government isn't used to moving at that speed. But

137
00:05:48,040 --> 00:05:51,319
I think even if there are not firm regulations, even

138
00:05:51,360 --> 00:05:53,959
if there's not even if there isn't an enforcement capability,

139
00:05:54,160 --> 00:05:56,879
so we're having insight and being able to highlight concerns

140
00:05:56,920 --> 00:05:59,879
to the public will be very powerful. So even if

141
00:06:00,000 --> 00:06:02,040
that's all that's accomplished, I think that will be very good.

142
00:06:02,160 --> 00:06:04,279
Speaker 3: Okay, y well, hopefully we can do better than that.

143
00:06:04,319 --> 00:06:04,680
Speaker 1: Hopefully.

144
00:06:04,759 --> 00:06:07,000
Speaker 2: Yeah, yeah, no, but that's how fair. Actually, we were

145
00:06:07,040 --> 00:06:09,639
talking before it was striking. You know, you're someone who

146
00:06:09,639 --> 00:06:13,879
spent their life in technology. They're living More's law and

147
00:06:13,959 --> 00:06:16,000
what it was interesting over the last couple of days

148
00:06:16,040 --> 00:06:18,680
talking to everyone who's doing the development of this, and

149
00:06:18,720 --> 00:06:21,000
I think you concur with this is just the pace

150
00:06:21,040 --> 00:06:24,600
of advancement here is unlike anything all of you have

151
00:06:24,680 --> 00:06:26,480
seen in your careers and technology.

152
00:06:26,519 --> 00:06:28,360
Speaker 3: Is that fair because you've got these kind.

153
00:06:28,199 --> 00:06:31,360
Speaker 2: Of compounding effects from the hardware and the data and

154
00:06:31,439 --> 00:06:32,040
the personnel.

155
00:06:32,279 --> 00:06:35,000
Speaker 1: Yeah, I mean the two currently the two leading centers

156
00:06:35,000 --> 00:06:37,720
for a development are the San Franusco area and the

157
00:06:37,759 --> 00:06:40,079
sort of London area, And there are many other places

158
00:06:40,079 --> 00:06:41,839
where it's being done, but those are the two leading areas.

159
00:06:41,879 --> 00:06:44,240
So I think if you know, if the United States

160
00:06:44,279 --> 00:06:48,240
and the UK and China are sort of aligned on safety,

161
00:06:48,279 --> 00:06:50,959
that's all going to be. That's really that's where that's

162
00:06:51,000 --> 00:06:52,199
where the leadership is general.

163
00:06:52,279 --> 00:06:55,079
Speaker 2: And you actually you mentioned China there. So I took

164
00:06:55,120 --> 00:06:58,079
a decision to invite China to summit over the last

165
00:06:58,079 --> 00:07:00,720
three days, and it was not easy decision. A lot

166
00:07:00,720 --> 00:07:03,120
of people criticize me for you know, my viewers, if

167
00:07:03,120 --> 00:07:05,720
you're going to try it essential serious conversation, you need to.

168
00:07:05,759 --> 00:07:07,800
But what would your thoughts? You do business all around

169
00:07:07,839 --> 00:07:10,560
the world. You just talked about it there. Should we

170
00:07:10,600 --> 00:07:12,639
be engaging with them? Can we trust them? Is that

171
00:07:12,680 --> 00:07:13,560
the right thing to have done?

172
00:07:14,040 --> 00:07:15,959
Speaker 1: If we don't, if China is not on board with

173
00:07:16,120 --> 00:07:19,439
AI safety, it's somewhat of a moot situation. The single

174
00:07:19,439 --> 00:07:22,560
biggest objection that I get to a kind of AI

175
00:07:22,680 --> 00:07:25,560
regulation or sort of safety controls are well, China is

176
00:07:25,600 --> 00:07:27,360
not going to do it, and therefore they will just

177
00:07:27,639 --> 00:07:30,040
jump into the lead and exceed us all. But actually

178
00:07:30,160 --> 00:07:33,680
China is willing to participate in a safety and thank

179
00:07:33,680 --> 00:07:36,000
you for inviting them, and they you know, I think

180
00:07:36,000 --> 00:07:38,519
we should thank China for for attending. When I was

181
00:07:38,519 --> 00:07:41,680
in China earlier this year, My main subject of discussion

182
00:07:41,800 --> 00:07:44,759
with the leadership in China was AI safety and saying

183
00:07:44,759 --> 00:07:46,839
that this is really something that they should care about,

184
00:07:47,040 --> 00:07:50,000
and they took it seriously and you are too, which

185
00:07:50,040 --> 00:07:52,800
is great. And having them here I think was essential. Really,

186
00:07:52,839 --> 00:07:55,279
if they're not participants, it's pointless.

187
00:07:54,800 --> 00:07:57,199
Speaker 2: It's yeah, I know that's and I think we were pleased.

188
00:07:57,199 --> 00:07:59,360
I think they were engaged yesterday in the discussions and

189
00:07:59,360 --> 00:08:02,519
actually ended up signing the same communicy that everyone else did,

190
00:08:02,920 --> 00:08:04,920
which is a good stop right. And I said, if

191
00:08:05,000 --> 00:08:07,319
we need everyone to approach us in a similar way,

192
00:08:07,360 --> 00:08:10,360
if we're going to have a realistic chance of resolving it.

193
00:08:10,600 --> 00:08:12,839
I was going to You talked about innovation earlier and

194
00:08:12,959 --> 00:08:14,040
regulation being annoying.

195
00:08:14,279 --> 00:08:16,680
Speaker 3: There was a good debate today we had about.

196
00:08:16,319 --> 00:08:18,399
Speaker 2: Open source, and I think you've kind of been a

197
00:08:18,439 --> 00:08:21,920
proponent of algorithmic transparency and making some of the ex

198
00:08:21,959 --> 00:08:25,399
algorithms public and actually we were talking about every hint

199
00:08:25,399 --> 00:08:28,519
and on the way in. He's particularly been very concerned

200
00:08:28,560 --> 00:08:31,079
about open source models being used.

201
00:08:30,879 --> 00:08:31,800
Speaker 3: By bad actors.

202
00:08:32,159 --> 00:08:33,799
Speaker 2: You've got a group of people who say they are

203
00:08:34,120 --> 00:08:37,200
critical to innovation happening in that distributed way.

204
00:08:37,440 --> 00:08:38,159
Speaker 3: Look, it's a trick.

205
00:08:38,039 --> 00:08:40,879
Speaker 2: There's probably no perfect answer, and there is a tricky balance.

206
00:08:40,960 --> 00:08:42,799
What are your thoughts on how we should approach this

207
00:08:42,840 --> 00:08:46,799
open source question or where should we be targeting whatever

208
00:08:46,960 --> 00:08:48,759
regulatory or monitoring that we're going.

209
00:08:48,759 --> 00:08:52,080
Speaker 1: To do well. The open source algorithms and data tend

210
00:08:52,080 --> 00:08:54,600
to lag the closed source by six to twelve. But

211
00:08:55,600 --> 00:08:57,759
given the rate of improvement that there's actually there for

212
00:08:57,919 --> 00:09:00,440
quite a big difference between the closed source in the

213
00:09:00,480 --> 00:09:02,799
open If things are improving by a factor of let's

214
00:09:02,840 --> 00:09:05,639
say five or more than being a year behind is

215
00:09:05,720 --> 00:09:07,960
you're five times worse. So it's a pretty big difference.

216
00:09:08,080 --> 00:09:10,320
And that might be actually an okay situation, but it

217
00:09:10,399 --> 00:09:12,480
certainly we'll get to the point where you've got open

218
00:09:12,519 --> 00:09:15,159
source AI that can do that, that will start to

219
00:09:15,200 --> 00:09:17,960
approach human level intelligence, Will PAPS succeeded. I don't know

220
00:09:18,039 --> 00:09:19,799
quite what to do about it. I think it's somewhat

221
00:09:19,799 --> 00:09:21,879
inevermentle there will be some amount of open source, and

222
00:09:22,000 --> 00:09:24,360
I guess I would have a slight bias towards open

223
00:09:24,399 --> 00:09:26,200
source because at least you can see what's going on.

224
00:09:26,320 --> 00:09:28,840
It wears closed source. You don't know what's going on now.

225
00:09:29,000 --> 00:09:30,759
It should be said with AI that even if it's

226
00:09:30,759 --> 00:09:32,840
an open source, do you actually know what's going on?

227
00:09:32,919 --> 00:09:35,679
Because if you've got a gigantic data file and you

228
00:09:35,679 --> 00:09:38,440
know sort of billions of days of weights and parameters,

229
00:09:38,600 --> 00:09:40,200
you can't just read it and see what it's going

230
00:09:40,200 --> 00:09:43,639
to do. It's a gigantic file of inscrutable numbers. You

231
00:09:43,679 --> 00:09:45,679
can test it when you run it. You can test it.

232
00:09:46,080 --> 00:09:47,519
You can run a bunch of tests to see what

233
00:09:47,559 --> 00:09:51,159
it's going to do. But it's probabilistic as opposed to deterministic.

234
00:09:51,240 --> 00:09:53,360
It's not like traditional programming, where you've got it you've

235
00:09:53,399 --> 00:09:56,080
got very discrete logic and the outcome is very predictable

236
00:09:56,080 --> 00:09:57,639
and you can read each line and see what each

237
00:09:57,639 --> 00:09:59,919
line's going to do. A neural net is just a

238
00:10:00,159 --> 00:10:02,559
much probabilities. I mean, it sort of ends up being

239
00:10:02,559 --> 00:10:05,639
a giant Comma separated value file. It's like our digital

240
00:10:05,679 --> 00:10:08,480
guide is a CSP file. Really, Okay, that is kind

241
00:10:08,519 --> 00:10:09,000
of what it is.

242
00:10:09,120 --> 00:10:11,600
Speaker 2: Yeah, Now, at that point you've just made is one

243
00:10:11,600 --> 00:10:14,679
that we have been talking about a lot, because again

244
00:10:15,080 --> 00:10:17,120
conversation with the people are developing their technology.

245
00:10:17,159 --> 00:10:19,120
Speaker 3: Make the point that you've just made. It is not

246
00:10:19,320 --> 00:10:20,159
like normal.

247
00:10:19,879 --> 00:10:24,240
Speaker 2: Software where there's predictability about inputs improving leading to this

248
00:10:24,240 --> 00:10:27,960
particular output improving, and as the models iterate and improve,

249
00:10:28,480 --> 00:10:30,200
we don't quite know what's going to come out the

250
00:10:30,240 --> 00:10:31,960
other end. I think we would agree with that, which

251
00:10:31,960 --> 00:10:33,960
is why I think there is this bias that we

252
00:10:34,000 --> 00:10:35,720
need to get in there while the training runs are

253
00:10:35,720 --> 00:10:38,600
being done before the models are released, to understand what

254
00:10:38,759 --> 00:10:41,679
is this new iteration broad about in terms of capability,

255
00:10:41,679 --> 00:10:44,240
which it sounds like you would agree with. I was

256
00:10:44,279 --> 00:10:46,440
going to shift goes a little bit on You know,

257
00:10:46,440 --> 00:10:49,120
you've talked a lot about human consciousness human agency, which

258
00:10:49,159 --> 00:10:52,200
it actually might strike people as strange given that you

259
00:10:52,480 --> 00:10:55,360
are known for being such a brilliant innovator and technologist,

260
00:10:55,480 --> 00:10:57,840
but it's quite heartfelt when I hear you talk about

261
00:10:57,879 --> 00:11:00,879
it and the importance of maintaining that agency in technology

262
00:11:01,679 --> 00:11:03,200
and preserving human consciousness.

263
00:11:03,480 --> 00:11:04,480
Speaker 3: Now it kind of links.

264
00:11:04,679 --> 00:11:06,039
Speaker 2: The thing I was going to ask is when I

265
00:11:06,159 --> 00:11:08,080
do interviews or talk to people out and about in

266
00:11:08,120 --> 00:11:10,279
this job about AI, the thing that comes up most

267
00:11:10,320 --> 00:11:12,480
actually is it probably not so much the stuff we've

268
00:11:12,519 --> 00:11:15,440
been talking about, but jobs. It's what is AI mean

269
00:11:15,480 --> 00:11:17,799
for my job? Is it going to mean that I

270
00:11:17,840 --> 00:11:19,759
don't have a job, or my kids are not going

271
00:11:19,759 --> 00:11:23,559
to have a job. Now, answer as a policymaker.

272
00:11:22,960 --> 00:11:25,440
Speaker 3: As a leader is actually, AI is already.

273
00:11:25,159 --> 00:11:27,799
Speaker 2: Creating jobs, and you can see that in the companies

274
00:11:27,799 --> 00:11:30,840
that are starting. Also, the way it's being used is

275
00:11:31,000 --> 00:11:33,840
a little bit more as a co pilot necessarily versus

276
00:11:34,159 --> 00:11:36,799
replacing the person. There's still human agency, but it's helping

277
00:11:36,879 --> 00:11:39,440
you do your job better, which is a good thing.

278
00:11:39,919 --> 00:11:43,200
And as we've seen with technological revolutions in the past,

279
00:11:43,480 --> 00:11:45,159
clearly there's change in the labor market.

280
00:11:45,559 --> 00:11:46,440
Speaker 3: The amount of jobs.

281
00:11:46,720 --> 00:11:49,200
Speaker 2: I was quoting an MIT study today that they did

282
00:11:49,200 --> 00:11:51,279
a couple of years ago, something like sixty percent of

283
00:11:51,279 --> 00:11:53,879
the jobs at that moment didn't exist forty years ago.

284
00:11:53,960 --> 00:11:54,759
Speaker 3: So hard to predict.

285
00:11:54,799 --> 00:11:57,320
Speaker 2: And my job is to create an incredible education system,

286
00:11:57,440 --> 00:12:00,360
whether it's at school, whether it's retraining people at any

287
00:12:00,360 --> 00:12:02,200
point in their career, because ultimately, if we've got a

288
00:12:02,200 --> 00:12:04,960
skill population, they'll be able to keep up with the

289
00:12:05,000 --> 00:12:06,960
pace of change and have a good life. But you

290
00:12:07,000 --> 00:12:10,440
know that it's still a concern, and you know, what

291
00:12:10,480 --> 00:12:13,960
would your kind of observation be on AI and the

292
00:12:14,000 --> 00:12:16,240
impact of labor markets and people's jobs and how they

293
00:12:16,240 --> 00:12:18,639
should feel about that as they think about this.

294
00:12:18,919 --> 00:12:21,360
Speaker 1: Well, I think we are seeing the most disruptive force

295
00:12:21,759 --> 00:12:24,200
in history here, you know, where we have for the

296
00:12:24,200 --> 00:12:26,480
first time, we will have the first time something that

297
00:12:26,559 --> 00:12:29,240
is smarter than the smartest human and that I mean

298
00:12:29,360 --> 00:12:31,240
it's hard to say exactly what that moment is, but

299
00:12:31,440 --> 00:12:34,159
there will come a point where no job is needed.

300
00:12:34,360 --> 00:12:35,879
You can have a job if you want to have

301
00:12:35,919 --> 00:12:39,759
a job for sort of personal satisfaction, but the AI

302
00:12:39,919 --> 00:12:42,000
will be able to do everything. So I don't know

303
00:12:42,000 --> 00:12:44,440
if that makes people comfortable uncomfortable. You know. That's why

304
00:12:44,480 --> 00:12:46,559
I say, if you wish for a magic genie that

305
00:12:46,639 --> 00:12:48,600
gives you any wishes you want and there's no limit,

306
00:12:48,639 --> 00:12:50,679
you don't have those three limits three wish limit, not

307
00:12:50,720 --> 00:12:52,919
since you have as many wishes as you want. So

308
00:12:53,480 --> 00:12:55,480
it's both good and bad. One of the challenges in

309
00:12:55,480 --> 00:12:57,480
the future will be how do we find meaning in

310
00:12:57,519 --> 00:13:00,080
life if you have a magic genie that can do

311
00:13:00,159 --> 00:13:02,879
everything you want. I do think we it's hard you

312
00:13:02,919 --> 00:13:05,120
know when when? When this new technology, it tends to

313
00:13:05,600 --> 00:13:07,919
have usually follow an S curve. In this case, we're

314
00:13:08,080 --> 00:13:09,679
going to be on the exponential portion of the S

315
00:13:09,720 --> 00:13:12,440
curve for a long time and we have to ask

316
00:13:12,679 --> 00:13:14,759
for anything, and it won't be if we won't have

317
00:13:14,879 --> 00:13:17,759
universal basic income, we'll have universal high income. So in

318
00:13:17,600 --> 00:13:19,039
some in some sense, it will be somewhat of a

319
00:13:19,120 --> 00:13:21,759
leveler or an equalizer, because really I think everyone will

320
00:13:21,799 --> 00:13:23,960
have access to this magic gene and you're able to

321
00:13:24,000 --> 00:13:26,000
ask any question. It will certainly be good for education.

322
00:13:26,200 --> 00:13:28,720
It'll be the best tutor you could, the most patient

323
00:13:28,840 --> 00:13:31,159
tutor sit there all day, and there will be no

324
00:13:31,200 --> 00:13:33,399
shortage of goods and services, will be an age of abundance.

325
00:13:33,600 --> 00:13:36,679
I think if I'd recommend people read in Banks, the

326
00:13:36,919 --> 00:13:40,279
Banks culture books are probably the best in visioning. In fact,

327
00:13:40,279 --> 00:13:42,440
not probably, they're definitely by far the best envisioning of

328
00:13:42,480 --> 00:13:45,200
an a future. There's nothing even close. So I'd recommend,

329
00:13:45,440 --> 00:13:48,000
really recommend Banks. I'm a very big fan. All these

330
00:13:48,039 --> 00:13:50,399
books are good. Does not say which one all of them,

331
00:13:50,519 --> 00:13:53,399
So that's that'll give you a sense of what is

332
00:13:53,440 --> 00:13:57,759
a I guess a fairly utopian pro toopian future with

333
00:13:57,759 --> 00:13:58,960
with AI, Yeah, which is.

334
00:13:58,960 --> 00:14:01,720
Speaker 2: Good from an as you said, it's a universal high income,

335
00:14:01,720 --> 00:14:03,360
which is a nice phrase, and that's it's good for

336
00:14:03,440 --> 00:14:07,559
a materialistic sense of abundance. Actually, that kind of then

337
00:14:07,759 --> 00:14:08,679
leads to the question.

338
00:14:08,519 --> 00:14:10,559
Speaker 3: That you pose. Right, I'm someone who believes, you know,

339
00:14:10,600 --> 00:14:12,639
work gives you meaning. I write a lot about that.

340
00:14:12,799 --> 00:14:14,240
As you know, I think work is a good thing.

341
00:14:14,320 --> 00:14:16,440
Speaker 2: It, you know, gives people purpose in their lives. And

342
00:14:16,759 --> 00:14:19,639
if you then remove a large chunk of that, you know,

343
00:14:19,679 --> 00:14:20,440
what does that mean?

344
00:14:20,519 --> 00:14:21,519
Speaker 3: And where do you get that?

345
00:14:21,679 --> 00:14:23,480
Speaker 2: You know, where do you get that drive, that motivation,

346
00:14:23,559 --> 00:14:25,519
that purpose. I mean, you're talking about it. You work

347
00:14:25,559 --> 00:14:26,759
a lot of our you know.

348
00:14:27,440 --> 00:14:29,279
Speaker 1: As I was mentioning when we were talking earlier, I

349
00:14:29,320 --> 00:14:31,720
have to somewhat engage in the liberty suspension of disbelief

350
00:14:31,799 --> 00:14:34,039
because I'm putting so much blood, sweat and tears into

351
00:14:34,080 --> 00:14:37,159
a work project in burning the you know, three am oil.

352
00:14:37,679 --> 00:14:39,480
Then I'm like, wait, why am I doing this? I

353
00:14:39,480 --> 00:14:40,919
can just wait for the AI to do it. I'm

354
00:14:40,960 --> 00:14:43,840
just lashing myself for no reason. It must be a

355
00:14:43,879 --> 00:14:45,080
glutton for punishment.

356
00:14:45,120 --> 00:14:46,759
Speaker 2: Called demos and tell them to hurry up and then

357
00:14:46,799 --> 00:14:47,799
you can have a holiday.

358
00:14:47,879 --> 00:14:50,039
Speaker 3: Right, it's a plan. Yeah, No, it's a Look, it's

359
00:14:50,039 --> 00:14:50,440
a tricky.

360
00:14:50,559 --> 00:14:52,679
Speaker 2: It's a tricky thing because I think, you know, part

361
00:14:52,679 --> 00:14:54,639
of our job is to make sure that we can

362
00:14:55,000 --> 00:14:58,720
navigate to that very I think largely positive place and

363
00:14:58,799 --> 00:15:01,399
help people through it between now and then, because these

364
00:15:01,399 --> 00:15:03,440
things bring a lot about a change in the labor

365
00:15:03,480 --> 00:15:04,399
market as we've seen.

366
00:15:04,600 --> 00:15:06,720
Speaker 1: Yeah, I think it's probably is generally a good thing

367
00:15:06,759 --> 00:15:08,360
because you know, there are a lot of jobs that

368
00:15:08,399 --> 00:15:11,879
are uncomfortable or dangerous or which sort of tedious, and

369
00:15:12,399 --> 00:15:14,399
the computer will have no problem doing that. We're happy

370
00:15:14,440 --> 00:15:16,159
to do that all day long. So you know, it's

371
00:15:16,360 --> 00:15:17,879
fun to cook food, but it's not that fun to

372
00:15:17,919 --> 00:15:20,240
wash the dishes, but the computer is perfectly happy to

373
00:15:20,240 --> 00:15:22,840
watch the dishes. I guess there is. You know, we

374
00:15:22,879 --> 00:15:25,879
still have a sports like where where humans compete and

375
00:15:25,919 --> 00:15:29,039
like the Olympics, and obviously a machine can can go

376
00:15:29,159 --> 00:15:31,399
faster than any human, but we still have we saw

377
00:15:31,600 --> 00:15:34,159
humans race against each other and have all, you know,

378
00:15:34,399 --> 00:15:36,679
have these sports competitions against each other where even though

379
00:15:36,720 --> 00:15:39,080
the machines are better, which there's all I guess computingity,

380
00:15:39,080 --> 00:15:41,240
who can be the best human at something and people

381
00:15:41,240 --> 00:15:43,240
do find fulfillment in that. So I guess that's perhaps

382
00:15:43,320 --> 00:15:45,759
a good example of how even when machines are faster

383
00:15:45,799 --> 00:15:46,759
than us, stronger than us.

384
00:15:46,639 --> 00:15:47,639
Speaker 3: We still find a way.

385
00:15:47,559 --> 00:15:49,840
Speaker 1: We still enjoy competing against other humans, so at least

386
00:15:49,919 --> 00:15:50,679
who's the best human.

387
00:15:50,720 --> 00:15:52,639
Speaker 2: It's a good it's a good analogy, and we've been

388
00:15:52,639 --> 00:15:55,600
talking a lot about managing the risks. Just before we

389
00:15:55,639 --> 00:15:57,799
move on and finish on AI, is just talk a

390
00:15:57,840 --> 00:16:01,320
little bit about the opportunities. You know, you're engaged in

391
00:16:01,320 --> 00:16:04,320
lots of different companies. You're an obvious one, which is

392
00:16:04,320 --> 00:16:06,440
doing some exciting stuff. You touched on the thing that

393
00:16:06,440 --> 00:16:09,840
I'm probably most excited about, which is an education. Yeah,

394
00:16:09,919 --> 00:16:12,519
and I think many people will have seen Suth's.

395
00:16:12,399 --> 00:16:13,480
Speaker 3: Video from earlier this year.

396
00:16:13,559 --> 00:16:15,399
Speaker 2: Is ted talk about as you talked about, it's like

397
00:16:15,399 --> 00:16:16,200
a personal tutor.

398
00:16:16,279 --> 00:16:18,720
Speaker 1: Yeah, an amazing personal tutor, an.

399
00:16:18,600 --> 00:16:19,519
Speaker 3: Amazing personal tutor.

400
00:16:19,559 --> 00:16:22,320
Speaker 2: And we know the difference in learning having that personal

401
00:16:22,559 --> 00:16:25,000
his tutor is incredible compared to class from learning. So

402
00:16:25,000 --> 00:16:27,240
if you can have every child have a personal tutor

403
00:16:27,519 --> 00:16:30,200
specifically for them that then just evolves with them over time,

404
00:16:30,320 --> 00:16:32,320
that could be extraordinary. So that you know, for me,

405
00:16:32,360 --> 00:16:34,440
I look at it, I think, gosh, that is within

406
00:16:34,519 --> 00:16:36,720
reach at this point, and that's one of the benefits

407
00:16:36,759 --> 00:16:39,639
I'm most excited about. When you look at the landscape

408
00:16:39,679 --> 00:16:41,759
of things that you see as possible, what is it

409
00:16:41,840 --> 00:16:44,559
that you know you are particularly excited about.

410
00:16:44,840 --> 00:16:47,679
Speaker 1: I think certainly a tutors are going to be the amazing,

411
00:16:47,919 --> 00:16:51,679
perhaps already are. I think there's also perhaps companionship, which

412
00:16:51,679 --> 00:16:53,879
may seem odd because how can the computer really be

413
00:16:54,279 --> 00:16:56,879
your friend? But if you haven't that has memory, you know,

414
00:16:56,919 --> 00:16:59,960
and remembers all of your interactions and has read every

415
00:17:00,240 --> 00:17:01,960
you're going to actually like give it permission to read

416
00:17:02,000 --> 00:17:03,759
everything you've ever done, so it really will know you

417
00:17:03,840 --> 00:17:06,400
better than anyone, perhaps even yourself. And where you can

418
00:17:06,440 --> 00:17:08,799
talk to it every day and those conversations spoiled upon

419
00:17:08,839 --> 00:17:11,559
each other, you will actually have a great friend, as

420
00:17:11,680 --> 00:17:13,480
long as that friend can stay your friend and not

421
00:17:13,480 --> 00:17:15,799
get turned off or something. Don't turn off my friends.

422
00:17:16,119 --> 00:17:17,759
But I think that will actually be a real thing.

423
00:17:18,440 --> 00:17:20,839
And I one of my sons is sort of has

424
00:17:20,880 --> 00:17:23,960
some learning disabilities and has trouble making friends actually, and

425
00:17:23,960 --> 00:17:26,799
I was like, well, you know he ai friend would

426
00:17:26,799 --> 00:17:27,720
actually be a great friend.

427
00:17:28,519 --> 00:17:31,519
Speaker 2: Okay, you know that was a surprising answer. That's actually

428
00:17:31,599 --> 00:17:34,440
it's worth worth reflecting on that. That's really interesting. I

429
00:17:34,480 --> 00:17:38,440
mean we're already seeing it actually as we deliver psychotherapy anyway,

430
00:17:38,839 --> 00:17:42,319
now doing far more by digitally and by telephone to

431
00:17:42,319 --> 00:17:44,039
people and it's making a huge difference, and you can

432
00:17:44,079 --> 00:17:46,359
see a world in which actually, you know, AI can

433
00:17:46,440 --> 00:17:50,599
provide that social benefit to people. Just a quick question

434
00:17:50,720 --> 00:17:53,319
on X and then we should open it up to everybody.

435
00:17:54,119 --> 00:17:56,599
You made a change when in one of the made

436
00:17:56,640 --> 00:17:58,440
many changes, but one of the one.

437
00:17:58,359 --> 00:18:01,559
Speaker 1: Of the change that letter. Yet thing about it, you really.

438
00:18:01,319 --> 00:18:03,839
Speaker 2: Do, really do one of the changes which you know

439
00:18:04,000 --> 00:18:05,799
kind of you know, it goes into the space that

440
00:18:05,880 --> 00:18:08,240
you know we have to operate in. And this balance

441
00:18:08,279 --> 00:18:11,720
between free speech and moderation is you know, we grapple

442
00:18:11,759 --> 00:18:14,640
with as politicians. You were grappling with your own version

443
00:18:14,640 --> 00:18:17,000
of that, and you moved away from a kind of

444
00:18:17,119 --> 00:18:20,759
manual human way of doing it the moderation.

445
00:18:20,440 --> 00:18:21,519
Speaker 3: To the community that.

446
00:18:22,000 --> 00:18:24,640
Speaker 2: Yeah, and I think that's it was an interesting change, right,

447
00:18:24,680 --> 00:18:27,200
It's not what everyone else has done. It would be

448
00:18:27,240 --> 00:18:29,400
good you know what's what was the reasoning behind that?

449
00:18:29,440 --> 00:18:31,079
And why do you think that is a better way

450
00:18:31,119 --> 00:18:31,519
to do that?

451
00:18:31,759 --> 00:18:33,440
Speaker 1: Yeah. Part of the problem is if you if you

452
00:18:33,559 --> 00:18:36,720
empower people as sensors, then well there's going to be

453
00:18:36,720 --> 00:18:39,079
some amount of bias that they have and then whoever

454
00:18:39,319 --> 00:18:42,440
points the sensors is effectively in control of information. So

455
00:18:42,480 --> 00:18:44,559
then the idea behind community notice, well, how do we

456
00:18:44,759 --> 00:18:47,200
have a consensus driven I mean, so it's not really

457
00:18:47,240 --> 00:18:50,000
censoring it, but consensus driven approach to truth. How do

458
00:18:50,039 --> 00:18:53,200
we or how do we make things the least amount untrue?

459
00:18:53,319 --> 00:18:56,480
Could you say, like you can't perhaps get to pure truth,

460
00:18:56,519 --> 00:18:59,319
but you can aspire to be more truthful. So the

461
00:18:59,359 --> 00:19:02,079
thing about comurnity notes is it doesn't actually delete anything.

462
00:19:02,279 --> 00:19:04,559
It simply adds context. Now that context could be this

463
00:19:04,599 --> 00:19:07,759
thing is untrue for the following reasons. But importantly with

464
00:19:07,759 --> 00:19:10,920
community notes, everything is open source actually, so you can

465
00:19:10,960 --> 00:19:13,680
see the software, every line of the software, you can

466
00:19:13,720 --> 00:19:16,319
see all of the data that went into a community note,

467
00:19:16,400 --> 00:19:19,519
and you can independently create that community note. So if

468
00:19:19,519 --> 00:19:21,799
you've got if you see manipulations of data, you can

469
00:19:21,799 --> 00:19:24,480
actually highlight that and say, well this this there appears

470
00:19:24,480 --> 00:19:26,839
to be some gaming of the system, and you can

471
00:19:26,880 --> 00:19:30,880
suggest improvement. So it's maximum transparency, which.

472
00:19:30,720 --> 00:19:32,640
Speaker 2: Is I think combined with the kind of wisdom of

473
00:19:32,680 --> 00:19:34,920
the crowds and trying to get to a better answer.

474
00:19:35,240 --> 00:19:37,359
Speaker 1: And really one of the key elements of community notes

475
00:19:37,400 --> 00:19:38,960
is that in order for a note to be shown,

476
00:19:39,079 --> 00:19:42,240
people who have historically disagreed must agree, and there is

477
00:19:42,240 --> 00:19:45,039
a bit of AI usage here. So this will populate

478
00:19:45,079 --> 00:19:48,720
a parameter space around each contributor to the community notes,

479
00:19:49,039 --> 00:19:52,559
and then a parameter space so everyone's got basically these

480
00:19:52,640 --> 00:19:55,559
vectors associated with them, which so it's not as simple

481
00:19:55,599 --> 00:19:58,279
as right or left, it's saying, it's more it's several

482
00:19:58,400 --> 00:20:01,599
hundred vectors. That that because things are more complicated than

483
00:20:01,599 --> 00:20:03,960
something right, right or left, and and they will we'll

484
00:20:03,960 --> 00:20:06,839
do sort of inverse correlation, say like, okay, these people

485
00:20:06,960 --> 00:20:09,559
generally disagree, but they agree about this note. Okay, so

486
00:20:09,640 --> 00:20:13,480
then that so then that that that gives the note credibility. Okay, yeah,

487
00:20:13,799 --> 00:20:15,839
that's that's the core of it. And it's working quite well.

488
00:20:16,359 --> 00:20:18,680
I get to see a note actually be present for

489
00:20:18,720 --> 00:20:21,039
more than a few hours, but that that is incorrect.

490
00:20:21,079 --> 00:20:23,200
So the batting average is extremely good. And when I

491
00:20:23,240 --> 00:20:25,880
ask people say, oh, they're worried about community notes sort

492
00:20:25,880 --> 00:20:28,079
of being disinformation, like send me one, and then they can't,

493
00:20:28,359 --> 00:20:29,960
So so I think it's I think it's quite good.

494
00:20:29,960 --> 00:20:33,240
I mean, the general aspiration is with the X platform

495
00:20:33,400 --> 00:20:36,279
is to inform and entertain the public and to be

496
00:20:36,319 --> 00:20:39,000
as accurate as possible. And as truthful as possible, even

497
00:20:39,039 --> 00:20:41,559
if someone doesn't like the truth. You know, people don't

498
00:20:41,559 --> 00:20:44,440
always like the truth at always, but that's the aspiration.

499
00:20:44,599 --> 00:20:47,400
And I think if we are, if we stay true

500
00:20:47,400 --> 00:20:49,440
to the truth, then I think we'll find that people

501
00:20:49,799 --> 00:20:52,960
use the system to learn what is going on and

502
00:20:53,079 --> 00:20:56,319
to I think actually truth pays, so I think it'll

503
00:20:56,319 --> 00:20:58,519
be well, I mean, assuming you don't want to engage

504
00:20:58,559 --> 00:21:01,319
in self delusion, then I think it's the smart move.

505
00:21:01,519 --> 00:21:04,440
Speaker 3: Excellent, very helpful. Right, let's open it up to all

506
00:21:04,480 --> 00:21:05,440
our guests here, and.

507
00:21:05,359 --> 00:21:07,799
Speaker 2: We've got some microphones that'll complete I find you have,

508
00:21:07,880 --> 00:21:09,519
we got yes, go for it.

509
00:21:09,799 --> 00:21:10,839
Speaker 3: Thank you, good evening.

510
00:21:11,039 --> 00:21:13,400
Speaker 4: Alice Pentink from an entrepreneur. First, thank you for a

511
00:21:13,400 --> 00:21:16,319
fascinating conversation. I suppose a question for each of you,

512
00:21:16,440 --> 00:21:19,279
Prime Minister. The UK has some of the best universities

513
00:21:19,279 --> 00:21:21,519
in the world, We have the talent. What will it

514
00:21:21,559 --> 00:21:24,759
take for the UK to be a real breeding unicorn companies?

515
00:21:25,680 --> 00:21:27,880
Being a founder in the UK is still a non

516
00:21:27,960 --> 00:21:31,279
obvious career choice for the most exceptional technical talent. What

517
00:21:31,279 --> 00:21:32,920
are the cultural elements that we need to put into

518
00:21:32,960 --> 00:21:33,960
place to change this?

519
00:21:34,119 --> 00:21:35,400
Speaker 3: Thank you? You want to go first?

520
00:21:35,440 --> 00:21:37,279
Speaker 1: Go for it. Sure, well, you're right that there are

521
00:21:37,640 --> 00:21:41,640
cultural elements where you know, the culture should celebrate creating

522
00:21:41,720 --> 00:21:45,440
new companies and there should be a bias towards supporting

523
00:21:45,680 --> 00:21:48,119
small companies because the ones that need nurturing. The larger

524
00:21:48,119 --> 00:21:51,079
companies really don't need nurturing. So you know, just you

525
00:21:51,079 --> 00:21:52,680
can think of it's sort of like a garden. If

526
00:21:52,720 --> 00:21:55,039
it's a little sprout that needs nurturing, if it's a

527
00:21:55,119 --> 00:21:57,599
mighty ogre, doesn't need quite as much. So I think

528
00:21:57,720 --> 00:22:00,440
that is a mindset change that is important. But I

529
00:22:00,440 --> 00:22:04,359
should mention that London is you know, London and San

530
00:22:04,400 --> 00:22:07,039
Francisco or the Bay Area are really the two centers

531
00:22:07,079 --> 00:22:09,759
for AI. So that so London is actually doing very

532
00:22:09,799 --> 00:22:12,440
well on that front. That the two most I said,

533
00:22:12,440 --> 00:22:15,519
the two leading locations on Earth. You know, San Francisco's

534
00:22:15,559 --> 00:22:18,000
probably head of London, but London's really very strong. But

535
00:22:18,079 --> 00:22:21,319
London Area greater London home counties, I guess. So I'm

536
00:22:21,319 --> 00:22:23,559
just saying objectively this is the case. But you do

537
00:22:23,640 --> 00:22:25,680
need that. You need the infrastructure, you need the landlords

538
00:22:25,680 --> 00:22:28,200
who are willing to rent to new companies. You need

539
00:22:28,319 --> 00:22:30,200
more firms and accounts that are willing to support new

540
00:22:30,240 --> 00:22:33,400
companies and it's generally it is a mindset change, and

541
00:22:33,440 --> 00:22:35,079
I think some of that is happening, but I think

542
00:22:35,440 --> 00:22:37,720
really it's just culturally people need to decide this is

543
00:22:37,720 --> 00:22:38,680
a good thing. Yeah.

544
00:22:38,839 --> 00:22:40,759
Speaker 2: Yeah, no, Actually, well thanks for what you said about

545
00:22:40,799 --> 00:22:42,480
the UK. It's something that we work hard on. Lots

546
00:22:42,480 --> 00:22:44,559
of people in the room are part of what makes

547
00:22:44,559 --> 00:22:48,279
this a fabulous place for companies, including Alice. So that's

548
00:22:48,279 --> 00:22:50,079
what i'd say is my job is to get all

549
00:22:50,119 --> 00:22:52,400
the you know they're nuts and bolts right, make sure

550
00:22:52,440 --> 00:22:54,640
that all of you are starting companies can raise the

551
00:22:54,799 --> 00:22:57,400
capital that you need everything from. You know, you'll see

552
00:22:57,400 --> 00:23:01,000
funding with our incredible you know EI tax reliefs all

553
00:23:01,000 --> 00:23:01,920
the way through.

554
00:23:01,720 --> 00:23:03,839
Speaker 3: To your late stage rounds.

555
00:23:03,880 --> 00:23:05,960
Speaker 2: And we need reform of our pension funds and the

556
00:23:06,039 --> 00:23:09,559
Chancellor's got a bunch of incredible reforms to unlock capital

557
00:23:09,680 --> 00:23:12,079
from all the people who have it and deployed into

558
00:23:12,119 --> 00:23:12,839
growth equity.

559
00:23:12,920 --> 00:23:13,079
Speaker 1: Right.

560
00:23:13,079 --> 00:23:14,960
Speaker 2: That is a work in progress. We're not there yet,

561
00:23:15,000 --> 00:23:17,720
but I think we're making good progress. We need talent,

562
00:23:17,839 --> 00:23:20,759
we need people, so that means an education system that

563
00:23:20,920 --> 00:23:22,480
prioritizes the things that matter.

564
00:23:22,559 --> 00:23:23,720
Speaker 3: And you've seen my reforms.

565
00:23:23,759 --> 00:23:26,000
Speaker 2: I go on about more maths, more maths, more maths,

566
00:23:26,279 --> 00:23:28,440
But I think that it's important but also attracting the

567
00:23:28,440 --> 00:23:30,519
best and the brightest fit. If you look at our

568
00:23:30,640 --> 00:23:33,279
fastest growing companies in this country, and I think it's

569
00:23:33,279 --> 00:23:36,000
probably the same in the US, over half of them

570
00:23:36,119 --> 00:23:39,079
have a non British founder, right, And so that tells

571
00:23:39,119 --> 00:23:40,920
you we've got to be a place that is open

572
00:23:41,000 --> 00:23:44,279
to the world's best and brightest entrepreneurial talent. So the

573
00:23:44,359 --> 00:23:46,680
visa regime that we've put in place, I think, does

574
00:23:46,720 --> 00:23:48,720
that makes it easy for those people to come here.

575
00:23:48,799 --> 00:23:50,720
And then actually it's the thing that we've spent the

576
00:23:50,759 --> 00:23:53,599
beginning of the session talking about the regulation right, making

577
00:23:53,640 --> 00:23:56,759
sure that we've got a regulatory system that's pro innovation.

578
00:23:57,240 --> 00:23:57,319
Speaker 1: That.

579
00:23:57,440 --> 00:23:59,839
Speaker 2: Yeah, of course we always need guardrails on the things

580
00:24:00,160 --> 00:24:02,240
that will worry us, but we've got to create a

581
00:24:02,279 --> 00:24:04,720
space for people to innovate and do different things.

582
00:24:04,880 --> 00:24:06,079
Speaker 3: Now, those are all my jobs.

583
00:24:06,279 --> 00:24:08,400
Speaker 2: The thing that is tougher is the thing that Elon

584
00:24:08,440 --> 00:24:10,400
talked about, which is culture. Right, It's how do you

585
00:24:10,839 --> 00:24:14,400
transpose that culture from places like Silicon Valley across the

586
00:24:14,440 --> 00:24:18,400
world where people are unafraid to give up the security

587
00:24:18,440 --> 00:24:21,519
of a regular paycheck to go and start something and

588
00:24:21,599 --> 00:24:22,720
be comfortable with failure.

589
00:24:22,759 --> 00:24:24,920
Speaker 3: You talk about that a lot. I think you talked about.

590
00:24:24,680 --> 00:24:27,480
Speaker 2: It more when you were playing games, right, But you've

591
00:24:27,519 --> 00:24:30,279
got to be comfortable failing and knowing that that's just

592
00:24:30,319 --> 00:24:32,839
part of the process. And that is a tricky cultural

593
00:24:32,880 --> 00:24:35,799
thing to do overnight, but it's an important part of

594
00:24:35,880 --> 00:24:36,720
I think creating that.

595
00:24:36,720 --> 00:24:37,480
Speaker 3: Kind of environment.

596
00:24:37,640 --> 00:24:40,039
Speaker 1: Yeah, if you don't succeed with your first startup, it

597
00:24:40,039 --> 00:24:43,240
shouldn't be a sort of a catastrophic career ending exactly thing,

598
00:24:43,279 --> 00:24:46,000
it should be you know, well good I think should

599
00:24:46,039 --> 00:24:47,960
like should be like, well, you know, you gave it

600
00:24:48,000 --> 00:24:49,880
a good shot, you know, in now try.

601
00:24:49,759 --> 00:24:50,680
Speaker 3: Again and exactly.

602
00:24:50,880 --> 00:24:53,720
Speaker 1: Yeah. And it's so one thing I'm going to mention is,

603
00:24:53,720 --> 00:24:55,559
like I've seen, creating a company is sort of a

604
00:24:55,640 --> 00:24:59,000
high risk, high reward situation. But I don't know quite

605
00:24:59,000 --> 00:25:00,839
what the how works. In the UK I think is

606
00:25:00,880 --> 00:25:03,880
probably better than contenttle Europe, but I'm not sure how

607
00:25:03,920 --> 00:25:06,079
does the new give it. But if somebody's basically going

608
00:25:06,079 --> 00:25:09,079
to risk their life savings and with the beast majority

609
00:25:09,079 --> 00:25:11,720
of startups fail, so I mean you hear about the

610
00:25:11,720 --> 00:25:15,400
startups that succeed, but most companies are most startups consist

611
00:25:15,480 --> 00:25:18,680
of you know, a massive amount of work followed by failure.

612
00:25:18,720 --> 00:25:21,640
But that's actually most companies, and so it's a high risk,

613
00:25:21,680 --> 00:25:23,960
high reward and so the higher reward part does need

614
00:25:24,000 --> 00:25:25,519
to be there for it to make sense.

615
00:25:25,680 --> 00:25:28,200
Speaker 2: I think that was a very soft pitch for a

616
00:25:28,279 --> 00:25:31,359
tax policy. I mean, but actually I can tell you so,

617
00:25:31,519 --> 00:25:34,440
like AI agree, and we have so we have, I

618
00:25:34,440 --> 00:25:37,799
think relative to certainly European countries, but certainly the US,

619
00:25:38,039 --> 00:25:41,480
definitely California a much lower rate of capital gains tax right,

620
00:25:41,480 --> 00:25:44,400
So for those people who are risking and growing something like,

621
00:25:44,440 --> 00:25:46,440
we think the reward should be there at the end.

622
00:25:46,480 --> 00:25:48,440
Speaker 3: So it's twenty percent capital gains tax right.

623
00:25:48,720 --> 00:25:50,519
Speaker 2: And on stock options, I don't know if we've got

624
00:25:50,559 --> 00:25:53,160
anyone from Index Ventures in the room.

625
00:25:53,319 --> 00:25:55,599
Speaker 3: So you know, Index one of our bleeding VC funds here.

626
00:25:55,680 --> 00:25:59,559
Speaker 2: Okay, they do a regular report looking at most countries

627
00:25:59,599 --> 00:26:02,880
tax of stock options. And you know, when I was

628
00:26:03,039 --> 00:26:05,960
a Chancellor of Treasury sexualy equivalent, you know, we were

629
00:26:06,079 --> 00:26:08,200
I think down we were pretty good, but we were

630
00:26:08,200 --> 00:26:10,319
fourth or fifth, and I said we need to for

631
00:26:10,359 --> 00:26:12,440
exactly the reason that you mentioned. I was like, this

632
00:26:12,480 --> 00:26:14,119
has got to be the best place for renovative We

633
00:26:14,160 --> 00:26:15,759
need to move that up and I think in the

634
00:26:15,799 --> 00:26:18,039
last iteration of that report we had because of the

635
00:26:18,119 --> 00:26:20,559
changes that Jeremy and I had made, we have moved

636
00:26:20,640 --> 00:26:23,400
up to I think second from memory. So hopefully they

637
00:26:23,400 --> 00:26:26,240
actually give you and everyone else some comfort that we recognize.

638
00:26:26,279 --> 00:26:29,359
That's important because when people work hard and risk things, yeah, they.

639
00:26:29,319 --> 00:26:31,599
Speaker 3: Should be able to enjoy the rewards and award.

640
00:26:31,720 --> 00:26:31,920
Speaker 1: Yeah.

641
00:26:31,920 --> 00:26:33,680
Speaker 2: And I think we have a We very much have

642
00:26:33,680 --> 00:26:36,079
a tax system that supports that. And those are the

643
00:26:36,119 --> 00:26:38,000
values that you know, I believe in, and I think

644
00:26:38,000 --> 00:26:40,480
most of us in this room probably do as well. Right,

645
00:26:40,640 --> 00:26:42,799
next next question, I've got seven front of me, and

646
00:26:42,839 --> 00:26:43,559
then I'll come over here.

647
00:26:43,640 --> 00:26:44,920
Speaker 1: Gone, thanks very much.

648
00:26:45,240 --> 00:26:48,519
Speaker 5: We've talked about some really big ideas, global changing ideas.

649
00:26:48,720 --> 00:26:51,240
I'm really interested, particularly in the context of creation of

650
00:26:51,319 --> 00:26:53,920
science and technology superhubs and so on. How does that

651
00:26:53,960 --> 00:26:58,519
map onto the everyday lives of people living in say Austin, Texas,

652
00:26:58,519 --> 00:27:00,759
to choose not aroundin the more in my case Nottingham,

653
00:27:00,880 --> 00:27:03,960
East Midlands. What is how do you see that evolving

654
00:27:03,960 --> 00:27:05,160
for people you know every.

655
00:27:05,000 --> 00:27:07,920
Speaker 2: Day sort of every day effective AI for context, Elon

656
00:27:08,000 --> 00:27:11,759
Sis said runs are equivalent of CDs, right or Walgreens.

657
00:27:11,799 --> 00:27:14,200
So you know, as I visited, right, so he's got

658
00:27:14,279 --> 00:27:16,240
millions of people coming in the shops every day, and

659
00:27:16,680 --> 00:27:18,480
it's making sure how do we make this relevant? I

660
00:27:18,480 --> 00:27:21,559
think that your question how is this relevant to that person?

661
00:27:21,839 --> 00:27:22,039
Speaker 3: You know?

662
00:27:22,039 --> 00:27:24,000
Speaker 2: Maybe actually let me go I'll go first on that

663
00:27:24,039 --> 00:27:25,839
because I think it's a fair point. I was just

664
00:27:25,839 --> 00:27:27,599
going over with the team a couple of things that

665
00:27:27,720 --> 00:27:29,680
we're doing because I was saying, how are we doing

666
00:27:29,799 --> 00:27:32,519
AI right now that it's making a difference to people's lives.

667
00:27:32,960 --> 00:27:35,200
And we have this thing called gov dot which is

668
00:27:35,480 --> 00:27:38,200
which actually when we when it happened several years ago,

669
00:27:38,319 --> 00:27:41,519
was a pioneering thing all the government information together on

670
00:27:41,519 --> 00:27:45,119
one website and so you need to get a driving license, passport,

671
00:27:45,160 --> 00:27:48,559
any interaction with government. It was centralized in a very easy,

672
00:27:48,839 --> 00:27:50,279
relatively easy to use way.

673
00:27:50,319 --> 00:27:53,640
Speaker 3: Better than most. So we're about to deploy AI across

674
00:27:53,640 --> 00:27:54,279
that platform.

675
00:27:54,319 --> 00:27:56,799
Speaker 2: So that is something that I think you know, several

676
00:27:56,839 --> 00:27:59,480
million people a day use, right, So a large chunk

677
00:27:59,519 --> 00:28:02,640
of the popular is interacting with gov dot UK every single.

678
00:28:02,440 --> 00:28:04,799
Speaker 3: Day to do all these day to day tasks.

679
00:28:04,839 --> 00:28:06,960
Speaker 2: Right, every one of your customers is doing all those things,

680
00:28:07,240 --> 00:28:10,440
and so we're about to deploy AI into that to

681
00:28:10,519 --> 00:28:13,920
make that whole process so much easier, because you know,

682
00:28:13,960 --> 00:28:16,319
some people will be like, look, well I'm currently here

683
00:28:16,519 --> 00:28:19,599
and I've lost my passport and my flights in five hours.

684
00:28:19,680 --> 00:28:21,759
At the moment, that would require you know how many

685
00:28:21,799 --> 00:28:24,200
steps to figure out what you do. Yeah, Actually, when

686
00:28:24,240 --> 00:28:25,880
we deploy the AI, it should be that you could

687
00:28:25,920 --> 00:28:28,359
just literally say that and boom, this is.

688
00:28:28,279 --> 00:28:29,279
Speaker 3: What we're going to do, walk.

689
00:28:29,079 --> 00:28:31,400
Speaker 2: You through it, and that's going to benefit millions and

690
00:28:31,440 --> 00:28:33,559
millions of people every single day, right, Because that's a

691
00:28:33,640 --> 00:28:36,599
very practical way in my seat, that I can start

692
00:28:36,680 --> 00:28:38,839
using this technology to help people in their day to

693
00:28:38,920 --> 00:28:42,359
day lives, not just healthcare discoveries and everything else that

694
00:28:42,400 --> 00:28:44,640
we're also doing. But I thought that's quite a powerful

695
00:28:44,640 --> 00:28:49,039
demonstration of literally your day to day customer seeing actually

696
00:28:49,079 --> 00:28:50,599
their just day to day life get a little bit

697
00:28:50,640 --> 00:28:52,920
easier because of something that you know, Elon Demis and

698
00:28:52,960 --> 00:28:54,200
others in this room have helped create.

699
00:28:54,359 --> 00:28:57,519
Speaker 1: Yeah, exactly. The most immediate thing is just being able

700
00:28:57,519 --> 00:28:59,799
to ask, like having a very smart friend that you

701
00:28:59,839 --> 00:29:02,839
can ask anything, you know, ask how to make something,

702
00:29:02,960 --> 00:29:05,279
how to solve any problem and it'll tell you so.

703
00:29:05,720 --> 00:29:08,519
And obviously companies are going to adopt this, so I

704
00:29:08,519 --> 00:29:10,559
think maybe you'll have much better customer service. I guess

705
00:29:10,680 --> 00:29:12,720
essentially that will probably be the first thing you notice.

706
00:29:12,759 --> 00:29:16,880
And then we talked about education. So having a tutor

707
00:29:16,880 --> 00:29:20,160
so that you're trying to understand a subject, like having

708
00:29:20,640 --> 00:29:23,240
a phenomenal tutor on any subject. Is that that's really

709
00:29:23,319 --> 00:29:26,480
pretty much there already almost. I mean we need to obviously,

710
00:29:26,599 --> 00:29:29,480
I need to stop hallucinating before you know it can't

711
00:29:29,480 --> 00:29:31,720
give you. I mean that we still have a little

712
00:29:31,720 --> 00:29:33,319
bit of a problem where it can give you an

713
00:29:33,359 --> 00:29:36,319
answer that's confidently wrong with great grammar, you know, bullet

714
00:29:36,319 --> 00:29:38,839
points and everything in citations. It was not real. So

715
00:29:38,880 --> 00:29:40,440
that's be okay, we need to make sure it's not

716
00:29:40,480 --> 00:29:43,359
it's not it's not giving you confidently wrong tutor answers.

717
00:29:43,440 --> 00:29:45,839
But but that's going to happen pretty quickly where it

718
00:29:46,079 --> 00:29:47,119
is actually correct.

719
00:29:47,440 --> 00:29:50,640
Speaker 2: Yeah, say for any parent who was homeschooling and realizing

720
00:29:50,720 --> 00:29:53,039
what their kids needed to be helped with, yeah, that

721
00:29:53,079 --> 00:29:54,400
will come as an enormous relief.

722
00:29:54,440 --> 00:29:55,279
Speaker 1: I think they're very good.

723
00:29:55,359 --> 00:29:57,720
Speaker 2: Right, have we got let's go questions over here, who

724
00:29:57,799 --> 00:30:01,319
have got only microphones or brand that perfect.

725
00:30:01,359 --> 00:30:04,759
Speaker 6: Brent Herman, So you know you've spoken eloquently about abundance

726
00:30:04,759 --> 00:30:06,960
in the Age of abundance, so it feels obviously with

727
00:30:07,039 --> 00:30:11,160
AI it's everything everywhere, all at once, but with robots,

728
00:30:11,359 --> 00:30:13,160
and to get the age of abundance will need a

729
00:30:13,160 --> 00:30:15,039
lot of robots. I know you're working out on robots

730
00:30:15,079 --> 00:30:17,680
as well. Are there sort of constraints that we should

731
00:30:17,680 --> 00:30:19,920
think of our politician should be thinking of that we

732
00:30:20,000 --> 00:30:23,079
might get one country might get heavily behind in robots

733
00:30:23,119 --> 00:30:24,240
that can do all these things and end to the

734
00:30:24,279 --> 00:30:26,519
age of bunds and therefore be it a strategic disadvantage.

735
00:30:26,599 --> 00:30:29,119
Speaker 1: Well, really, anything that can be actuated by a computer

736
00:30:29,519 --> 00:30:32,720
is effectively a robot. So you can think of frankly

737
00:30:32,720 --> 00:30:36,519
Tesla cars or robots on wheels. Anything that's connected to

738
00:30:36,559 --> 00:30:40,519
the Internet is effectively an endpoint actuator for artificial intelligence.

739
00:30:40,720 --> 00:30:43,400
So you've got Bust and Dynamics. Obviously they've been making

740
00:30:43,519 --> 00:30:45,960
impressive robots for a while. I think they are at

741
00:30:45,960 --> 00:30:48,759
this point mostly owned by Hundai, So I guess I

742
00:30:48,799 --> 00:30:51,519
is probably gonna make robots that are humanoid and some

743
00:30:52,200 --> 00:30:54,160
interesting shapes that I wasn't just beating like the one

744
00:30:54,160 --> 00:30:56,599
that looks like a has wheels and looks sort of

745
00:30:56,599 --> 00:30:58,440
like a kangaroo on wheels. I'm sure what that is,

746
00:30:58,480 --> 00:31:00,640
but it's a little demanded, frankly, but there's going to

747
00:31:00,680 --> 00:31:03,079
be all sorts of robots. You've got the company Dison,

748
00:31:03,440 --> 00:31:07,200
in which I think there's some pretty impressive things, and

749
00:31:07,279 --> 00:31:09,440
I think the UK will not be behind. Actually on

750
00:31:09,799 --> 00:31:12,640
that front, the UK also has ARM, which is really

751
00:31:12,680 --> 00:31:15,319
the best once one of the best peraps, the best

752
00:31:15,480 --> 00:31:18,400
in chip design in the world. Tesla uses a lot

753
00:31:18,400 --> 00:31:22,000
of ARM technology almost every winds actually, so I think

754
00:31:22,000 --> 00:31:24,319
the UK is in a strong position. Germany obviously makes

755
00:31:24,359 --> 00:31:26,960
a lot of robots, industrial robots. I mean, I think

756
00:31:27,119 --> 00:31:30,279
generally countries that make robots of any kind, even if

757
00:31:30,319 --> 00:31:32,960
they seem somewhat conventional, will be fine. I do think

758
00:31:33,039 --> 00:31:38,160
there is a safety concern, especially with humanoid robots, because

759
00:31:38,680 --> 00:31:41,039
at least the car can't chase you into this building,

760
00:31:41,160 --> 00:31:43,240
not very easily, you know, or chase you over a tree,

761
00:31:43,359 --> 00:31:44,960
or you know, you can sort of run up a

762
00:31:44,960 --> 00:31:46,559
flight of stairs and get away from a Tesla. I

763
00:31:46,559 --> 00:31:48,440
think it's a Stephen King movie about that pere car

764
00:31:48,480 --> 00:31:51,000
gets possessed so, but if you have a humanoid robot,

765
00:31:51,039 --> 00:31:53,680
it can basically chase you anywhere. So I think we

766
00:31:53,720 --> 00:31:57,480
should have some kind of hardwired local cutoff that you

767
00:31:57,519 --> 00:31:59,839
can't update from the internet. So anything that could be

768
00:32:00,000 --> 00:32:02,559
after updated from the internet obviously could be overridden. But

769
00:32:02,599 --> 00:32:04,920
if you have a local sort of off switch where

770
00:32:04,920 --> 00:32:06,880
you have to say a keyword or something, and then

771
00:32:06,920 --> 00:32:09,240
that puts the robot into a safe state, it's some

772
00:32:09,319 --> 00:32:12,920
kind of localized safe state ability and off switch, you know,

773
00:32:13,039 --> 00:32:14,599
where you don't have to get too close to the robot.

774
00:32:14,839 --> 00:32:16,680
I don't know. So we've got millions of these things

775
00:32:16,720 --> 00:32:17,440
going over the place.

776
00:32:17,599 --> 00:32:20,400
Speaker 3: You're not selling it, just you know, I know.

777
00:32:20,440 --> 00:32:22,400
Speaker 1: I'm saying this is something we should be quite concerned

778
00:32:22,440 --> 00:32:25,039
about because Robert can follow you anywhere. Then you know

779
00:32:25,279 --> 00:32:27,759
what if they just one day get a software update

780
00:32:27,880 --> 00:32:30,000
and they're not so friendly anymore. We've got a James

781
00:32:30,079 --> 00:32:31,039
Cameron movie on our house.

782
00:32:31,079 --> 00:32:33,160
Speaker 2: It's actually that's it's funny you're saying that, because in

783
00:32:33,200 --> 00:32:36,799
our session that we had today, I just would say

784
00:32:36,839 --> 00:32:39,359
who they made exactly the same point right then, And

785
00:32:39,440 --> 00:32:42,119
so we're talking about they're talking about movies actually about

786
00:32:42,119 --> 00:32:44,799
mentioning James Cameron. They're talking about James cameraon movies the same.

787
00:32:44,920 --> 00:32:47,440
If you think about it, it's not just those movies, but

788
00:32:47,559 --> 00:32:52,680
any of these movies, trains, subways, metros. They said, all

789
00:32:52,759 --> 00:32:55,160
these movies with the same plot fundamentally all and with

790
00:32:55,400 --> 00:32:58,480
the person turning it off right or finding a way

791
00:32:58,559 --> 00:33:00,559
to shut the thing down. And they were making the

792
00:33:00,599 --> 00:33:03,640
same point that you were about the importance of actual.

793
00:33:03,559 --> 00:33:04,759
Speaker 3: Physical off switches.

794
00:33:05,000 --> 00:33:07,319
Speaker 2: Yeah, and so all the technology is great, but undamentally,

795
00:33:07,400 --> 00:33:10,039
the same movie has played out fifty times. We've all

796
00:33:10,079 --> 00:33:12,480
watched it and they're all fundamentally you know, you know

797
00:33:12,640 --> 00:33:14,880
point I'm revenge right, it all ends in pretty much

798
00:33:14,920 --> 00:33:16,759
the same way, with someone finding their way to just.

799
00:33:17,400 --> 00:33:18,000
Speaker 1: Do it wrong.

800
00:33:18,039 --> 00:33:20,480
Speaker 3: Which is kind of interesting that you've said a similar point. Right,

801
00:33:20,599 --> 00:33:23,279
it's not the it's not the obvious place you'd go to, but.

802
00:33:23,799 --> 00:33:25,400
Speaker 1: Be one of the tests for the AI, which is

803
00:33:25,400 --> 00:33:28,319
so like blank is your favorite Gamers camera movie for

804
00:33:28,440 --> 00:33:29,000
Linda Blank?

805
00:33:29,079 --> 00:33:32,680
Speaker 2: Yeah, excellent, right, Yes, we got over there, yep, perfect, Hi.

806
00:33:32,920 --> 00:33:35,359
Speaker 7: Question for you both. So I'm a founder of a

807
00:33:35,359 --> 00:33:37,559
AI and mL scale up in the third Center for

808
00:33:37,599 --> 00:33:39,359
AI which is leads in the North of England and

809
00:33:39,440 --> 00:33:43,039
bit biased since the launch of chat GBT. Three months

810
00:33:43,039 --> 00:33:46,000
after that, we saw a real increase in phishing attacks

811
00:33:46,039 --> 00:33:48,920
using much more sophisticated language patterns. What do we do

812
00:33:49,039 --> 00:33:53,039
to protect businesses consumers so they trust this technology better

813
00:33:53,359 --> 00:33:54,799
and how do we bring them along that.

814
00:33:54,799 --> 00:33:57,279
Speaker 1: Journey with this, Well, I think we shouldn't trusted that much. Actually,

815
00:33:57,920 --> 00:34:00,319
it is actually quite quite a significant allenge because we're

816
00:34:00,319 --> 00:34:02,480
getting to the point where even open source AI can

817
00:34:02,640 --> 00:34:05,920
pass human capture tests. So you know, this is are

818
00:34:05,960 --> 00:34:08,920
you're human identify all the traffic lights in this picture.

819
00:34:09,199 --> 00:34:12,039
You're like, okay, yeah, it's going to have no problem

820
00:34:12,119 --> 00:34:13,559
doing that. In fact, it'll do it better than the

821
00:34:13,639 --> 00:34:15,840
human and faster than human. So we're like, how do

822
00:34:15,960 --> 00:34:17,840
you know it's the point which is a better human

823
00:34:18,039 --> 00:34:21,280
better passing human tests than humans? Then well, what tests

824
00:34:21,400 --> 00:34:23,760
actually make sense? That is a real problem. I don't

825
00:34:23,760 --> 00:34:24,840
actually have a good solution to it.

826
00:34:25,199 --> 00:34:25,239
Speaker 3: That.

827
00:34:25,480 --> 00:34:27,079
Speaker 1: One of the things we're trying to figure out on

828
00:34:27,199 --> 00:34:30,239
the X platform is how to deal with that, because

829
00:34:30,519 --> 00:34:32,760
it really we really are at the point where even

830
00:34:32,880 --> 00:34:35,519
with open source you know, readily available AI, you don't

831
00:34:35,559 --> 00:34:37,360
need to be sort of leading in the field. You

832
00:34:37,440 --> 00:34:40,039
can actually be better than humans at passing these tests,

833
00:34:40,480 --> 00:34:42,239
and that's sort of why we think, well, perhaps we

834
00:34:42,280 --> 00:34:44,239
should sort of charge a dollar or a pound a year.

835
00:34:44,480 --> 00:34:46,800
It's a very tiny amount of money, but it's still

836
00:34:46,840 --> 00:34:49,559
makes it privatively expensive to make a million bots. So

837
00:34:50,079 --> 00:34:52,119
and especially if you need a million payment methods, then

838
00:34:52,159 --> 00:34:54,760
you run out of sort of stolen credit cards pretty quickly.

839
00:34:55,280 --> 00:34:57,199
So that's that's sort of where we're thinking, like we

840
00:34:57,280 --> 00:34:59,039
might have to sort of just charge some very tiny

841
00:34:59,079 --> 00:35:02,719
amount of money three cents a day effectively to deal

842
00:35:02,800 --> 00:35:07,199
with the onsought of AI power advance. And that is

843
00:35:07,239 --> 00:35:08,800
not a growing problem, but it will be I think,

844
00:35:08,920 --> 00:35:11,679
perhaps an insurmountable problem next year. So and then you

845
00:35:11,760 --> 00:35:15,039
have to worry about, well, manipulation of information is making

846
00:35:15,119 --> 00:35:17,079
something seem very popular when in fact it it's not,

847
00:35:17,559 --> 00:35:20,920
because it's getting boosted by all these likes and reposts

848
00:35:21,000 --> 00:35:23,719
from AI powered barts. So that's why I sort of

849
00:35:23,760 --> 00:35:27,480
think somewhat inevitably it leads to some small payment in

850
00:35:27,639 --> 00:35:30,639
order to dramatically increase the cost of a bot. So

851
00:35:31,360 --> 00:35:33,679
I frankly I think probably any social media system that

852
00:35:33,719 --> 00:35:35,519
doesn't do that will simply be over on by bot.

853
00:35:35,679 --> 00:35:38,559
Speaker 2: You know, I think my general answer would be, you know,

854
00:35:38,840 --> 00:35:40,880
we need to show that we are on top of

855
00:35:41,000 --> 00:35:44,519
mitigating the risks right so people can trust the technology.

856
00:35:44,599 --> 00:35:46,320
That's what actually the last couple of days has been

857
00:35:46,360 --> 00:35:49,039
about on the Safety Summit is just showing, you know,

858
00:35:49,119 --> 00:35:52,239
we're investing in the Safety Institute, having the people who

859
00:35:52,280 --> 00:35:54,320
can do the research on these things to figure out

860
00:35:54,320 --> 00:35:57,119
how we mitigate against them, and we have to do

861
00:35:57,199 --> 00:35:59,960
it fast and we have to keep iterating it because

862
00:36:00,519 --> 00:36:02,079
all of us probably in this room, believe that the

863
00:36:02,119 --> 00:36:04,800
technology can be incredibly powerful, but we've got to make

864
00:36:04,800 --> 00:36:06,679
sure we bring people along that journey with us, that

865
00:36:06,800 --> 00:36:09,960
we're handling the risks that are there. And as there's

866
00:36:09,960 --> 00:36:12,159
a job to do, and the last couple of days,

867
00:36:12,400 --> 00:36:14,280
I think we make good progress on it because we

868
00:36:14,360 --> 00:36:16,880
want to focus on the positives and manage these things.

869
00:36:16,960 --> 00:36:19,960
But that requires action, and that's what the last couple

870
00:36:19,960 --> 00:36:20,400
of days.

871
00:36:20,239 --> 00:36:22,440
Speaker 3: Has been about. Your story, you know, analogy.

872
00:36:22,480 --> 00:36:25,400
Speaker 2: There was part of the research that actually, you know,

873
00:36:25,480 --> 00:36:28,880
the team working on the task force here published and

874
00:36:29,039 --> 00:36:30,159
presented yesterday.

875
00:36:30,159 --> 00:36:30,639
Speaker 3: I don't know if you.

876
00:36:30,679 --> 00:36:33,599
Speaker 2: Saw it was, which is essentially that it was using

877
00:36:33,639 --> 00:36:37,639
AI to do to create a ton of fake profiles

878
00:36:37,960 --> 00:36:43,039
on social media and then infiltrate particular groups with particular information.

879
00:36:43,239 --> 00:36:44,800
And actually, at the moment that, as I said, to

880
00:36:44,800 --> 00:36:46,119
your point, and there's a cost.

881
00:36:46,280 --> 00:36:48,800
Speaker 1: Free, it's getting to the point where it's like, really,

882
00:36:48,800 --> 00:36:50,400
you're going to have one hundred for a penny sort

883
00:36:50,440 --> 00:36:51,199
of thing. Ridiculous.

884
00:36:51,239 --> 00:36:52,920
Speaker 2: And if you think about some of these social networks

885
00:36:53,079 --> 00:36:55,519
quite a neighborhood or town level, it's not that many

886
00:36:56,000 --> 00:36:56,639
fake profiles.

887
00:36:57,559 --> 00:36:58,920
Speaker 3: Suddenly they're everywhere and.

888
00:36:58,960 --> 00:37:01,280
Speaker 2: There's some local issue that might be of importance, and

889
00:37:01,440 --> 00:37:03,599
you know, the team of run versions of how that

890
00:37:03,639 --> 00:37:05,880
would look like, and suddenly they're interacting with everybody and

891
00:37:05,920 --> 00:37:09,320
then spreading and misinformation around. Yeah, challenge, and we literally

892
00:37:09,360 --> 00:37:11,000
as part of the research that we published on this

893
00:37:11,119 --> 00:37:13,559
information yesterday, it's a real challenge.

894
00:37:13,719 --> 00:37:16,679
Speaker 1: Yeah, exactly to your point. I mean the images, it's

895
00:37:17,000 --> 00:37:19,440
you don't even need to steal somebody's picture because then

896
00:37:19,599 --> 00:37:22,519
that's traceable. But you can actually just say, create a

897
00:37:22,599 --> 00:37:25,440
new image of a person, realistic looking it doesn't exist,

898
00:37:25,800 --> 00:37:29,760
and then create a biography realistic but doesn't exist, and

899
00:37:30,159 --> 00:37:32,199
do that en mass and practically the only way of

900
00:37:32,239 --> 00:37:33,559
able to tell us that the grammar is too good

901
00:37:33,880 --> 00:37:35,519
give away typhos.

902
00:37:35,599 --> 00:37:37,760
Speaker 2: Come on, now, I'm getting waved at because I think

903
00:37:37,760 --> 00:37:39,440
we are out of time. I don't we take one

904
00:37:39,800 --> 00:37:42,599
very brief last question and let's make it go on. Yes,

905
00:37:42,639 --> 00:37:43,840
sir going, you're right in front of me, go.

906
00:37:44,639 --> 00:37:47,039
Speaker 8: Question for you related to the X platform. Are there

907
00:37:47,079 --> 00:37:49,039
simple things we can do, especially when it comes to

908
00:37:49,119 --> 00:37:51,800
visual media. You alluded to the fact that it's fairly

909
00:37:51,840 --> 00:37:55,039
straightforward and effectively free to make people like yourselves say

910
00:37:55,360 --> 00:37:57,199
and do things that you never said or did. Can

911
00:37:57,239 --> 00:37:59,719
we do something like cryptographically signed media? I'm from Adobe

912
00:38:00,039 --> 00:38:02,679
working on this project. Twitter was a member. Love to

913
00:38:02,719 --> 00:38:05,360
see X come back digitally sign media to indicate not

914
00:38:05,480 --> 00:38:07,400
only what was created by AI, but what came from

915
00:38:07,440 --> 00:38:09,760
a camera, what was real to imview, A sense of

916
00:38:09,840 --> 00:38:11,719
trust in media that can go viral.

917
00:38:11,880 --> 00:38:13,960
Speaker 1: That sounds like a good idea. Actually, so if some

918
00:38:14,079 --> 00:38:17,519
way of authenticating would be good. So, yeah, that sounds

919
00:38:17,519 --> 00:38:18,599
like a good idea, we should probably do it.

920
00:38:18,760 --> 00:38:21,760
Speaker 2: There you go, and actually on that on that point already,

921
00:38:21,800 --> 00:38:24,119
And this is particularly partner for people in my job,

922
00:38:24,360 --> 00:38:27,079
right and I've already had a situation happened to me

923
00:38:27,159 --> 00:38:30,480
with adopted image that goes everywhere negative.

924
00:38:30,159 --> 00:38:31,960
Speaker 3: By the time everyone realizes all.

925
00:38:31,880 --> 00:38:35,480
Speaker 2: That's fake and we should stop sending it the damages damage.

926
00:38:35,519 --> 00:38:37,440
Speaker 3: And actually we were again reflecting today.

927
00:38:37,480 --> 00:38:40,599
Speaker 2: If you think next year, you've got elections in you know,

928
00:38:40,679 --> 00:38:44,079
I think you know, the US, India, I think Indonesia.

929
00:38:43,639 --> 00:38:48,079
Speaker 3: Probably here there you go for massive news and actually you've.

930
00:38:47,960 --> 00:38:50,880
Speaker 2: Got just an enormous junk of the world's population is

931
00:38:50,960 --> 00:38:54,320
voting next year, right, and you've got EU elections as well.

932
00:38:54,920 --> 00:38:55,079
Speaker 1: You know.

933
00:38:55,159 --> 00:38:57,880
Speaker 2: Actually, just these issues are right in front of ours.

934
00:38:58,440 --> 00:39:00,760
Next year is where a big election across the globe,

935
00:39:00,760 --> 00:39:02,599
probably the first set of elections.

936
00:39:02,199 --> 00:39:04,000
Speaker 3: Where this has been a real issue.

937
00:39:04,119 --> 00:39:07,239
Speaker 2: Yeah, So figuring out how we manage that is I

938
00:39:07,320 --> 00:39:09,559
think kind of mission critical for the people who want

939
00:39:09,760 --> 00:39:11,199
the integrity of our democracy.

940
00:39:11,400 --> 00:39:13,800
Speaker 1: Yeah. I mean some of it is quite entertaining, like

941
00:39:14,000 --> 00:39:15,760
the pope and the puffer jacket. Have you seen that one?

942
00:39:16,280 --> 00:39:18,199
That's amazing, But I mean I still write to people

943
00:39:18,199 --> 00:39:20,360
who think that's real. I'm like, well, one of the

944
00:39:20,400 --> 00:39:23,800
answer is wearing a puffer jacket in July, right, be sweating,

945
00:39:23,840 --> 00:39:26,960
but it actually looks quite quite dashing of things. In fact,

946
00:39:27,000 --> 00:39:28,719
I think AI fashion is going to be a real thing.

947
00:39:28,760 --> 00:39:31,559
So doing gloom like we learn in the most interesting times.

948
00:39:31,599 --> 00:39:33,440
And I think this is it is, you know, like

949
00:39:33,559 --> 00:39:36,000
eighty percent likely to be good and twenty percent bad.

950
00:39:36,039 --> 00:39:39,000
And I think if we're cognizant and careful about the

951
00:39:39,039 --> 00:39:41,480
bad part, on balance, actually it will be the future

952
00:39:41,519 --> 00:39:44,119
that we want or for the future that is preferable,

953
00:39:44,440 --> 00:39:47,320
and it actually will be somewhat of a leveler, an

954
00:39:47,360 --> 00:39:49,440
equalizer in the sense that you know, I think everyone

955
00:39:49,480 --> 00:39:52,239
will have access to goods and services and education, and

956
00:39:52,400 --> 00:39:55,000
so you know, I think probably it leads to more

957
00:39:55,079 --> 00:39:57,599
human happiness. So I guess i'd probably leave on an

958
00:39:57,599 --> 00:39:58,360
optimistic note.

959
00:39:58,440 --> 00:39:58,800
Speaker 3: Tough act.

960
00:39:59,079 --> 00:40:01,440
Speaker 2: Yeah, well, that's that is a great note to end on.

961
00:40:01,599 --> 00:40:04,760
I think we all want that better future. I think

962
00:40:04,800 --> 00:40:06,760
it's that the promise of it is certainly there. Lots

963
00:40:06,760 --> 00:40:09,239
of people in this room, including yourselves, are working hard to.

964
00:40:09,239 --> 00:40:09,719
Speaker 3: Make it happen.

965
00:40:09,800 --> 00:40:11,840
Speaker 2: Our job in government is to make sure it happen safely.

966
00:40:12,000 --> 00:40:14,159
But on the basis of this conversation in the last

967
00:40:14,199 --> 00:40:16,639
couple of days, I'm certainly leaving more confident.

968
00:40:17,159 --> 00:40:19,760
Speaker 1: Thanks for listening. See you in the next episode.