Latest / Elon Musk Thinking / Latest: Elon Musk Predictions!!!

Transcript

Latest: Elon Musk Predictions!!! · Elon Musk Thinking

0:00
Speaker 1: Latest interview of Elon Musk.
0:02
Speaker 2: Bill Gates said there is no one in our time
0:04
Speaker 2: who has done more to push the bounds of science.
0:06
Speaker 3: Innovation than you.
0:08
Speaker 2: What's kind of going to say, well, that's a nice
0:10
Speaker 2: thing to have anyone say about you. Nice coming from
0:12
Speaker 2: Bill Gates. But oddly enough, when it comes to AI, actually,
0:15
Speaker 2: for around a decade, you've almost been doing the opposite
0:18
Speaker 2: and saying, hang on, you need to think about what
0:21
Speaker 2: we're doing and what we're pushing here, and what do
0:23
Speaker 2: we do to make this safe? And actually maybe we
0:25
Speaker 2: shouldn't be pushing as faster or as hard as we are.
0:28
Speaker 2: I mean, you've been doing it for a decade, Like
0:30
Speaker 2: what was it that caused you to think about it
0:32
Speaker 2: that way? And you know, why do we need to
0:34
Speaker 2: be worried?
0:34
Speaker 4: Yeah? I've been somewhat of a Cassandra for quite a
0:36
Speaker 4: while where people would I tell you, like, we should
0:39
Speaker 4: really be concerned about AI, and the'd be like, what
0:40
Speaker 4: are you talking about? Like I've never really had any
0:42
Speaker 4: experience with it since I was immersed in technology. I
0:46
Speaker 4: have been immersed in technology for a long time. I
0:48
Speaker 4: could see it coming, so, but I think this year
0:51
Speaker 4: was that there's been a number of breakthroughs. I mean,
0:53
Speaker 4: you know, at the point in which someone can see
0:55
Speaker 4: a dynamically created video of themselves, you know, like so
0:58
Speaker 4: you can make a video of you say anything in
1:00
Speaker 4: real time or me, and so there's sort of the
1:03
Speaker 4: deep videos, which are really incredibly good, in fact, sometimes
1:05
Speaker 4: more convincing than real ones and deep real and then
1:09
Speaker 4: and then obviously things like chat GPT were quite remarkable,
1:12
Speaker 4: and now I saw a GPT one, GPT two, GPT three,
1:15
Speaker 4: GPD four that you know, the whole sort of lead
1:17
Speaker 4: up to that, So it was easy for me to
1:20
Speaker 4: kind of see where it's going. If you just sort
1:21
Speaker 4: of extrapolate the points on a curve and as seeing
1:24
Speaker 4: that tread will continue, then we will have profound artificial intelligence,
1:28
Speaker 4: and obviously at the level that far exceeds human intelligence.
1:31
Speaker 4: So but I'd see at this point that people are
1:34
Speaker 4: taking safety seriously, and I like, I say thank you
1:36
Speaker 4: for holding this AI safety conference. I think actually it
1:39
Speaker 4: will go down in history as being very important. I
1:41
Speaker 4: think it's really quite profound, and I do think overall
1:44
Speaker 4: that the potential is there for artificial intelligence AI to
1:48
Speaker 4: have most likely a positive effect and to create a
1:51
Speaker 4: future of abundance where there is no scarcity of goods
1:54
Speaker 4: and services. But it is somewhat of the magic gene
1:57
Speaker 4: problem where if you have a magic gene that can
1:59
Speaker 4: grab all the wishes, usually those stories don't end well.
2:03
Speaker 4: Who careful what you wish for? Including wishes?
2:05
Speaker 2: So you talked a little bit about the summit and
2:08
Speaker 2: thank you for being engaged in it, which has been
2:10
Speaker 2: great and people enjoyed having you there pulling in this dialogue. Now,
2:13
Speaker 2: one of the things that we achieved today in the
2:16
Speaker 2: meetings between the companies and the leaders was an agreement
2:21
Speaker 2: that externally, ideally, governments should be doing safety testing of
2:26
Speaker 2: models before their relief. I think this is something that
2:28
Speaker 2: you've spoken about a little bit. It was something we
2:30
Speaker 2: worked really hard on because you know, my job in
2:32
Speaker 2: government is to say, hang on, there is a potential risk,
2:35
Speaker 2: and not a definite risk, but a potential risk of
2:38
Speaker 2: something that could be bad. And my job is to
2:40
Speaker 2: protect the country and we can only do that if
2:43
Speaker 2: we develop the capability we need in our safety institute
2:46
Speaker 2: and then go in and make sure we can test
2:47
Speaker 2: the models before they are released. Delight, did that happened today.
2:50
Speaker 2: But you know, what's your view on what we should
2:52
Speaker 2: be doing. Right, You've talked about the potential risk. Right again,
2:55
Speaker 2: we don't know, but you know, what are the types
2:57
Speaker 2: of things governments like ours should be doing to manage
2:59
Speaker 2: and de gate against those risks.
3:01
Speaker 4: Well, I generally think that it is good for government
3:03
Speaker 4: to play a role when the public safety is at risk. So,
3:07
Speaker 4: you know, really, for the vast majority of software, the
3:10
Speaker 4: public safety is not at risk. If the app crashes
3:12
Speaker 4: on your phone, your laptop, it's not a massive catastrophe.
3:15
Speaker 4: But when talking about digital superintelligence, I think, which does
3:20
Speaker 4: pose a risk to the public, then there is a
3:23
Speaker 4: role for government to play to safeguard the interest of
3:25
Speaker 4: the public. And this is of course true in many fields,
3:29
Speaker 4: you know, aviation, cars, you know, I deal with regulators
3:32
Speaker 4: throughout the world because of stalling being communications, rockets being aerospace,
3:37
Speaker 4: and cars, you know, being vehicle transport. So I'm very
3:40
Speaker 4: familiar with dealing with with regulators, and I actually agree
3:42
Speaker 4: with the vast majority of regulations. There's a few that
3:44
Speaker 4: I disagree with from time to time, but zero point
3:46
Speaker 4: one percent, probably less than one percent of regulations I
3:49
Speaker 4: disagree with. So there is some concern from people in
3:52
Speaker 4: Silicon Valley who who have never dealt with regulators before,
3:55
Speaker 4: and they think that this is going to just crush
3:57
Speaker 4: innovation and slow them down and be annoying. But and
4:00
Speaker 4: it will be annoying, It's true. They're not wrong about that.
4:03
Speaker 4: But I think there's we've learnt over the years that
4:06
Speaker 4: having a referee is a good thing. And if you
4:09
Speaker 4: look at any sports game, there's always a referee, and
4:12
Speaker 4: nobody's suggesting, I think, to have a sports game without one.
4:15
Speaker 4: And I think that's the right way to think about this,
4:17
Speaker 4: is for governed to be a referee, to make sure
4:20
Speaker 4: the sportsmanlike conduct and that the public safety is you know,
4:24
Speaker 4: is addressed. That we care about the public safety because
4:26
Speaker 4: I think there might be at times too much optimism
4:29
Speaker 4: about technology. And I speak I say that as a technologist.
4:31
Speaker 4: I mean, so I ought to know and like I
4:33
Speaker 4: said that, on balance, I think that the AI will
4:36
Speaker 4: be a forcible good most likely, but the probability of
4:40
Speaker 4: it going bad it's not zero percent. Yeah, so we
4:43
Speaker 4: just need to mitigate the downside potential.
4:45
Speaker 3: And then how you talk about referee and that's what
4:47
Speaker 3: we're demonstrate.
4:48
Speaker 2: The yeah, well there we go, I mean, you know,
4:50
Speaker 2: and we talked about this in Demos and I discussed
4:52
Speaker 2: this a long time.
4:52
Speaker 4: Ago, like literally facing right and actually.
4:56
Speaker 3: Demoster his credit and the credit of people in the industry,
4:58
Speaker 3: did say that to us.
4:59
Speaker 2: You say it's not right. Yeah, that Demis and his
5:01
Speaker 2: colleagues are marking their own homework.
5:03
Speaker 4: Right.
5:03
Speaker 2: There needs to be someone independent, and that's why we've
5:05
Speaker 2: developed the safety Institute here. I mean, do you think
5:09
Speaker 2: governments can develop the expertise. One of the things we
5:11
Speaker 2: need to do is they hang on you know, Demis
5:14
Speaker 2: and Exam all the others have got a lot of
5:16
Speaker 2: very smart people doing this. Governments need to quickly tool
5:19
Speaker 2: up capability wise, personnel wise, which is what we're doing.
5:22
Speaker 3: I mean, do you think it is possible.
5:23
Speaker 2: For governments to do that fast enough, given how quickly
5:26
Speaker 2: the technology is developing, or what do we need to
5:28
Speaker 2: do to make sure we do it quick enough?
5:29
Speaker 4: No, I think it's a great point you're making. The
5:31
Speaker 4: pace of AI is faster than any technology I've seen
5:35
Speaker 4: in history by far, and it seems to be growing
5:37
Speaker 4: in capability by at least fivefold peraps tenfold per year.
5:41
Speaker 4: It'll certainly grow by an automatitude next year.
5:43
Speaker 3: Yeah.
5:44
Speaker 4: So and government isn't used to moving at that speed.
5:47
Speaker 4: But I think even if there are not firm regulations,
5:51
Speaker 4: even if there's not even if there isn't an enforcement capability,
5:54
Speaker 4: so we're having insight and being able to highlight concerns
5:56
Speaker 4: to the public will be very powerful. So even if
6:00
Speaker 4: that's all that's accomplished, I think that will be very good.
6:02
Speaker 3: Okay, y, well, hopefully we can do better than that.
6:04
Speaker 4: Hopefully.
6:04
Speaker 2: Yeah, yeah, no, but that's how fair. Actually we were
6:07
Speaker 2: talking before it was striking. You know, you're someone who
6:09
Speaker 2: spent their life in technology. They're living More's law. And
6:13
Speaker 2: what it was interesting over the last couple of days
6:16
Speaker 2: talking to everyone who's doing the development of this, and
6:18
Speaker 2: I think you concur with this is just the pace
6:21
Speaker 2: of advancement here is unlike anything all of you have
6:24
Speaker 2: seen in your careers and technology.
6:26
Speaker 3: Is that fair because you've got these kind of.
6:28
Speaker 2: Compounding effects from the hardware and the data and the personnel.
6:32
Speaker 4: Yeah, I mean the two currently the two leading centers
6:35
Speaker 4: for a development are the San Francisco Bay area and
6:37
Speaker 4: the sort of London area, And there are many other
6:39
Speaker 4: places where it's being done, but those are the two
6:41
Speaker 4: leading areas. So I think if you know, if the
6:43
Speaker 4: United States and the UK and China are sort of
6:47
Speaker 4: aligned on safety, that's all going to be. That's really
6:50
Speaker 4: that's where that's where the leadership is general.
6:52
Speaker 3: And you actually you mentioned China there.
6:54
Speaker 2: So I took a decision to invite China to summit
6:57
Speaker 2: over the last three days, and it was not easy decision.
7:00
Speaker 2: A lot of people criticize me for you know, my viewers,
7:03
Speaker 2: if you're going to try it essential serious conversation, you
7:05
Speaker 2: need to. But what would your thoughts? You do business
7:07
Speaker 2: all around the world. You just talked about it there.
7:10
Speaker 2: Should we be engaging with them? Can we trust them?
7:12
Speaker 2: Is that the right thing to have done?
7:14
Speaker 4: If we don't, If China is not on board with
7:16
Speaker 4: AI safety, it's somewhat of a moot situation. The single
7:19
Speaker 4: biggest objection that I get to any kind of AI
7:22
Speaker 4: regulation or sort of safety controls are well, China is
7:25
Speaker 4: not going to do it, and therefore they will just
7:27
Speaker 4: jump into the lead and exceed us all. But actually
7:30
Speaker 4: China is willing to participay in safety and thank you
7:33
Speaker 4: for inviting them, and they you know, I think we
7:36
Speaker 4: should thank China for for attending. When I was in China,
7:38
Speaker 4: earlier this year. My main subject of discussion with the
7:42
Speaker 4: leadership in China was AI safety and saying that this
7:45
Speaker 4: is really something that they should care about, and they
7:47
Speaker 4: took it seriously and you are too, which is great.
7:50
Speaker 4: And having them here I think was essential. Really, if
7:52
Speaker 4: they're not participants, it's pointless.
7:55
Speaker 3: Yeah, that's and I think we were pleased.
7:57
Speaker 2: I think they were engaged yesterday in the discussions and
7:59
Speaker 2: actually ended up signing the same communicy that everyone else did. Great,
8:02
Speaker 2: which is a good stop, right, And I said, if
8:05
Speaker 2: we need everyone to approach us in a similar way,
8:07
Speaker 2: if we're going to have a realistic chance of resolving it.
8:10
Speaker 2: I was going to You talked about innovation earlier and
8:12
Speaker 2: regulation being annoying.
8:14
Speaker 3: There was a good debate today we had about.
8:16
Speaker 2: Open source and I think you've kind of been a
8:18
Speaker 2: proponent of algorithmic transparency and making some of the ex
8:21
Speaker 2: algorithms public and actually we were talking about every hint
8:25
Speaker 2: and on the way in. He's particularly been very concerned
8:28
Speaker 2: about open source models being used.
8:30
Speaker 3: By bad actors.
8:32
Speaker 2: You've got a group of people who say they are
8:34
Speaker 2: critical to innovation happening in that distributed way.
8:37
Speaker 3: Look, it's a trick.
8:38
Speaker 2: There's probably no perfect answer, and there is a tricky balance.
8:40
Speaker 2: What are your thoughts on how we should approach this
8:42
Speaker 2: open source question or where should we be targeting whatever
8:46
Speaker 2: regulatory or monitoring that we're going.
8:48
Speaker 4: To do well. The open source algorithms and data tend
8:52
Speaker 4: to lag the closed source by six to twelve, but
8:55
Speaker 4: given the rate of improvement that there's actually there for
8:57
Speaker 4: quite a big difference between the closed source in the
9:00
Speaker 4: open If things are improving by a factor of let's
9:02
Speaker 4: say five or more than being a year behind is
9:05
Speaker 4: you're five times worse. So it's a pretty big difference.
9:08
Speaker 4: And that might be actually an okay situation, but it
9:10
Speaker 4: certainly we'll get to the point where you've got open
9:12
Speaker 4: source AI that can do that, that will start to
9:15
Speaker 4: approach human level intelligence or PAPS succeeded. I don't know
9:18
Speaker 4: quite what to do about it. I think it's somewhat
9:19
Speaker 4: in neeverment, there'll be some amount of open source, and
9:22
Speaker 4: I guess I would have a slight bias towards open
9:24
Speaker 4: source because at least you can see what's going on.
9:26
Speaker 4: It wears a closed source. You don't know what's going
9:28
Speaker 4: on now. It should be said with AI that even
9:30
Speaker 4: if it's an open source, do you actually know what's
9:32
Speaker 4: going on. Because if you've got a gigantic data file
9:35
Speaker 4: and you know sort of billions of days of weights
9:37
Speaker 4: and parameters, you can't just read it and see what
9:39
Speaker 4: it's going to do. It's a gigantic file of inscrutable numbers.
9:43
Speaker 4: You can test it when you run it. You can
9:45
Speaker 4: test it. You can run a bunch of tests to
9:47
Speaker 4: see what it's going to do. But it's probabilistic as
9:49
Speaker 4: opposed to deterministic. It's not like traditional programming where you've
9:52
Speaker 4: got you've got very discrete logic and the outcome is
9:55
Speaker 4: very predictable and you can read each line and see
9:57
Speaker 4: what each line's going to do. A neural net is
9:59
Speaker 4: just a much of probabilities. I mean, it sort of
10:02
Speaker 4: ends up being a giant Comma separated value file. It's
10:04
Speaker 4: like our digital guide is a CSP file. Really, Okay,
10:08
Speaker 4: that is kind of what it is.
10:09
Speaker 2: Yeah, Now at that point you've just made is one
10:11
Speaker 2: that we have been talking about a lot, because again
10:15
Speaker 2: conversation with the people are developing their technology.
10:17
Speaker 3: Make the point that you've just made. It is not
10:19
Speaker 3: like normal.
10:19
Speaker 2: Software where there's predictability about inputs improving leading to this
10:24
Speaker 2: particular output improving, and as the models iterate and improve,
10:28
Speaker 2: we don't quite know what's going to come out the
10:30
Speaker 2: other end. I think we'd agree with that, which is
10:32
Speaker 2: why I think there is this bias that we need
10:34
Speaker 2: to get in there while the training runs are being
10:35
Speaker 2: done before the models are released, to understand what is
10:38
Speaker 2: this new iteration broad about in terms of capability, which
10:41
Speaker 2: it sounds like you would agree with.
10:44
Speaker 3: I was going to shift goes a little bit on.
10:46
Speaker 2: You know, you've talked a lot about human consciousness human agency,
10:48
Speaker 2: which it actually might strike people as strange given that
10:52
Speaker 2: you are known for being such a brilliant innovator and technologist,
10:55
Speaker 2: but it's quite heartfelt when I hear you talk about
10:57
Speaker 2: it and the importance of maintaining that agency in technology and.
11:01
Speaker 3: Preserving human consciousness. Now it kind of links.
11:04
Speaker 2: The thing I was going to ask is when I
11:06
Speaker 2: do interviews or talk to people out and about in
11:08
Speaker 2: this job about AI, the thing that comes up most
11:10
Speaker 2: actually is it probably not so much the stuff we've
11:12
Speaker 2: been talking about, but jobs. It's what is AI mean
11:15
Speaker 2: for my job? Is it going to mean that I
11:17
Speaker 2: don't have a job or my kids are not going
11:19
Speaker 2: to have a job.
11:20
Speaker 4: Now.
11:21
Speaker 2: Answer as a policymaker, as a leader is actually, AI
11:24
Speaker 2: is already creating jobs, and you can see that in
11:27
Speaker 2: the companies.
11:27
Speaker 3: That are starting.
11:28
Speaker 2: Also, the way it's being used is a little bit
11:31
Speaker 2: more as a co pilot necessarily versus replacing the person.
11:35
Speaker 2: There's still human agency, but it's helping you do your
11:37
Speaker 2: job better, which is a good thing. And as we've
11:40
Speaker 2: seen with technological revolutions in the past, clearly there's change
11:44
Speaker 2: in the labor market.
11:45
Speaker 3: The amount of jobs.
11:46
Speaker 2: I was quoting an MIT study today that they did
11:49
Speaker 2: a couple of years ago, something like sixty percent of
11:51
Speaker 2: the jobs at that moment didn't exist forty years ago,
11:53
Speaker 2: So hard to predict. And my job is to create
11:55
Speaker 2: an incredible education system, whether it's at school, whether it's
11:58
Speaker 2: retraining people at any point in their career, because ultimately,
12:01
Speaker 2: if we've got a skill population, they'll be able to
12:03
Speaker 2: keep up with the pace of change and have a
12:06
Speaker 2: good life. But you know that it's still a concern,
12:09
Speaker 2: and you know, what would your kind of observation be
12:12
Speaker 2: on AI and the impact of labor markets and people's
12:15
Speaker 2: jobs and how they should feel about that as they
12:18
Speaker 2: think about.
12:18
Speaker 4: This, well, I think we are seeing the most disruptive
12:21
Speaker 4: force in history here. You know, where we have for
12:24
Speaker 4: the first time, we will have the first time something
12:26
Speaker 4: that is smarter than the smartest human and that I
12:29
Speaker 4: mean it's hard to say exactly what that moment is,
12:31
Speaker 4: but there will come a point where no job is needed.
12:34
Speaker 4: You can have a job if you want to have
12:35
Speaker 4: a job for sort of personal satisfaction, but the AI
12:39
Speaker 4: will be able to do everything. So I don't know
12:42
Speaker 4: if that makes people comfortable uncomfortable. You know. That's why
12:44
Speaker 4: I say, if you wish for a magic genie that
12:46
Speaker 4: gives you any wishes you want and there's no limit,
12:48
Speaker 4: you don't have those three limits three wish limits, not
12:50
Speaker 4: since you have many wishes as you want. So it's
12:53
Speaker 4: both good and bad. One of the challenges in the
12:55
Speaker 4: future will be how do we find meaning in life
12:57
Speaker 4: if you have a magic genie that can do everything
13:00
Speaker 4: you want. I do think we's it's hard, you know.
13:03
Speaker 4: Whene's when when there's new technology, it tends to have
13:06
Speaker 4: usually follow an S curve. In this case, we're going
13:08
Speaker 4: to be on the exponential portion of the S curve.
13:10
Speaker 4: For a long time and we have to ask for anything,
13:13
Speaker 4: and it won't be if we won't have universal basic income,
13:15
Speaker 4: we'll have universal high income. So in some in some sense,
13:18
Speaker 4: it will be somewhat of a leveler or an equalizer,
13:20
Speaker 4: because really, I think everyone will have access to this
13:22
Speaker 4: magic gene and you're able to ask any question. It
13:24
Speaker 4: will certainly be good for education. It'll be the best
13:26
Speaker 4: tutor you could, the most patient tutor sit there all day,
13:30
Speaker 4: and there will be no shortage of goods and services.
13:32
Speaker 4: Will be an age of abundance. I think if I'd
13:34
Speaker 4: recommend people read in Banks, the Banks culture books are
13:38
Speaker 4: probably the best in visioning. In fact, not probably, they're
13:40
Speaker 4: definitely by far the best envisioning of an a future.
13:43
Speaker 4: There's nothing even close. So I'd recommend, really recommend Banks.
13:46
Speaker 4: I'm a very big fan. All these books are good.
13:48
Speaker 4: There's not Jay which one all of them. So that's
13:51
Speaker 4: that'll give you a sense of what is a I
13:54
Speaker 4: guess a fairly utopian pro toopian future with with Ai, Yeah,
13:58
Speaker 4: which is.
13:58
Speaker 2: Good from an as you said, it's a universal high income,
14:01
Speaker 2: which is a nice phrase, and that's it's good for
14:03
Speaker 2: a materialistic sense of abundance. Actually, that kind of then
14:07
Speaker 2: leads to the question.
14:08
Speaker 3: That you pose. Right, I'm someone who believes, you know,
14:10
Speaker 3: work gives you meaning. I write a lot about that.
14:12
Speaker 3: As you know, I think work is a good thing.
14:14
Speaker 2: It, you know, gives people purpose in their lives. And
14:16
Speaker 2: if you then remove a large chunk of that, you know,
14:19
Speaker 2: what does that mean?
14:20
Speaker 3: And where do you get that?
14:21
Speaker 2: You know, where do you get that drive, that motivation,
14:23
Speaker 2: that purpose. I mean, you're talking about it. You work
14:25
Speaker 2: a lot of our you know, as I was.
14:27
Speaker 4: Mentioning when we were talking earlier, I have to somewhat
14:29
Speaker 4: engage in the liberty suspension of disbelief because I'm putting
14:32
Speaker 4: so much blood, sweat and tears into a work project
14:35
Speaker 4: and burning the you know, three am oil. Then I'm like, wait,
14:38
Speaker 4: why am I doing this? I just wait for the
14:40
Speaker 4: AI to do it. I'm just lashing myself for no reason.
14:43
Speaker 4: It must be a glutton for punishment.
14:45
Speaker 3: Called demos and tell them to hurry up and then
14:46
Speaker 3: you can have a holiday. Right, it's a plan. Yeah, No,
14:49
Speaker 3: it's a Look, it's a tricky.
14:50
Speaker 2: It's a tricky thing because I think, you know, part
14:52
Speaker 2: of our job is to make sure that we can
14:55
Speaker 2: navigate to that very I think largely positive place and
14:58
Speaker 2: help people through it between now and then, because these
15:01
Speaker 2: things bring a lot about change in the labor market
15:03
Speaker 2: as we've seen.
15:04
Speaker 4: Yeah, I think it's probably is generally a good thing
15:06
Speaker 4: because you know, there are a lot of jobs that
15:08
Speaker 4: are uncomfortable or dangerous or which sort of tedious, and
15:12
Speaker 4: the computer will have no problem doing that. We're happy
15:14
Speaker 4: to do that all day long. So you know, it's
15:16
Speaker 4: fun to cook food, but it's not that fun to
15:17
Speaker 4: wash the dishes, but the computer is perfectly happy to
15:20
Speaker 4: watch the dishes. I guess there is. You know, we
15:22
Speaker 4: still have a sports like where where humans compete and
15:25
Speaker 4: like the Olympics, and obviously a machine can can go
15:29
Speaker 4: faster than any human, but we still have we saw
15:31
Speaker 4: humans race against each other and have all, you know,
15:34
Speaker 4: have these sports competitions against each other where even though
15:36
Speaker 4: the machines are better, which there's all I guess compunity
15:39
Speaker 4: who can be the best human at something, and people
15:41
Speaker 4: do find pulforment in that. So I guess that's perhaps
15:43
Speaker 4: a good example of how even when machines are faster than.
15:45
Speaker 3: Us, stronger than us, we still find a way.
15:47
Speaker 4: We still enjoy competing against other humans, so at least
15:49
Speaker 4: who's the best human.
15:50
Speaker 2: It's a good it's a good analogy, and we've been
15:52
Speaker 2: talking a lot about managing the risks. Just before we
15:55
Speaker 2: move on and finish on AI is just talk a
15:57
Speaker 2: little bit about the opportunities. You know, you're engaged in
16:01
Speaker 2: lots of different companies. You're being an obvious one, which
16:04
Speaker 2: is doing some exciting stuff. You touched on the thing
16:06
Speaker 2: that I'm probably most excited about, which is an education. Yeah,
16:09
Speaker 2: and I think many people will have seen sous video
16:12
Speaker 2: from earlier this year. Is ted talk about as you
16:14
Speaker 2: talked about it's like a personal tutor.
16:16
Speaker 4: Yeah, an amazing personal tutor.
16:18
Speaker 3: An amazing personal tutor.
16:19
Speaker 2: And we know the difference in learning having that personal
16:22
Speaker 2: his tutor is incredible compared to class from learning. So
16:25
Speaker 2: if you can have every child have a personal tutor
16:27
Speaker 2: specifically for them that then just evolves with them over time,
16:30
Speaker 2: that could be extraordinary. So that you know, for me,
16:32
Speaker 2: I look at it, I think, gosh, that is within
16:34
Speaker 2: reach at this point, and that's one of the benefits
16:36
Speaker 2: I'm most excited about. When you look at the landscape
16:39
Speaker 2: of things that you see as possible, what is it
16:41
Speaker 2: that you know you are particularly excited about.
16:44
Speaker 4: I think certainly AI tutors are going to be the
16:47
Speaker 4: amazing perhaps already are. I think there's also perhaps companionship,
16:51
Speaker 4: which may seem odd because how can the computer really
16:53
Speaker 4: be your friend? But if you haven't that has memory,
16:56
Speaker 4: you know, and remembers all of your interactions and has
16:59
Speaker 4: read every you're going to actually like give it permission
17:01
Speaker 4: to read everything you've ever done, so really will know
17:03
Speaker 4: you better than anyone, perhaps even yourself. And where you
17:06
Speaker 4: can talk to it every day and those conversations spoiled
17:08
Speaker 4: upon each other, you will actually have a great friend
17:11
Speaker 4: as long as that friend can stay your friend and
17:13
Speaker 4: not get turned off or something. Don't turn off my friends.
17:16
Speaker 4: But I think that will actually be a real thing.
17:18
Speaker 4: And I one of my sons is sort of has
17:20
Speaker 4: some learning disabilities and has trouble making friends actually, and
17:24
Speaker 4: I was like, well, you know, he AI friend would
17:26
Speaker 4: actually be a great friend.
17:28
Speaker 2: Okay, you know that was a surprising answer. That's actually
17:31
Speaker 2: it's worth worth reflecting on that. That's really interesting. I
17:34
Speaker 2: mean we're already seeing it actually as we deliver psychotherapy anyway,
17:38
Speaker 2: now doing far more by digitally and by telephone to
17:42
Speaker 2: people and it's making a huge difference, and you can
17:44
Speaker 2: see a world in which actually, you know, AI can
17:46
Speaker 2: provide that social benefit to people. Just a quick question
17:50
Speaker 2: on X and then we should open it up to everybody.
17:54
Speaker 2: You made a change when in one of the made
17:56
Speaker 2: many changes, but one of the one of.
17:58
Speaker 4: The changes that letter. Yet thing about it, you really.
18:01
Speaker 2: Do, really do one of the changes which you know
18:03
Speaker 2: kind of you know, it goes into the space that
18:05
Speaker 2: you know we have to operate in. And this balance
18:08
Speaker 2: between free speech and moderation is you know, we grapple.
18:11
Speaker 3: With as politicians.
18:13
Speaker 2: You were grappling with your own version of that, and
18:15
Speaker 2: you moved away from a kind of manual human.
18:18
Speaker 3: Way of doing it the moderation to the community and that.
18:22
Speaker 2: Yeah, and I think that's it was an interesting change, right,
18:24
Speaker 2: It's not what everyone else has done. It would be good,
18:27
Speaker 2: you know what's what was the reasoning behind that and
18:29
Speaker 2: why do you think that is a better way to
18:31
Speaker 2: do that?
18:31
Speaker 4: Yeah. Part of the problem is if you if you
18:33
Speaker 4: empower people as sensors, then well there's going to be
18:36
Speaker 4: some amount of bias that they have. And then whoever
18:39
Speaker 4: appoints the sensors is effectively in control of information. So
18:42
Speaker 4: then the idea behind community notice, well, how do we
18:44
Speaker 4: have a consensus driven I mean, so it's not really
18:47
Speaker 4: censoring it, but consensus driven approach to truth. How do
18:50
Speaker 4: we how do we make things the least amount untrue?
18:53
Speaker 4: Could you say, like you can't perhaps get to pure truth,
18:56
Speaker 4: but you can aspire to be more truthful. So the
18:59
Speaker 4: thing about comurnity notes, it doesn't actually delete anything, It
19:02
Speaker 4: simply adds context. Now that context could be this thing
19:04
Speaker 4: is untrue for the following reasons. But importantly, with community notes,
19:08
Speaker 4: everything is open source actually, so you can see the software,
19:12
Speaker 4: every line of the software, you can see all of
19:14
Speaker 4: the data that went into a community note, and you
19:16
Speaker 4: can independently create that community note. So if you've got
19:20
Speaker 4: if you see manipulations of data, you can actually highlight
19:22
Speaker 4: that and say, well this this appears to be some
19:24
Speaker 4: gaming of the system, and you can suggest improvement. So
19:28
Speaker 4: it's maximum transparency, which.
19:30
Speaker 2: Is I think combined with the kind of wisdom of
19:32
Speaker 2: the crowds and trying to get to a better answer.
19:35
Speaker 4: And really one of the key elements of community notes
19:37
Speaker 4: is that in order for a note to be shown,
19:39
Speaker 4: people who have historically disagreed must agree, and there is
19:42
Speaker 4: a bit of AI usage here. So this will populate
19:45
Speaker 4: a parameter space around each contributor to the community notes,
19:49
Speaker 4: and then a parameter space so everyone's got basically these
19:52
Speaker 4: vectors associated with them, which so it's not as simple
19:55
Speaker 4: as right or left, it's saying, it's more it's several
19:58
Speaker 4: hundred vectors. That that because things are more complicated than
20:01
Speaker 4: something right, right or left. And then we'll do sort
20:04
Speaker 4: of inverse correlation, say like, okay, these people generally disagree,
20:07
Speaker 4: but they agree about this note, so then that so
20:10
Speaker 4: then that that that gives the note credibility. Okay, yeah,
20:13
Speaker 4: that's that's the core of it, and it's working quite well.
20:16
Speaker 4: I get to see a note actually be present for
20:18
Speaker 4: more than a few hours, but that that is incorrect.
20:21
Speaker 4: So the batting average is extremely good. And when I
20:23
Speaker 4: ask people say, oh, they're worried about community notes sort
20:25
Speaker 4: of being disinformation, like send me one, and then they can't,
20:28
Speaker 4: so so I think it's I think it's quite good.
20:29
Speaker 4: I mean, the general aspiration is with the X platform
20:33
Speaker 4: is to inform and entertain the public and to be
20:36
Speaker 4: as accurate as possible and as truthful as possible, even
20:39
Speaker 4: if someone doesn't like the truth. You know, people don't
20:41
Speaker 4: always like the truth always, but that's the aspiration. And
20:44
Speaker 4: I think if we are, if we stay true to
20:47
Speaker 4: the truth, then I think we'll find that people use
20:50
Speaker 4: the system to learn what is going on and to
20:53
Speaker 4: I think actually truth pays, so I think it'll be well,
20:56
Speaker 4: I mean, assuming you don't want to engage your self delusion,
20:59
Speaker 4: then I think it's the smart move.
21:01
Speaker 3: Excellent, very helpful. Right, let's open it up to all
21:04
Speaker 3: our guests here, and.
21:05
Speaker 2: We've got some microphones that'll company I find you have,
21:07
Speaker 2: We've got yes, go for it.
21:09
Speaker 3: Thank you, good evening.
21:11
Speaker 5: Alice Bentink from Entrepreneur. First, thank you for a fascinating conversation.
21:14
Speaker 5: I suppose a question for each of you, Prime Minister.
21:17
Speaker 5: The UK has some of the best universities in the world,
21:19
Speaker 5: we have the talent. What will it takes the UK
21:22
Speaker 5: to be a real breeding with unicorn companies? Being a
21:26
Speaker 5: founder in the UK is still a non obvious career
21:28
Speaker 5: choice for the most exceptional technical talent. What are the
21:31
Speaker 5: cultural elements that we need to put into place to
21:33
Speaker 5: change this?
21:34
Speaker 3: Thank you? You want to go first?
21:35
Speaker 4: Go for it? Sure? Well, You're right that there are
21:37
Speaker 4: cultural elements where you know, the culture should celebrate creating
21:41
Speaker 4: new companies and there should be a bias towards supporting
21:45
Speaker 4: small companies because the ones that need nurturing. The larger
21:48
Speaker 4: companies really don't need nurturing. So you know, just you
21:51
Speaker 4: can think of it's sort of like a garden. If
21:52
Speaker 4: it's a little sprout that needs needs nurturing, if it's
21:55
Speaker 4: a mighty ogre, doesn't need quite as much. So I
21:57
Speaker 4: think that is a mindset change that is important. But
22:00
Speaker 4: I should mention that London is you know, London and
22:04
Speaker 4: San Francisco or the Bay Area are really the two
22:06
Speaker 4: centers for AI. So that so London is actually doing
22:09
Speaker 4: very well on that front. That the two I said,
22:12
Speaker 4: the two leading locations on Earth. You know, San Francisco's
22:15
Speaker 4: probably head of London, but London's really very strong. But
22:18
Speaker 4: London Area greater London home counties, I guess. So I'm
22:21
Speaker 4: just saying objectively this is the case. But you do
22:23
Speaker 4: need that. You need the infrastructure, you need the landlords
22:25
Speaker 4: who are willing to rent to new companies. You need
22:28
Speaker 4: more firms and accounts that are willing to support new companies.
22:30
Speaker 4: And it's generally it is a mindset change, and I
22:33
Speaker 4: think some of that is happening, but I think really
22:35
Speaker 4: it's just culturally people need to decide this is a
22:37
Speaker 4: good thing. Yeah.
22:38
Speaker 2: Yeah, no, Actually, well thanks for what you said about
22:40
Speaker 2: the UK. It's something that we work hard on. Lots
22:42
Speaker 2: of people in the room are part of what makes
22:44
Speaker 2: this a fabulous place for companies, including Alice. So that's
22:48
Speaker 2: what i'd say is my job is to get all
22:50
Speaker 2: the you know they're nuts and volts right, make sure
22:52
Speaker 2: that all of you are starting companies can raise the capital.
22:55
Speaker 3: That you need everything from.
22:56
Speaker 2: You know, you'll see funding with our incredible you know
22:59
Speaker 2: EI tax release all the way through.
23:01
Speaker 3: To your late stage rounds.
23:03
Speaker 2: And we need reform of our pension funds and the
23:06
Speaker 2: Chancellor's got a bunch of incredible reforms to unlock capital
23:09
Speaker 2: from all the people who have it and deploy it
23:11
Speaker 2: into growth equity.
23:12
Speaker 4: Right.
23:13
Speaker 2: That is a work in progress. We're not there yet,
23:15
Speaker 2: but I think we're making good progress. We need talent,
23:17
Speaker 2: we need people, so that means an education system that
23:20
Speaker 2: prioritizes the things that matter.
23:22
Speaker 3: And you've seen my reforms.
23:23
Speaker 2: I go on about more maths, more maths, more maths,
23:26
Speaker 2: but I think that it's important but also attracting the
23:28
Speaker 2: best and the brightest fit. If you look at our
23:30
Speaker 2: fastest growing companies in this country, and I think it's
23:33
Speaker 2: probably the same in the US, over half of them
23:36
Speaker 2: have a non British founder, right, And so that tells
23:39
Speaker 2: you we've got to be a place that is open
23:41
Speaker 2: to the world's best and brightest entrepreneurial talent. So the
23:44
Speaker 2: visa regime that we've put in place, I think, does
23:46
Speaker 2: that makes it easy for those people to come here.
23:48
Speaker 2: And then actually it's the thing that we've spent the
23:50
Speaker 2: beginning of the session talking about the regulation, right, making
23:53
Speaker 2: sure that we've got a regulatory system that's pro innovation.
23:57
Speaker 4: That.
23:57
Speaker 2: Yeah, of course we always need guardrails on the things
24:00
Speaker 2: that will worry us, but we've got to create a
24:02
Speaker 2: space for people to innovate and do different things. Now,
24:05
Speaker 2: those are all my jobs. The thing that is tougher
24:07
Speaker 2: is the thing that Elon talked about, which is culture. Right,
24:09
Speaker 2: It's how do you transpose that culture from places like
24:12
Speaker 2: Silicon Valley across the world where people are unafraid to
24:17
Speaker 2: give up the security of a regular paycheck to go
24:20
Speaker 2: and start something and be comfortable with failure.
24:22
Speaker 3: You talk about that a lot.
24:23
Speaker 2: I think you talked about it more in when you're
24:25
Speaker 2: playing games, right, But you've got to be comfortable failing
24:29
Speaker 2: and knowing that.
24:29
Speaker 3: That's just part of the process.
24:31
Speaker 2: And that is as a tricky cultural thing to do overnight,
24:34
Speaker 2: but it's an important part of I think creating that.
24:36
Speaker 3: Kind of environment.
24:37
Speaker 4: Yeah, if you don't succeed with your first startup, it
24:40
Speaker 4: shouldn't be a sort of a catastrophic career ending exactly thing,
24:43
Speaker 4: it should be you know, well good I think should
24:46
Speaker 4: like should be like, well, you know, you gave it
24:48
Speaker 4: a good shot, you know, in now try again and exactly. Yeah.
24:51
Speaker 4: And it's so one thing I'm going to mention is,
24:53
Speaker 4: like I've seen, creating a company is sort of a
24:55
Speaker 4: high risk, high reward situation. But I don't know quite
24:59
Speaker 4: what the how works In the UK, I think probably
25:01
Speaker 4: better than contenttle Europe, but I'm not sure how does
25:04
Speaker 4: the new give it. But if somebody's basically going to
25:06
Speaker 4: risk the life savings and with the beast story of
25:09
Speaker 4: startups fail, so I mean you hear about the startups
25:12
Speaker 4: that succeed, but most companies are most startups consist of
25:16
Speaker 4: you know, a massive amount of work followed by failure.
25:18
Speaker 4: But that's actually most companies, and so it's a high risk,
25:21
Speaker 4: high reward and so the higher reward part does need
25:24
Speaker 4: to be there for it to make sense.
25:25
Speaker 2: I think that was a very soft pitch for a
25:28
Speaker 2: tax policy. I mean, but actually I can tell you
25:31
Speaker 2: so like AI agree, and we have so we have
25:34
Speaker 2: I think relative to certainly European countries, but certainly the
25:37
Speaker 2: US is definitely California a much lower rate of capital
25:40
Speaker 2: gains tax, right, So for those people who are risking
25:43
Speaker 2: and growing something like, we think the reward should be
25:45
Speaker 2: there at the end. So it's twenty percent capital gains
25:47
Speaker 2: tax right.
25:48
Speaker 3: And on stock options, I.
25:49
Speaker 2: Don't know if we've got anyone from Index Ventures in
25:52
Speaker 2: the room. So Index one of our bleeding VC funds here, Okay,
25:56
Speaker 2: they do a regular report looking at most countries tax
26:00
Speaker 2: moment of stock options. And you know, when I was
26:03
Speaker 2: a Chancellor of Treasury sexual equivalent, you know, we were
26:06
Speaker 2: I think down we were pretty good, but we were
26:08
Speaker 2: fourth or fifth, and I said we need to for
26:10
Speaker 2: exactly the reason that you mentioned. I was like, this
26:12
Speaker 2: has got to be the best place for renovative. We
26:14
Speaker 2: need to move that up and I think in the
26:15
Speaker 2: last iteration of that report we had because of the
26:18
Speaker 2: changes that Jeremy and I had made, we have moved
26:20
Speaker 2: up to I think second from memory. Hopefully that actould
26:23
Speaker 2: give you and everyone else some comfort that we recognize.
26:26
Speaker 2: That's important because when people work hard and risk things, yeah,
26:29
Speaker 2: they should be able to.
26:29
Speaker 4: Enjoy the rewards and pigh award. Yeah.
26:31
Speaker 2: And I think we have a We very much have
26:33
Speaker 2: a tax system that supports that. And those are the
26:36
Speaker 2: values that you know, I believe in, and I think
26:38
Speaker 2: most of us in this room probably do as well. Right, next,
26:41
Speaker 2: next question, I've got seven front of me, and then
26:42
Speaker 2: I'll come over here.
26:43
Speaker 4: Gone, thanks so much.
26:45
Speaker 6: We've talked about some really big ideas, global changing ideas.
26:48
Speaker 6: I'm really interested, particularly in the context of creation of
26:51
Speaker 6: science and technology superhubs and so on. How does that
26:53
Speaker 6: map onto the everyday lives of people living in say Austin, Texas,
26:58
Speaker 6: to choose not aroundin the more in my case Nottingham,
27:00
Speaker 6: East Midlands. What is how do you see that evolving
27:03
Speaker 6: for people you know every.
27:05
Speaker 4: Day sort of every day effective AI for.
27:07
Speaker 2: Context, Elon so Seib runs are equivalent of CDs, right
27:11
Speaker 2: or Walgreens. So you know, as I visited, right, so
27:13
Speaker 2: he's got millions of people coming in the shops every day,
27:16
Speaker 2: and it's making sure how do we make this relevant?
27:18
Speaker 2: I think that your question how is this relevant to
27:21
Speaker 2: that person? You know, maybe actually let me go I'll
27:23
Speaker 2: go first on that because I think it's a fair point.
27:25
Speaker 2: I was just going over with the team a couple
27:27
Speaker 2: of things that we're doing because I was saying, how
27:29
Speaker 2: are we doing AI right now that it's making a
27:31
Speaker 2: difference to people's lives. And we have this thing called
27:34
Speaker 2: gov dot which is which actually when we when it
27:36
Speaker 2: happened several years ago, was a pioneering thing all the
27:39
Speaker 2: government information together on one website, so you need to
27:43
Speaker 2: get a driving license, passport, any interaction with government. It
27:46
Speaker 2: was centralized in a very easy, relatively easy to use way.
27:50
Speaker 3: Better than most.
27:51
Speaker 2: So we're about to deploy AI across that platform. So
27:54
Speaker 2: that is something that I think you know, several million
27:57
Speaker 2: people a day use, right, So a large chunk of
27:59
Speaker 2: the popular is interacting with gov dot UK every single
28:02
Speaker 2: day to do all these day to day tasks. Right,
28:04
Speaker 2: every one of your customers is doing all those things,
28:07
Speaker 2: and so we're about to deploy AI into that to
28:10
Speaker 2: make that whole process so much easier, because you know,
28:13
Speaker 2: some people will be like, look, well I'm currently here
28:16
Speaker 2: and I've lost my passport and my flights in five hours.
28:19
Speaker 2: At the moment, that would require you know how many
28:21
Speaker 2: steps to figure out what you do. Yeah, Actually, when
28:24
Speaker 2: we deploy the AI, it should be that you could
28:25
Speaker 2: just literally say that and boom, this is what we're
28:28
Speaker 2: going to do, walk you through it, and that's going
28:30
Speaker 2: to benefit millions and millions of people every single day, right,
28:33
Speaker 2: Because that's a very practical way in my seat, that
28:36
Speaker 2: I can start using this technology to help people in
28:38
Speaker 2: their day to day lives, not just healthcare discoveries and
28:41
Speaker 2: everything else that we're also doing. But I thought that's
28:43
Speaker 2: quite a powerful demonstration of literally your day to day
28:47
Speaker 2: customer seeing actually their just day to day life get
28:50
Speaker 2: a little bit easier because of something that you know,
28:52
Speaker 2: Elon Demis and others in this room of help.
28:53
Speaker 4: Create Yeah, exactly. The most immediate thing is just being
28:57
Speaker 4: able to ask, like having a very smart friend that
28:59
Speaker 4: you can ask anything, you know, ask how to make something,
29:02
Speaker 4: how to solve any problem, and it'll tell you so.
29:05
Speaker 4: And obviously companies are going to adopt this, so I
29:08
Speaker 4: think maybe you'll have much better customer service. I guess
29:10
Speaker 4: essentially that'll probably be the first thing you notice. And
29:13
Speaker 4: then if we talked about education, so having a tutor,
29:16
Speaker 4: so if you're trying to understand a subject, like having
29:20
Speaker 4: a phenomenal tutor on any subject, is that that's really
29:23
Speaker 4: pretty much there already almost. I mean we need to obviously,
29:26
Speaker 4: I need to stop hallucinating before you know it can't
29:29
Speaker 4: give you. I mean that we still have a little
29:31
Speaker 4: bit of the problem where it can give you an
29:33
Speaker 4: answer that's confidently wrong with great grammar, you know, bullet
29:36
Speaker 4: points and everything in citations. It was not real. So
29:38
Speaker 4: that's to be Okay, we need to make sure it's
29:40
Speaker 4: not it's not it's not giving you confidently wrong tutor answers.
29:43
Speaker 4: But but that's going to happen pretty quickly where it
29:46
Speaker 4: is actually correct.
29:47
Speaker 2: Yeah, say for any parent who was homeschooling and realizing
29:50
Speaker 2: what their kids needed to be helped with, yeah, that will.
29:53
Speaker 3: Come as an enormous relief.
29:54
Speaker 4: I think they're very good.
29:55
Speaker 2: Right, have we got let's go questions over here, who
29:57
Speaker 2: have got only microphones or brand that perfect.
30:01
Speaker 7: Brent Herman, So you know you've spoken eloquently about abundance
30:04
Speaker 7: in the Age of abundance, so it feels obviously with
30:07
Speaker 7: AI it's everything everywhere, all at once, but with robots,
30:11
Speaker 7: and to get the age of abundance will need a
30:13
Speaker 7: lot of robots. I know you're working out on robots
30:15
Speaker 7: as well. Are there sort of constraints that we should
30:17
Speaker 7: think of our politician should be thinking of that we
30:20
Speaker 7: might get one country might get heavily behind in robots
30:23
Speaker 7: that can do all these things and end to the
30:24
Speaker 7: age of the bunds and therefore be it a strategic disadvantage.
30:26
Speaker 4: Well, really, anything that can be actuated by a computer
30:29
Speaker 4: is effectively a robot. So you can think of frankly
30:32
Speaker 4: Tesla cars or robots on wheels. Anything that's connected to
30:36
Speaker 4: the Internet is effectively an endpoint actuator for artificial intelligence.
30:40
Speaker 4: So you've got Bust and Dynamics. Obviously they've been making
30:43
Speaker 4: impressive robots for a while. I think they are at
30:45
Speaker 4: this point mostly owned by Hundai, So I guess I
30:48
Speaker 4: was probably gonna make robots that are humanoid and some
30:52
Speaker 4: interesting shapes that I wasn't just beating, like the one
30:54
Speaker 4: that looks like a has wheels and looks sort of
30:56
Speaker 4: like a kangaroo on wheels. I'm sure what that is,
30:58
Speaker 4: but it's a little demanded, frankly, but there's going to
31:00
Speaker 4: be all sorts of robots. You've got the company Dison,
31:03
Speaker 4: in which I think there's some pretty impressive things, and
31:07
Speaker 4: I think the UK will not be behind. Actually on
31:09
Speaker 4: that front, the UK also has ARM, which is really
31:12
Speaker 4: the best one, one of the best peraps, the best
31:15
Speaker 4: in trip design in the world. Tesla uses a lot
31:18
Speaker 4: of ARM technology. Almost everyone is actually, so I have
31:21
Speaker 4: the UK is in a strong position. Germany obviously makes
31:24
Speaker 4: a lot of robots, industrial robots. I mean, I think
31:27
Speaker 4: generally countries that make robots of any kind, even if
31:30
Speaker 4: they seem somewhat conventional, will be fine. I do think
31:33
Speaker 4: there is a safety concerned, especially with humanoid robots, because
31:38
Speaker 4: at least the car can't chase you into this building,
31:41
Speaker 4: not very easily, you know, or chase you over a tree,
31:43
Speaker 4: or you know, you can sort of run up a
31:44
Speaker 4: flight of stairs and get away from a tesla. I
31:46
Speaker 4: think it's a Stephen King movie about that. If your
31:48
Speaker 4: car gets possessed so, but if you have a humanoid robot,
31:51
Speaker 4: it can basically chase you anywhere. So I think we
31:53
Speaker 4: should have some kind of hardwired local cutoff that you
31:57
Speaker 4: can't update from the internet. So anything that could be
32:00
Speaker 4: after updated from the internet obviously could be overridden. But
32:02
Speaker 4: if you have a local sort of off switch where
32:04
Speaker 4: you have to say a keyword or something, and then
32:06
Speaker 4: that puts the robot into a safe state, it's some
32:09
Speaker 4: kind of localized safe state ability and off switch, you know,
32:13
Speaker 4: where you don't have to get too close to the robot.
32:14
Speaker 4: I don't know. So we've got millions of these things
32:16
Speaker 4: going over the place.
32:17
Speaker 3: You're not selling it, just you know, I know.
32:20
Speaker 4: I'm saying this is something we should be quite concerned
32:22
Speaker 4: about because Robert can follow you anywhere. Then you know
32:25
Speaker 4: what if they just one day get a software update
32:27
Speaker 4: and they're not so friendly anymore. We've got a James
32:30
Speaker 4: Cameron movie on our house.
32:31
Speaker 2: It's actually, that's it's funny you're saying that, because in
32:33
Speaker 2: our session.
32:33
Speaker 3: That we had today, I just I would say, who they.
32:37
Speaker 2: Made exactly the same point right then, So we're talking
32:40
Speaker 2: about they talking about movies, actually about mentioning James Cameron.
32:42
Speaker 2: They're talking about James cameraon movies the same. If you
32:45
Speaker 2: think about it, it's not just those movies, but any of
32:48
Speaker 2: these movies, trains, subways, metros. They said, all these movies
32:53
Speaker 2: with the same plot fundamentally all and with the person
32:56
Speaker 2: turning it off right or finding a way to shut
32:59
Speaker 2: the thing down. And they were making the same point
33:01
Speaker 2: that you were about the importance of actual.
33:03
Speaker 3: Physical off switches.
33:05
Speaker 2: Yeah, and so all the technology is great, but undamentally,
33:07
Speaker 2: the same movie has played out fifty times. We've all
33:10
Speaker 2: watched it and it's all fundamentally you know, you know
33:12
Speaker 2: the point I'm revenge right, all ends in pretty much
33:14
Speaker 2: the same way, with someone finding their way to just
33:17
Speaker 2: do the wrong. Which is kind of interesting that you've
33:19
Speaker 2: said a similar point. Right, it's not the it's not
33:21
Speaker 2: the obvious place you'd go to.
33:23
Speaker 4: But I be one of the tests for the AI,
33:25
Speaker 4: which is so like blank is your favorite Gamers camera movie? Blank?
33:29
Speaker 3: Excellent? Right, Yes, we got over there, yep, perfect, Hi.
33:32
Speaker 4: Question for you both.
33:33
Speaker 8: So I'm a founder of a AI and mL scale
33:36
Speaker 8: up in the third Center for AI which is leads
33:38
Speaker 8: in the North of England and bit biased. Since the
33:40
Speaker 8: launch of chat GBT. Three months after that, we saw
33:43
Speaker 8: a real increase in phishing attacks using much more sophisticated
33:47
Speaker 8: language patterns. What do we do to protect businesses consumers
33:51
Speaker 8: they trust this technology better, and how do we bring
33:54
Speaker 8: them along that journey with us?
33:55
Speaker 4: Well, I think we shouldn't trusted that much. Actually, it
33:58
Speaker 4: is actually quite quite a significant allenge because we're getting
34:00
Speaker 4: to the point where even open source AI can pass
34:03
Speaker 4: human capture tests. So you know, this is are you're
34:06
Speaker 4: human identify all the traffic lights in this picture. You're like, okay, yeah,
34:10
Speaker 4: it's going to have no problem doing that. In fact,
34:12
Speaker 4: it'll do it better than the human and faster than human.
34:14
Speaker 4: So we're like, how do you know it's the point
34:16
Speaker 4: which is a better human better passing human tests than humans?
34:20
Speaker 4: Then well, what tests actually make sense? That is a
34:22
Speaker 4: real problem. I don't actually have a good solution to it.
34:25
Speaker 7: That.
34:25
Speaker 4: One of the things we're trying to figure out on
34:27
Speaker 4: the X platform is how to deal with that, because
34:30
Speaker 4: it really we really are at the point where even
34:32
Speaker 4: with open source you know, readily available AI, you don't
34:35
Speaker 4: need to be sort of leading in the field. You
34:37
Speaker 4: can actually be better than humans at passing these tests,
34:40
Speaker 4: and that's sort of why we think, well, perhaps we
34:42
Speaker 4: should sort of charge a dollar or a pound a year.
34:44
Speaker 4: It's a very tiny amount of money, but it's still
34:46
Speaker 4: makes it privatively expensive to make a million bots, So
34:50
Speaker 4: especially if you need a million payment methods, then you
34:52
Speaker 4: run out of sort of stolen credit cards pretty quickly.
34:55
Speaker 4: So that's that's sort of where we're thinking, like we
34:57
Speaker 4: might have to sort of just charge some very tiny
34:59
Speaker 4: amount of money three cents a day effectively to deal
35:02
Speaker 4: with the onslought of AI power advance. And that is
35:07
Speaker 4: not a growing problem, but it will be I think,
35:08
Speaker 4: perhaps an insurmountable problem next year. So and then you
35:11
Speaker 4: have to worry about, well, manipulation of information is making
35:15
Speaker 4: something seem very popular when in fact it it's not,
35:17
Speaker 4: because it's getting boosted by all these likes and reposts
35:21
Speaker 4: from AI powered barts. So that's why I sort of
35:23
Speaker 4: think somewhat inevitably it leads to some small payment in
35:27
Speaker 4: order to dramatically increase the cost of a bot. So
35:31
Speaker 4: I frankly I think probably any social media system that
35:33
Speaker 4: doesn't do that will simply be over OneD by bot.
35:35
Speaker 2: You know, I think my general answer would be you know,
35:38
Speaker 2: we need to show that we are on top of
35:41
Speaker 2: mitigating the risks right so people can trust the technology.
35:44
Speaker 2: That's what actually the last couple of days has been
35:46
Speaker 2: about on the Safety Summit is just showing, you know,
35:49
Speaker 2: we're investing in the Safety Institute, having the people who
35:52
Speaker 2: can do the research on these things to figure out
35:54
Speaker 2: how we mitigate against them, and we have to do
35:57
Speaker 2: it fast and we have to keep iterating it is.
36:00
Speaker 2: All of us probably in this room, believe that the
36:02
Speaker 2: technology can be incredibly powerful, but we've got to make
36:04
Speaker 2: sure we bring people along that journey with us, that
36:06
Speaker 2: we're handling the risks that are there and as there's
36:09
Speaker 2: a job to do, and the last couple of days,
36:12
Speaker 2: I think we make good progress on it because we
36:14
Speaker 2: want to focus on the positives and manage these things.
36:16
Speaker 2: But that requires action, and that's what the last couple
36:19
Speaker 2: of days has been about. Your story, you know, analogy.
36:22
Speaker 2: There was part of the research that actually, you know,
36:25
Speaker 2: the team working on the task force here published and
36:29
Speaker 2: presented yes day. I don't know if you saw it was,
36:31
Speaker 2: which is essentially that it was using AI to do
36:35
Speaker 2: to create a ton of fake profiles on social media
36:39
Speaker 2: and then infiltrate particular groups with particular information. And actually,
36:43
Speaker 2: at the moment that, as I said, to your point,
36:45
Speaker 2: and there's a cost free.
36:47
Speaker 4: It's getting to the point where it's like, really, you're
36:48
Speaker 4: going to have one hundred for a penny sort of thing. Ridiculous.
36:51
Speaker 2: And if you think about some of these social networks
36:52
Speaker 2: at quite a neighborhood or town level, it's not that
36:55
Speaker 2: many fake profiles. Quickly suddenly they're everywhere, and there's some
36:59
Speaker 2: local issue that might be of importance, and you know,
37:01
Speaker 2: the team of run versions of how that would look like,
37:04
Speaker 2: and suddenly they're interacting with everybody and then spreading and
37:06
Speaker 2: misinformation around. Yeah, challenge we literally as part of the
37:09
Speaker 2: research that we published on this information yesterday. It's a
37:12
Speaker 2: real challenge.
37:13
Speaker 4: Yeah, exactly to your point. I mean, the images. You
37:17
Speaker 4: don't even need to steal somebody's picture because then that's traceable.
37:20
Speaker 4: But you can actually just say, create a new image
37:23
Speaker 4: of a person, realistic looking it doesn't exist, and then
37:26
Speaker 4: create a biography realistic but doesn't exist, and do that
37:30
Speaker 4: en mass and practically the only way of able to
37:32
Speaker 4: tell us that the grammar is too good. Don't give
37:34
Speaker 4: away typhos come on.
37:36
Speaker 3: Now, I'm getting waved at because I think we are
37:38
Speaker 3: out of time. I don't we take one.
37:39
Speaker 2: Very brief last question and let's make it go on? Yes, sir, going,
37:43
Speaker 2: you're right in front of me, go on.
37:44
Speaker 9: Question for you related to the X platform. Are there
37:47
Speaker 9: simple things we can do, especially when it comes to
37:49
Speaker 9: visual media. You alluded to the fact that it's fairly
37:51
Speaker 9: straightforward and effectively free to make people like yourselves say
37:55
Speaker 9: and do things that you never said or did. Can
37:57
Speaker 9: we do something like cryptographically signed media? I'm from Adobe
38:00
Speaker 9: working on this project. Twitter was a member. Love to
38:02
Speaker 9: see X come back. Digitally sign media to indicate not
38:05
Speaker 9: only what was created by AI, but what came from
38:07
Speaker 9: a camera, what was real to imview, a sense of
38:09
Speaker 9: trust in media that can go viral.
38:11
Speaker 4: That sounds like a good idea. Actually, so if some
38:14
Speaker 4: way of authenticating would be good. So, yeah, that sounds
38:17
Speaker 4: like a good idea, which probably do it.
38:18
Speaker 2: There you go, and actually on that on that point already,
38:21
Speaker 2: And this is particularly partner for people in my job,
38:24
Speaker 2: right and I've already had a situation happened to me
38:27
Speaker 2: with adopted image that goes everywhere negative.
38:30
Speaker 3: By the time everyone realizes well.
38:31
Speaker 2: That's fake and we should stop sending it the damages damage.
38:35
Speaker 2: And actually we were again reflecting today. If you think
38:38
Speaker 2: next year, you've got elections in you know, I think
38:41
Speaker 2: you know, the US, India, I think Indonesia probably here
38:44
Speaker 2: there you go for massive news and actually you've got
38:48
Speaker 2: just an enormous junk of the world's population is voting
38:51
Speaker 2: next year, right, and you've got EU elections as well.
38:54
Speaker 4: You know.
38:55
Speaker 2: Actually, just these issues are right in front of ours.
38:58
Speaker 2: Next year is where a big election across the globe,
39:00
Speaker 2: probably the first set of elections where this has been
39:03
Speaker 2: a real issue. Yeah, So figuring out how we manage
39:06
Speaker 2: that is I think kind of mission critical for the
39:09
Speaker 2: people who want the integrity of our democracy.
39:11
Speaker 4: Yeah, I mean some of it it's quite entertaining, Like
39:14
Speaker 4: the pope and the puffer jacket. Have you seen that one?
39:16
Speaker 4: That's amazing. But I mean I still write too people
39:18
Speaker 4: who think that's real. I'm like, well, one of the
39:20
Speaker 4: answer is wearing a puffer jacket in July, right, be sweating,
39:23
Speaker 4: but it actually looks quite quite dashing of things. In fact,
39:27
Speaker 4: I think AI fashion is going to be a real thing.
39:28
Speaker 4: So a bit of doing gloom, like we learn in
39:30
Speaker 4: the most interesting times. And I think this is it is,
39:33
Speaker 4: you know, like eighty percent likely to be good and
39:35
Speaker 4: twenty percent bad. I think if we're cognizant and careful
39:38
Speaker 4: about the bad part, on balance, actually it will be
39:41
Speaker 4: the future that we want or for the future that
39:43
Speaker 4: is preferable, and it actually will be somewhat of a leveler,
39:47
Speaker 4: an equalizer in the sense that you know, I think
39:49
Speaker 4: everyone will have access to goods and services and education,
39:52
Speaker 4: and so you know, I think probably it leads to
39:54
Speaker 4: more human happiness. So I guess i'd probably leave on
39:57
Speaker 4: an optimistic note.
39:58
Speaker 3: Tough act. Yeah, well, that's it.
40:00
Speaker 2: That is a great note to end on. I think
40:01
Speaker 2: that we all want that better future. I think it's
40:05
Speaker 2: that the promise of it is certainly there. Lots of
40:06
Speaker 2: people in this room, including yourselves, are working hard to
40:09
Speaker 2: make it happen. Our job in government is to make
40:10
Speaker 2: sure it happens safely. But on the basis of this
40:13
Speaker 2: conversation in the last couple of days, I'm certainly leaving
40:16
Speaker 2: more confident that we can make that happen. It's been
40:18
Speaker 2: a huge privilege and pleasure to have you here.
40:20
Speaker 4: Thank you very much.
40:21
Speaker 1: For me, Thanks for listening, See you in the next episode.