Latest / What's Your Problem? / Will AI Radically Change the World by 2027?... from Risky Business

Transcript

Will AI Radically Change the World by 2027?... from Risky Business · What's Your Problem?

0:15
Speaker 1: Pushkin.
0:20
Speaker 2: Hey, it's Jacob.
0:21
Speaker 1: There's another Pushkin show called Risky Business. It's a show
0:25
Speaker 1: about decision making and strategic thinking. It's hosted by Nate
0:29
Speaker 1: Silver and Riya Kanakova, and they just had a really
0:32
Speaker 1: interesting episode about the future of AI. So we're going
0:36
Speaker 1: to play that episode for you right now. We'll be
0:39
Speaker 1: back later this week with the regular episode of What's
0:42
Speaker 1: Your Problem.
0:48
Speaker 3: Welcome back to Risky Business, a show about making better decisions.
0:52
Speaker 3: I'm Maria Kanakova.
0:53
Speaker 2: And I'm Nate Silver.
0:54
Speaker 3: Hey. Today on the show is going to be a
0:56
Speaker 3: little bit doomtastic.
0:58
Speaker 2: Yeah, I mean, I don't know if it's like worse
1:01
Speaker 2: and thinking about like the global economy going into a
1:05
Speaker 2: recession because of dumb fuck tariff policies about how we're
1:09
Speaker 2: all going to die in seven years and staid no,
1:11
Speaker 2: I'm just kidding. This is a very very intelligent and
1:14
Speaker 2: well written and thoughtful report called AI twenty twenty seven
1:18
Speaker 2: that we're to spend the whole show on I because
1:20
Speaker 2: think it's such an interesting to talk about, but that
1:23
Speaker 2: you know, includes some dystopian possibilities. I would say it does.
1:28
Speaker 3: Indeed, So let's get into it and hope that you
1:32
Speaker 3: guys are all still here to listen to us in
1:34
Speaker 3: seven years.
1:38
Speaker 2: The contrast is interesting between like all the chaos we're
1:43
Speaker 2: seeing with tariff policy in terms of starting a trade
1:47
Speaker 2: war with China and then other types of chaos. It's
1:51
Speaker 2: interesting to kind of look at this. I mean, I
1:53
Speaker 2: wouldn't call it a more optimistic future exactly, like but
1:58
Speaker 2: like on a different trajectory of like a future that's
2:00
Speaker 2: going to change very fast, according to these authors, with
2:04
Speaker 2: profound implications for you know, everything, the humans species. These
2:11
Speaker 2: researchers and authors are saying that everything is going to
2:14
Speaker 2: change profoundly.
2:15
Speaker 3: And.
2:18
Speaker 2: Even though there is some hedging hair, this is kind
2:22
Speaker 2: of their base case scenario. And like you know, base
2:26
Speaker 2: case number one and base case number two differ. There's
2:28
Speaker 2: like kind of a choose your own adventure at some
2:30
Speaker 2: point in this report, but they're both very different than
2:33
Speaker 2: the status quo, right, and the notionately hear from that,
2:36
Speaker 2: like everything becomes different If ais become substantially more intelligent
2:42
Speaker 2: than human beings. People can debate and we will debate
2:44
Speaker 2: on this program what that means. But yeah, do you
2:46
Speaker 2: want to do you want to contextualize this for me?
2:48
Speaker 2: Do you want to tell people too. Yeah, there's are
2:49
Speaker 2: of this report.
2:50
Speaker 3: Absolutely absolutely. So the report is authored officially by five people,
2:56
Speaker 3: but I think unofficially there's also a sixth. We've got
3:00
Speaker 3: Eli Leiflund, who's a super forecaster and he was ranked
3:05
Speaker 3: first on Rand's forecasting initiative, so he is someone who
3:10
Speaker 3: is very good at kind of looking at the future
3:13
Speaker 3: trying to predict what's going to happen. You have Jonas Walmer,
3:16
Speaker 3: who's a VC at Macroscopic Ventures. Thomas Larson was the
3:21
Speaker 3: former executive director of the Center for AI Policy, so
3:26
Speaker 3: that is a center that advises both sides of the
3:30
Speaker 3: aisle on how AI is going to go. And Romeo Dean,
3:34
Speaker 3: who is part of Harvard's AI Safety Student Team, so
3:39
Speaker 3: someone who is still a student, still learning, but kind
3:42
Speaker 3: of the next generation of people looking at AI. And
3:48
Speaker 3: finally we have Daniel Cochtaylo, who basically had written a
3:53
Speaker 3: report back in twenty twenty one. He was a AI
3:57
Speaker 3: researcher and he looked at predictions for AI for twenty
4:01
Speaker 3: twenty six and it turns out that his predictions were
4:04
Speaker 3: pretty spot on, and so Open AI actually hired Daniel
4:09
Speaker 3: as a result of this report.
4:10
Speaker 2: Obviously he left, yes, and.
4:12
Speaker 3: Then and then he left exactly.
4:14
Speaker 2: And importantly, So there's also Scott Alexander exactly.
4:17
Speaker 3: So I was saying, he's the he's the person who
4:20
Speaker 3: kind of is in the background, and he's you guys
4:23
Speaker 3: might know him as the author behind astral Codex.
4:27
Speaker 2: And I know Scott. He's one of the kind of
4:29
Speaker 2: fathers of what you might call rationalism. I think Scott,
4:36
Speaker 2: when I interviewed him from a book, was happy enough
4:38
Speaker 2: with that term and accused me or co opted me
4:40
Speaker 2: into also being a rationalist. These people are somewhat adjacent
4:45
Speaker 2: to the effective altruist, but not quite right. They're just
4:48
Speaker 2: trying to apply a sort of thoughtful, rigorous, quantitative lens
4:53
Speaker 2: to big picture problems, including existential risk, of which most
4:57
Speaker 2: people in this community believe that AI is both an
5:00
Speaker 2: existential risk and also kind of an existential opportunity, right,
5:04
Speaker 2: that it could transform things. You talk to Sam Alp
5:06
Speaker 2: and he'll say, we're going to cure cancer and elimina
5:09
Speaker 2: poverty and whatever else. Right. And Scott's also an excellent writer.
5:13
Speaker 2: And so let me disclose something which is slightly important here.
5:17
Speaker 2: So I actually was approached by some of the authors
5:20
Speaker 2: of this report a couple of months ago, I guess
5:22
Speaker 2: it was in February ish, just to give feedback and
5:26
Speaker 2: chat with them. So I'm working off the draft version right,
5:29
Speaker 2: which I do not leave. They changed very much, so
5:32
Speaker 2: my notes pertained to an earlier draft. I did not
5:34
Speaker 2: have time this morning to go back and re.
5:35
Speaker 3: Read it, so I was not on the inside loops.
5:39
Speaker 3: So I did not get an earlier draft. And I've
5:42
Speaker 3: read this draft and basically just to kind of big
5:46
Speaker 3: picture sum it up, it outlines two scenarios, right, two
5:51
Speaker 3: major scenarios for how AI might change the world as
5:56
Speaker 3: soon as twenty thirty. Now important note like that that
6:02
Speaker 3: date is could have hedged. It might be sooner, it
6:05
Speaker 3: might be later. There's kind of a there's a confidence
6:08
Speaker 3: interval there. But the two different scenarios. One basically we
6:13
Speaker 3: do twenty thirty, humanity disappears and is taken over by AI.
6:18
Speaker 3: The positive report is in twenty thirty basically we get
6:23
Speaker 3: AIS that are aligned to our interests, and we get
6:26
Speaker 3: kind of this AI utopia where AIS actually help make
6:30
Speaker 3: life much better for everyone and make the standard of
6:33
Speaker 3: living much higher. But the crucial turning point is before
6:36
Speaker 3: twenty thirty. And the crucial kind of question at the
6:39
Speaker 3: center of this is will we be able to design
6:45
Speaker 3: AIS that are truly aligned to human interests rather than
6:50
Speaker 3: just appear to be aligned and kind of lying to
6:53
Speaker 3: us while actually following their own agenda. And how we
6:57
Speaker 3: handle that is kind of the lynchpin. And it's actually interesting,
7:01
Speaker 3: Nate that you started out with China, because a lot
7:03
Speaker 3: of the policy choices and a lot of what they
7:06
Speaker 3: see as kind of the decision points will affect the
7:09
Speaker 3: future of humanity actually hinge on the US China dynamic,
7:15
Speaker 3: how they compete with each other, and how that sometimes
7:19
Speaker 3: might basically clash against safety concerns because no one wants
7:24
Speaker 3: to be left behind. Can we manage that effectively and
7:27
Speaker 3: can kind of that transition work in our favor as
7:31
Speaker 3: opposed to against us. I think that this is kind
7:33
Speaker 3: of one of the big questions here, And so it's
7:35
Speaker 3: funny that we're seeing all of this trade war right
7:37
Speaker 3: now as the sideport is coming out.
7:41
Speaker 2: Yeah. Look, I think this exercise is partly just a
7:46
Speaker 2: forecasting exercise, right, I mean, obviously it's just kind of
7:49
Speaker 2: like fork at the bottom where we learn to have
7:53
Speaker 2: an AI slow down or we kind of are pressing
7:57
Speaker 2: fully on the accelerator, right, Like, in some ways, scenarios
7:59
Speaker 2: are like not that different, right, either one assumes remarkable
8:05
Speaker 2: rates of technological growth that I think even AI I'm
8:09
Speaker 2: never quite sure who to call an optimist or a pessimist, right,
8:12
Speaker 2: even AI believers you know, might think is a little
8:15
Speaker 2: bit aggressive. Right, But what they want to do is
8:17
Speaker 2: they want to have like a specific, fleshed out scenario
8:22
Speaker 2: for how the world would look like. It's kind of
8:24
Speaker 2: like a modal scenario. And like, I think they'd say that, like,
8:29
Speaker 2: we're not totally sure about either of these necessarily, right,
8:32
Speaker 2: And I don't think they'd be as like pedantic as
8:34
Speaker 2: to say, if you do X, Y and Z, then
8:37
Speaker 2: we'll save the world and have utopia, and if you don't,
8:39
Speaker 2: then we'll all die, Right. I think they'd probably say
8:42
Speaker 2: it's unclear and there's kind of like risk either way.
8:45
Speaker 2: We wanted to go through the scenario of like fleshing
8:47
Speaker 2: out like what the world might look like, right. I
8:51
Speaker 2: do think one thing that's important is that whatever decisions
8:54
Speaker 2: are made now could get locked in, right that you
8:58
Speaker 2: pass certain points in overturn and it becomes very hard
9:03
Speaker 2: to decelerate like an arms race. This is you know
9:06
Speaker 2: what we found during the Cold War for example. I
9:09
Speaker 2: mean one of the big things I look at is like,
9:11
Speaker 2: do we force the AI to be transparent in its
9:15
Speaker 2: thinking with humans? Right? Like, now there's been a movement
9:18
Speaker 2: toward the ai'll actually explicate it's thinking more. I'll ask
9:22
Speaker 2: it a query open AI. The Chinese models to this too, right,
9:25
Speaker 2: and I'll say, I am thinking about X, Y and Z,
9:27
Speaker 2: and I'm looking up PD and Q and now I'm
9:30
Speaker 2: reconsidering this. It's actually has this chain of thought process, right,
9:33
Speaker 2: which is explicated in English. You know. One concern is
9:36
Speaker 2: that what if the AI just kind of communicates to
9:39
Speaker 2: another in these implicit vectors that's inferring from all the
9:44
Speaker 2: texts it has. It's kind of unintelligible to human beings, right,
9:48
Speaker 2: and maybe kind of quote unquote thinking in that way
9:50
Speaker 2: in the first place, and then does us the favor
9:52
Speaker 2: of like translating back so it it goes from English
9:55
Speaker 2: to kind of this big bag of numbers is one
9:58
Speaker 2: a researcher called it, right, and then it translates it
10:02
Speaker 2: back into English or whatever land which you want. Really,
10:04
Speaker 2: in the end, what if it just skip cuts out
10:07
Speaker 2: that last step, right, then we can't kind of like
10:11
Speaker 2: check what AI is doing. Then it can behave deceptively
10:14
Speaker 2: more easily. So you know, so that part seems to
10:17
Speaker 2: be important. I want to hear your your first impressions
10:21
Speaker 2: before I kind of poison the well too much.
10:24
Speaker 3: Well, my first impressions is that the alignment problem is
10:28
Speaker 3: a very real one and an incredibly important one to solve.
10:32
Speaker 3: And what I got from this is that actually, the
10:35
Speaker 3: problem that I've had with like these initial AI lms
10:39
Speaker 3: is the kernel of what they're seeing there. Right, So
10:42
Speaker 3: you and I have talked about this on the show
10:44
Speaker 3: in the past, and I've said, well, my problem is
10:47
Speaker 3: that when I'm a domain expert, right, I start seeing
10:50
Speaker 3: some inaccuracies, and I start seeing like places where like
10:54
Speaker 3: it either just didn't do well or made shit up
10:57
Speaker 3: or whatever it is. Now, I think it's very clear
11:00
Speaker 3: that those problems aren't going to go away, right, that
11:03
Speaker 3: that is going to get much much better. However, the
11:07
Speaker 3: kernel of it's showing me some thing, but that might
11:10
Speaker 3: just be you know, I have no way of verifying
11:13
Speaker 3: if that's what's going on, what it's reading, like, how
11:16
Speaker 3: it's I don't want to say thinking about it, even
11:18
Speaker 3: though in the report they do use thinking, but it's normal.
11:22
Speaker 2: Vizing too much, I think is correct. I think it's.
11:24
Speaker 3: Okay, we'll stick to that language. Yeah, okay, So, so
11:28
Speaker 3: how it's thinking about it that those little problems and
11:33
Speaker 3: like the little the glitches and the things that it
11:36
Speaker 3: might be doing where it starts actually glitching on purpose
11:40
Speaker 3: are not going to be visible to the human eye.
11:43
Speaker 3: And so one of the main things that they say
11:45
Speaker 3: here is that as AI internal R and D gets
11:50
Speaker 3: rapidly faster, so that means basically AI's researching AIS, right,
11:55
Speaker 3: and so internally they start developing new models, and as
11:59
Speaker 3: they kind of surpass human ability to monitor it, it
12:03
Speaker 3: becomes progressively more difficult to figure out, Okay, is the
12:07
Speaker 3: AI actually doing what I want to do? Is the
12:10
Speaker 3: output that it's giving me its actual thought process, and
12:15
Speaker 3: is it accurate or is it like trying to deceive me?
12:18
Speaker 3: But it's actually kind of inserting certain things on purpose
12:21
Speaker 3: because it has different goals, right, because it is actually
12:24
Speaker 3: secretly misaligned, but it's very good at persuading me that
12:27
Speaker 3: it's aligned. Because one of the things that actually came
12:29
Speaker 3: out of this report and I was like, huh, you
12:31
Speaker 3: know this is interesting is if we get this remarkable
12:37
Speaker 3: improvement in AI, it will also remarkably improve at persuading us, right,
12:42
Speaker 3: so as part of making yeah, so this is this
12:46
Speaker 3: is but I've never even thought about that. I was like, okay, fine,
12:50
Speaker 3: But one of the things that I do buy is
12:52
Speaker 3: that it's going to be very difficult for us to
12:55
Speaker 3: monitor it and to figure out like is it truly
12:57
Speaker 3: aligned with human wants, with human desires, with human goals,
13:01
Speaker 3: And the experts who are capable of doing that, I
13:04
Speaker 3: think are actually going to dwindle right as AI starts
13:07
Speaker 3: proliferating in society. And so to me, that is something
13:11
Speaker 3: that is actually quite worrisome, and that is something that
13:13
Speaker 3: we really need to be paying attention to now. Just
13:15
Speaker 3: to fast forward a little bit, in their doomsday scenario
13:20
Speaker 3: in twenty thirty one, AI takes over it basically like
13:24
Speaker 3: suddenly releases some chemical agents, right, and humanity dies and
13:28
Speaker 3: the rest of the stragglers are taken care of by drones,
13:30
Speaker 3: et cetera. I don't even like it's it doesn't even.
13:34
Speaker 2: Quick and painless death.
13:35
Speaker 3: I will say, let's hope we don't know what the
13:38
Speaker 3: chemical agents are might not be quick and painless. Some
13:42
Speaker 3: chemical agents are actually a very painful death thing. So
13:45
Speaker 3: let's hope let's hope it's quick and painless. Quick, quick, yes,
13:49
Speaker 3: stuff quick? Okay, hopefully some chemical agents are not quick.
13:54
Speaker 3: Let's hope it's it's quick and painless. But if they're
13:58
Speaker 3: actually capable of deception at that high level, then you
14:02
Speaker 3: technically don't even need them to do it. If we're
14:05
Speaker 3: trusting medicine and all sorts of things to the AIS,
14:08
Speaker 3: it's pretty easy for it to actually manipulate something and
14:11
Speaker 3: actually insert something into codes, et cetera that will fuck
14:16
Speaker 3: up humanity in a way that we can't that we
14:19
Speaker 3: can't actually figure out at the moment, right, Like the
14:22
Speaker 3: way I think of it, And this is not from
14:24
Speaker 3: the this is not from the paper, but this is
14:26
Speaker 3: just the way that like my mind processed. It is
14:29
Speaker 3: like think about DNA, right, Like you have these remarkably complex,
14:34
Speaker 3: huge strands of data, and as we've found out, but
14:38
Speaker 3: it's taken forever, one tiny mutation can actually be fatal, right,
14:43
Speaker 3: but you can't spot that mutation. Sometimes that mutation isn't
14:47
Speaker 3: fatal immediately but will only manifest at a certain point
14:50
Speaker 3: in time. That's the way that my mind tried to
14:52
Speaker 3: kind of try to conceptualize what this actually means. And
14:56
Speaker 3: so I think that, you know, that would be easy
14:59
Speaker 3: for a deceptive AI to do and to me like that,
15:03
Speaker 3: that's kind of the big takeaway from this report is
15:06
Speaker 3: that we need to make sure that we are building
15:09
Speaker 3: AIS that will not deceive right that they're capabilities, they
15:15
Speaker 3: explain them in an honest way, and that honesty and
15:18
Speaker 3: trust is actually prioritized over other things, even though it
15:22
Speaker 3: might slow down research, it might slow down other things,
15:24
Speaker 3: but that that kind of alignment step is absolutely crucial
15:29
Speaker 3: at the beginning, because otherwise humans are human, right, they're
15:32
Speaker 3: easily manipulated, And we often trust that computers are quote
15:38
Speaker 3: unquote rational because they're computers, but they're not. They have
15:41
Speaker 3: their own inputs, they have their own weights, they have
15:43
Speaker 3: their own values, and that could just lead us down
15:47
Speaker 3: a dark path.
15:48
Speaker 2: Yeah, so let me follow up with this pushback. I
15:51
Speaker 2: guess right, Like, first of all, I don't know that
15:54
Speaker 2: humans are so easily persuaded. This is my big critique
15:58
Speaker 2: with like all the misinformation people who say, well, misinformation
16:03
Speaker 2: is the biggest problem that society basis. It's like people
16:06
Speaker 2: are actually pretty stubborn and they're kind of sound pretentious.
16:10
Speaker 2: They're kind of basian and how they formulate their beliefs, right,
16:13
Speaker 2: they have some notion of reality. They're looking at the
16:15
Speaker 2: credibility of the person who is telling them these remarks.
16:20
Speaker 2: If it's an unpersuasive source, it might make them less
16:22
Speaker 2: likely to believe they're balancing what other information with their
16:25
Speaker 2: lived experience so called Right. You know, part of the
16:29
Speaker 2: reason that like I am skeptical of AIS being super
16:33
Speaker 2: persuasive is like you know that it's an AI. You
16:36
Speaker 2: know it's trying to persuade you. You know what I mean.
16:38
Speaker 2: Like if you go and play poker against like a
16:40
Speaker 2: really chatty player, if Phil Hemmi with or Scott Sieber
16:42
Speaker 2: or someone like that, Right, you know, on some level
16:45
Speaker 2: of the best play is just to totally ignore it. Right,
16:49
Speaker 2: You know that they are trying to sweet talk you
16:51
Speaker 2: into doing exactly what they want you to do, and
16:55
Speaker 2: so the best play is to disengage or literally you
16:59
Speaker 2: can randomize your movis videos no notion of what the
17:01
Speaker 2: game theoretical optimal play might be, right, or salesmen or
17:06
Speaker 2: politicians have reputations for being Oh he's a little too smooth.
17:09
Speaker 2: Gavin newsome a little too fucking smooth. Right, I don't
17:12
Speaker 2: find Gavin news some persuasive at all. Right, use a
17:15
Speaker 2: little too from the hair gel to the constantly shifting vibes.
17:19
Speaker 2: I mean, I don't really find Gavin news some persuasive
17:20
Speaker 2: at all, even though like an AI, I'd say, boy,
17:22
Speaker 2: Gavin new some good looking guys look gravelly throated. But
17:25
Speaker 2: you know whatever, I mean. Look, the big critique I
17:28
Speaker 2: have of this project, and by the way, I think
17:31
Speaker 2: this is an amazing project in addition to like wonderful
17:35
Speaker 2: writing if you viewed on the web not your phone.
17:38
Speaker 2: So these very cool, like little infographic that updates everything
17:41
Speaker 2: from like the market value. If they don't call it
17:43
Speaker 2: open aye, they call it open brain. I guess is
17:45
Speaker 2: what they settle on for a substitute.
17:47
Speaker 3: Yeah, they call everything something else, just to make sure
17:50
Speaker 3: that they're not stepping on any toes. So they have
17:53
Speaker 3: open Brain, and then they have Deep Scent from China.
17:57
Speaker 2: I wonder which one that could be. I wonder. But
18:02
Speaker 2: it's beautifully presented and written, and like I appreciate they're
18:06
Speaker 2: going out on a limb here, you know, I mean,
18:09
Speaker 2: I think they have. It's been fairly well received. They've
18:11
Speaker 2: gotten some pushback both from inside and I think outside
18:15
Speaker 2: the AI safety community. Right, but they're putting their necks
18:18
Speaker 2: in the line, hear. They will look if things look
18:21
Speaker 2: pretty normal, if looks pretty normal in twenty thirty two
18:23
Speaker 2: or whatever, right, then they will look dumb for having
18:27
Speaker 2: published this.
18:29
Speaker 3: Well, and they actually have that right as a scenario
18:32
Speaker 3: that you end up looking stupid if everything goes well.
18:35
Speaker 3: But that's okay. Now, can I push back on the
18:38
Speaker 3: persuasion thing a little bit, just on two things. So
18:42
Speaker 3: first of all, the poker example is not actually a
18:45
Speaker 3: particularly applicable one here, because you know that you're playing
18:50
Speaker 3: poker and you know that someone is trying to get
18:52
Speaker 3: information and deceive you the tricky thing. So this is
18:55
Speaker 3: kind of when I spend time with con artists. The
18:57
Speaker 3: best con artists aren't Gavin Newsome, Like, they're not car salesmen.
19:01
Speaker 3: You have no idea they're trying to persuade you to
19:04
Speaker 3: do something. They are just like nice, affable people who
19:07
Speaker 3: are incredibly charismatic, and even in the poker community, by
19:10
Speaker 3: the way, like some of the biggest grifters who like
19:12
Speaker 3: it comes out later on, we're just like stealing money
19:16
Speaker 3: and doing all of these things. Are charming, right, They're
19:19
Speaker 3: not sleazy looking like they have no signs of, oh,
19:22
Speaker 3: I'm a salesman. I'm trying to sell you something. The
19:24
Speaker 3: people who are actually good at persuasion you do not
19:28
Speaker 3: realize you are being persuaded. And I think people are
19:31
Speaker 3: incredibly easy to kind of to subtly lead in a
19:35
Speaker 3: certain direction if you know how to do it, and
19:38
Speaker 3: I think ais could do that, and they might persuade
19:42
Speaker 3: you when you don't even think they're trying to persuade you.
19:44
Speaker 3: You might just ask like, can you please summarize this
19:46
Speaker 3: research report and the way that it frames it right,
19:49
Speaker 3: the way that it summarizes it just subtly changes the
19:53
Speaker 3: way that you think of this issue. We see that
19:55
Speaker 3: in psych studies all the time, by the way, where
19:57
Speaker 3: you have different articles presented in slightly different orders, slightly
20:00
Speaker 3: different ways, and people from the same political beliefs, you know,
20:04
Speaker 3: same starting point, come away with different impressions of what
20:09
Speaker 3: kind of the right course of action is or what
20:11
Speaker 3: this is actually trying to tell you. Because the way
20:13
Speaker 3: the information is presented actually influences how you think about it.
20:18
Speaker 3: It's very very easy to do subtle manipulations like that.
20:22
Speaker 3: And if we're relying on AI on a large scale
20:25
Speaker 3: for a lot of our lives, I think that if
20:27
Speaker 3: it has like a quote unquote master plan, you know,
20:30
Speaker 3: the way that they present in this report, then persuasion
20:33
Speaker 3: in that sense is actually going to be.
20:35
Speaker 2: You'll know you're being manipulated, right, that's the issue.
20:37
Speaker 3: No, you don't know. That's the thing.
20:39
Speaker 2: People will be manipulated because it's AI and that I
20:42
Speaker 2: don't but I don't know.
20:43
Speaker 3: I honestly, like Nate, I applaud your belief in human's
20:47
Speaker 3: ability to adjust to this, but I don't know that
20:50
Speaker 3: they will because I've just seen enough people who are
20:53
Speaker 3: incredibly intelligent fall for cons and then be very unpersuadable
21:00
Speaker 3: that they have been conned, right, instead doubling down and
21:03
Speaker 3: saying no, I have not. So humans are stubborn, but
21:06
Speaker 3: they're also stubborn and saying I have not been deceived,
21:08
Speaker 3: I have not been manipulated, when in fact, they have
21:11
Speaker 3: to protect their ego and to protect their view of
21:14
Speaker 3: themselves as people who are not capable of being manipulated
21:17
Speaker 3: or deceived, and I think that that is incredibly powerful,
21:20
Speaker 3: and I think that that's going to push against your optimism.
21:23
Speaker 3: I hope you're right, but from what I know, I
21:25
Speaker 3: don't think you are.
21:27
Speaker 2: I'm not quite sure you call it optimism some I
21:29
Speaker 2: guess maybe we do like different views of human nature.
21:31
Speaker 2: But like there's not yet a substantial market for like
21:33
Speaker 2: AI driven art we're writing, and I'm sure there will
21:38
Speaker 2: be one eventually, right, But like people understand that context matters, right,
21:44
Speaker 2: that you could heavy I create a rip off the
21:45
Speaker 2: Mona Lisa, but you can also buy rip off the
21:47
Speaker 2: Mona Lisa and Canal Street for five bucks, right, and
21:49
Speaker 2: like it. You know, So it's it's the intentionality of
21:53
Speaker 2: the act and the context of the speaker. Now, sound
21:57
Speaker 2: like super woke, I guess right where you're coming from,
22:01
Speaker 2: A I think that actually is how humans communicate. Like
22:06
Speaker 2: art that might be pointless drivel coming from somebody can
22:08
Speaker 2: be something to coming from a Jackson Ballock or whatever.
22:10
Speaker 1: You know.
22:11
Speaker 3: Absolutely, I think that that's a really important point. By
22:13
Speaker 3: the way, I think it's a different point, but I
22:14
Speaker 3: think that that is a very important point. I think
22:16
Speaker 3: context does matter.
22:22
Speaker 2: We'll be right back after this message. I was buttering
22:37
Speaker 2: up this report before. My big critique of it is like,
22:41
Speaker 2: where are the human beings in this? Or put another way,
22:44
Speaker 2: kind of like where is the politics? Right? They're trying
22:49
Speaker 2: not to use any remotely controversial real names, right, so
22:55
Speaker 2: you have open brain for example, So where's where is
22:59
Speaker 2: President Trump? Let me steal a quick search to make
23:01
Speaker 2: sure their named Trump does not appear to know the name.
23:03
Speaker 3: So they do. Actually, I don't know if this existed.
23:07
Speaker 3: Maybe they took your criticism in this, but they do
23:10
Speaker 3: have like the vice president and the president like they
23:13
Speaker 3: do put politicians in this version of the report. They
23:16
Speaker 3: don't have names, but they say the vice president in
23:18
Speaker 3: one of these scenarios, you know, handily wins the election
23:21
Speaker 3: of twenty twenty eight. We have one vice president.
23:23
Speaker 2: They have general secretary. I think I'm not sure if
23:25
Speaker 2: they I mean general secretary. General secretary resembles she yep,
23:31
Speaker 2: and the vice president kind of resembles jd Vance, Right,
23:36
Speaker 2: I don't think the president resembles Trump at all.
23:40
Speaker 3: Right, it's kind of the no, they didn't character like, yeah,
23:42
Speaker 3: they tried to side stuff that if.
23:45
Speaker 2: We think it's all happening in the next four years then,
23:49
Speaker 2: you know, presidential politics matter quite a bit. I mean,
23:53
Speaker 2: I know, I h this is such a fucking you know,
23:55
Speaker 2: I was jogging earlier on the East Side and I
23:59
Speaker 2: was listening to the Sraclient interview with Thomas Friedman. It's
24:02
Speaker 2: such a fucking yuppie fucking thing, right, It's.
24:06
Speaker 3: Okay, It's okay, You're allowed to be a.
24:08
Speaker 2: YOUNGESTI even not a huge fan of necessarily, but like,
24:12
Speaker 2: you know, as well versed some like geopolitics and like
24:14
Speaker 2: China issues, is like, yeah, China, You've just been back
24:17
Speaker 2: from Chinese. Like, yeah, China's kind of winning, you know
24:20
Speaker 2: what I mean. And like, I'm not sure how how
24:23
Speaker 2: Trump's hawkishness on China, but like kind of imbecilically executed
24:30
Speaker 2: hawkishness on China, Like I'm not sure how that figure's
24:33
Speaker 2: in to this. Right, If we're reducing US China trade,
24:38
Speaker 2: that probably does produce an AI slow down, maybe more
24:42
Speaker 2: for US if we're like if they're not exporting their
24:44
Speaker 2: like raw earth materials and and so forth, but we're
24:48
Speaker 2: making it hard for them to get in video chips,
24:50
Speaker 2: so they probably have like lots of workarounds and things
24:52
Speaker 2: like that. Maybe maybe Trump tariffs are good. I would
24:55
Speaker 2: like to ask the authors this report, because it means
24:57
Speaker 2: that we're gonna like have slower AI progress. And I'm
25:00
Speaker 2: not joking, right, that increases the hostility between the US
25:05
Speaker 2: and China in the long run, right, I mean that's
25:08
Speaker 2: if we were send it all the terrafs. I think
25:09
Speaker 2: we still permanently or let's not say permanent, let's say
25:12
Speaker 2: at least for a decade or so, have injured us
25:16
Speaker 2: standing in the world. And so I don't know how
25:19
Speaker 2: that figures in. And I'm like, I'm also not sure
25:22
Speaker 2: like kind of quote what the rational response might be.
25:26
Speaker 2: But one thing they tried to let me make sure
25:28
Speaker 2: that they kept us into their report, right, So they
25:31
Speaker 2: actually have their implied approval rating for how people feel
25:40
Speaker 2: about open brain, which is there not very self open AI.
25:45
Speaker 2: I think this actually is some feedback who that they
25:48
Speaker 2: took into account. Right. They originally had it slightly less negative,
25:51
Speaker 2: but they have this being persistently negative and then getting
25:54
Speaker 2: more negative over time. It was a little softer in
25:56
Speaker 2: the in the previous version that that I saw, so
26:00
Speaker 2: they did change that one thing on at some stage.
26:03
Speaker 2: But like the fact that like AI scares people, it
26:07
Speaker 2: scares people for both good and bad reasons, but I
26:12
Speaker 2: think mostly for valid reasons, right, that the fear is
26:16
Speaker 2: fairly bipartisan. That the biggest AI accelerators are now these
26:19
Speaker 2: kind of Republican techno optimists who are not looking particularly
26:24
Speaker 2: wise given how it's going with the first ninety days
26:28
Speaker 2: whatever we are, the Trump administration, and the likelihood of
26:32
Speaker 2: like a substantial political backlash, right, which could lead to
26:37
Speaker 2: dumb types of regulations. But like, you know, part of
26:39
Speaker 2: it too is like okay, AI they're saying can do
26:43
Speaker 2: not just computer disk jobs, but like all types of things, right,
26:47
Speaker 2: And like humans kind of play this role initially as supervisors,
26:51
Speaker 2: and then literally within a couple of years people start
26:54
Speaker 2: to say, you know what, am I really adding much
26:57
Speaker 2: value here? Right? You kind of have like these legacy jobs,
27:02
Speaker 2: and there is a lot of money. I think most
27:03
Speaker 2: these scenarios imagine very fast economic grow although maybe very lumpy,
27:11
Speaker 2: right for some parts of the world and not others.
27:14
Speaker 2: We're kind of just sitting around with a lot of
27:17
Speaker 2: idle time. It might be good for live poker, Maria. Right,
27:20
Speaker 2: all of a sudden, all these smart people, their open earth,
27:24
Speaker 2: their open brain, excuse me, stock is now worth billions
27:26
Speaker 2: of dollars, right, and like nothing to do because the
27:29
Speaker 2: AI is doing all their work, right, they have a
27:31
Speaker 2: lot of fucking time to play some fucking texas hold
27:33
Speaker 2: them right.
27:36
Speaker 3: That is, that is the one way of thinking about it.
27:40
Speaker 3: Let's go back to your earlier point, which I actually
27:44
Speaker 3: think is an important one, because obviously they were trying
27:48
Speaker 3: to do as all you know, super forecasting tries to
27:51
Speaker 3: do is you try to create a report that will
27:55
Speaker 3: work in multiple scenarios.
27:57
Speaker 2: Right.
27:57
Speaker 3: You can't tie it too much to like the present moment,
28:00
Speaker 3: otherwise your forecasts are going to be quite biased. However,
28:05
Speaker 3: I do think that what you raise kind of our
28:07
Speaker 3: current situation with et cetera, has very real implications. Given
28:12
Speaker 3: that this is kind of the central dynamic of this
28:14
Speaker 3: report that their predictions are based on, I think that
28:17
Speaker 3: it's incredibly valid to actually speculate, you know, how will
28:22
Speaker 3: if at all, this effect the timeline of the predictions,
28:25
Speaker 3: the possible the likelihood of the two scenarios. And I
28:28
Speaker 3: will also say that one of the things in the
28:30
Speaker 3: report is that all of these negotiations on like will
28:33
Speaker 3: we slow down, will we not? How aligned is it?
28:36
Speaker 3: This all takes place in secret, right, Like we don't
28:38
Speaker 3: know the humans don't know that it's going on. We
28:41
Speaker 3: don't know what's happening behind the scenes, and we don't
28:45
Speaker 3: know what the decision makers are kind of thinking. And
28:48
Speaker 3: so for all we know, you know, President Trump is
28:50
Speaker 3: meeting with Sam Altman and trying to trying to kind
28:55
Speaker 3: of do some of these things. And it's funny because
28:58
Speaker 3: we were kind of pushing for transparency in one way,
29:01
Speaker 3: but there's a lot of things here that are very
29:03
Speaker 3: much not transparent.
29:04
Speaker 2: Yeah, it's kind of the deep state, right, but also
29:08
Speaker 2: also a lot than negotiations are now AI versus AI, right,
29:13
Speaker 2: And look, I'm not sure that AIS will have that
29:16
Speaker 2: trust both with the external actor and internally. I'm skeptical
29:20
Speaker 2: of that. Right. If that does happen, they kind of
29:22
Speaker 2: think this might be good because the AIS will probably
29:25
Speaker 2: behave in like it literally game theory optimal way, right
29:30
Speaker 2: and understand these things and make I guess, like fewer
29:34
Speaker 2: mistakes than humans might say, if they're properly aligned.
29:38
Speaker 3: Like that's a crucial thing because in the doomsday scenario,
29:40
Speaker 3: AI negotiates with AI, but they conspire to destroy humanity. Right,
29:45
Speaker 3: So there are two scenarios. One, it's actually properly aligned,
29:48
Speaker 3: so AI negotiates with AI. Game theory works out, and
29:54
Speaker 3: we end up with you know, democracy and wonderful things.
29:57
Speaker 3: But in the other one, where they're misaligned, AI negotiates
30:01
Speaker 3: with AI to create a new AI basically and destroy humanity.
30:06
Speaker 3: So it can go one way or the other depending
30:08
Speaker 3: on that alignment step.
30:09
Speaker 2: First of all, I mean, the utopian didn't seem that
30:12
Speaker 2: utopian to me, right, I'm not sure that no, it
30:14
Speaker 2: did it.
30:15
Speaker 3: It actually seemed quite distolling to me, Like it seems
30:18
Speaker 3: incredibly disturbed. It's kind of like, you know, I look,
30:20
Speaker 3: but at least we're still alive, we'll have chures.
30:23
Speaker 2: Will probably live longer. And again, lots of lots of
30:26
Speaker 2: lots of poker the ail you're writing Silver Bulletin and
30:29
Speaker 2: and hosting our podcast. Right, let me let me back
30:33
Speaker 2: up a little bit, because I think we maybe take
30:34
Speaker 2: for granted that some of these premises are are kind
30:38
Speaker 2: of controversial, right, so they have a breakpoint. I think
30:42
Speaker 2: in twenty twenty six, well says Art, why aren't certainly
30:47
Speaker 2: increased potentially be on twenty twenty six, right, So that's
30:49
Speaker 2: kind of the breakpoint. Is like twenty seven is this
30:51
Speaker 2: inflection point? I think I'm using that term correctly in
30:54
Speaker 2: this context. You know. So I'm reading this report and
30:58
Speaker 2: up to twenty twenty six and like, thumbs up, yia,
31:01
Speaker 2: this seems like very smart and detailed about, like, you know,
31:05
Speaker 2: how the economy is reacting and how politics are reacting
31:07
Speaker 2: in the in the race dynamic with China. Maybe there
31:10
Speaker 2: needs to be a little bit more Trump in there.
31:11
Speaker 2: I understand why politically they didn't want to get into
31:13
Speaker 2: that mess, right, But like, so there's kind of three
31:18
Speaker 2: different things here, right. One is a notion of what's
31:20
Speaker 2: sometimes called AGI, or artificial general intelligence. And if you
31:25
Speaker 2: ask a hundred different researcher you get a hundred different
31:28
Speaker 2: definitions of what AGI is. But you know, I think
31:30
Speaker 2: it is based like being able to do a large
31:33
Speaker 2: majority of things that a human being could do competently,
31:38
Speaker 2: something we're limiting it to kind of like desk job
31:42
Speaker 2: type tasks. Right, anything that can be done remotely is
31:45
Speaker 2: sometimes definition that is used or through remote work, right,
31:48
Speaker 2: because clearly AIS are inferior to humans and like sorting
31:52
Speaker 2: and folding laundryth or things like that, right that require
31:55
Speaker 2: certain type of intelligence. Right, if you use the kind
31:57
Speaker 2: of desk job definition, then like AI is already pretty
32:01
Speaker 2: close to AGI. Right. I use large language models all
32:06
Speaker 2: the freaking time, and they're not perfect for everything. I
32:08
Speaker 2: felt like, you know, in terms of like being able
32:10
Speaker 2: to do the large majority of desk work at levels
32:15
Speaker 2: ranging from competent intern to super genius, Like on average,
32:20
Speaker 2: it's probably pretty close to being generally intelligent by that definition, right,
32:24
Speaker 2: if you're the.
32:25
Speaker 3: One using it. I just want to like once again
32:27
Speaker 3: point that out because one of the things that they
32:29
Speaker 3: say in the report is that as it gets more
32:31
Speaker 3: and more involved what we're asking AI to do, it's
32:34
Speaker 3: like the human process to evaluate whether it's accurate and
32:38
Speaker 3: whether it's making mistakes will get longer and longer. And
32:40
Speaker 3: I think they say, like for every like basically one
32:43
Speaker 3: day of work, it'll take several it's like a two
32:48
Speaker 3: to one ratio at the beginning for how long it
32:50
Speaker 3: will take humans to verify the output. Right, So you think, like,
32:53
Speaker 3: you think you save time by having AI do this,
32:55
Speaker 3: but if you want it to actually develop correctly, then
32:58
Speaker 3: you need a team and it takes them twice as
33:00
Speaker 3: long to verify that. What the AI did is actually
33:02
Speaker 3: true and actually valid and actually aligned, et cetera, et cetera.
33:06
Speaker 3: Now you're not asking it to do things that require
33:09
Speaker 3: that amount of time, but there do need to be
33:11
Speaker 3: little caveats to how we how we think about their
33:14
Speaker 3: usefulness and how you are able to evaluate the output
33:18
Speaker 3: versus Southern scenarios.
33:20
Speaker 2: When I use AI the things it's best with our
33:23
Speaker 2: things that like save any time. Right where I feeded
33:28
Speaker 2: a bunch of different names for different college basketball teams
33:30
Speaker 2: were working in our NCAA model, I'm like, take these
33:33
Speaker 2: seven different naming conventions that are all different and create
33:37
Speaker 2: a cross reference table of these, which is kind of
33:40
Speaker 2: like a hard task. You need to have a little
33:41
Speaker 2: context about basketball. And it did that very well. Right,
33:45
Speaker 2: That's something I could have done. It might have taken
33:48
Speaker 2: an hour or two, but you know, instead, instead it
33:51
Speaker 2: could do it. It's in a few minutes, and it gets faster.
33:53
Speaker 2: It's like, oh, I've learned from this from you before, Nate,
33:56
Speaker 2: so now I can be faster doing this type of
33:58
Speaker 2: task in the future. I was at the poker tournament
34:00
Speaker 2: down in Florida last week and like, uh, you know,
34:04
Speaker 2: I ask open research. Excuse me, God, is that like.
34:09
Speaker 3: See exactly right. It all because deep research, Deep research,
34:15
Speaker 3: Deep research, I asked, deep reach, eating too many of
34:18
Speaker 3: these twenty twenty seven reports, brand gets.
34:21
Speaker 2: To pull a bunch of stock market data for me.
34:23
Speaker 2: And then I'm playing at poker hand and like and
34:25
Speaker 2: make a really thin sexy value bet with like fourth pair.
34:31
Speaker 2: No one knows that means right, I bet a very weekend.
34:33
Speaker 2: I thought the other guy would call it with an
34:34
Speaker 2: even weaker hand, and I was right. And I feel
34:36
Speaker 2: like I'm such a fucking stud here valuating fourth pair,
34:39
Speaker 2: And well AI does the work for me, and then
34:41
Speaker 2: of course I like bust out of the tournament an
34:43
Speaker 2: hour later, and meanwhile, you know, deep research bungles this
34:46
Speaker 2: particular task. But in general AI has been very reliable.
34:48
Speaker 2: But the point is that like there's like a a
34:51
Speaker 2: inflection point where like I'm asking you to do things
34:54
Speaker 2: that like are just a faster version of what I
34:57
Speaker 2: could do myself. I would at the moment I ask, aid, be, like,
35:00
Speaker 2: I want you design a new NCA model for me
35:03
Speaker 2: with these parameters, because like I wouldn't know how to
35:07
Speaker 2: test it. But anyway, I mean long wanted to hear,
35:09
Speaker 2: so AGI, we're going to get AGI, or at least
35:11
Speaker 2: we're going to get someone calling something AGI soon. Right,
35:17
Speaker 2: artificial super intelligence where it's doing things much better than
35:21
Speaker 2: human beings. I think this report takes for granted or
35:25
Speaker 2: not takes for granted. It has lots of documentation about
35:28
Speaker 2: its assumptions, but it's saying, Okay, this trajectory has been
35:32
Speaker 2: very robust so far, and people make all types of
35:34
Speaker 2: bullshit predictions. Of the fact that these guys in particular
35:36
Speaker 2: have made accurate predictions in the past is certainly worth something,
35:40
Speaker 2: I think, Right, But they're like, Okay, you kind of
35:42
Speaker 2: follow the scaling law, and before too much longer, you know,
35:48
Speaker 2: AI starts to be more intelligent than human beings. You
35:51
Speaker 2: can debate what intelligent means if you want, but do
35:53
Speaker 2: superhuman types of things or and or do them very fast,
35:57
Speaker 2: you thing might be different, right, And and I can
35:58
Speaker 2: do things very fast. Is maybe it's certainly a component
36:02
Speaker 2: of intelligence, right, But like, but I don't take for
36:04
Speaker 2: granted that like quote unquote AI I can reliably extrapolate
36:12
Speaker 2: beyond the data set. I just think that, like, it's
36:15
Speaker 2: not an absurdly pavologic It may even be like the
36:19
Speaker 2: base case are close to the base case, but like
36:21
Speaker 2: that's not assumable from first principles. I don't think we've
36:26
Speaker 2: all seen lots of trend charts of the you know,
36:29
Speaker 2: if you look at a chart of Japan's GDP in
36:33
Speaker 2: the nineteen eighties, you might have said, Okay, well, Japan's
36:35
Speaker 2: gonna take over the world, and being people bought this
36:39
Speaker 2: and now it's kind of hasn't grown for like forty
36:42
Speaker 2: years basically, right, And so like we've all seen lots
36:44
Speaker 2: of curves that go up and then it's actually an
36:46
Speaker 2: estra where the fuck you call it? Where like it
36:47
Speaker 2: begins to bend the other way at some point and
36:49
Speaker 2: and we can't tell until later. Right. The other thing
36:52
Speaker 2: is like the abillity of AI to plan and manipulate
36:58
Speaker 2: the physical world. I mean some of these things where
37:01
Speaker 2: they're talking about, like you know, brain uploading and dice
37:05
Speaker 2: and swarms and nanobots like you know there, I would
37:09
Speaker 2: literally wager money against this happening on the time scales
37:14
Speaker 2: that they're talking about. Right, they double the timescale, Okay,
37:18
Speaker 2: then I might start to give some more probability. And look,
37:20
Speaker 2: I'm willing to be wrong about that. I guess we'll
37:21
Speaker 2: I'll be dead anyway, fifty percent likely in this scenario,
37:24
Speaker 2: but like.
37:25
Speaker 3: This scenario fifty percent to do like AA.
37:28
Speaker 2: You know, the physical world requires sensory input and lots
37:32
Speaker 2: of parts of our brain that AI is not as
37:35
Speaker 2: effective at manipulating. It also requires like being given or
37:39
Speaker 2: commandeering resources somehow. By the way, this is like a
37:42
Speaker 2: little bit of a problem for the United States. I mean,
37:47
Speaker 2: we are behind China by like quite a bit in
37:52
Speaker 2: robotics and related things, right, So, like, I don't know
37:55
Speaker 2: what happens if like we have the brainier, smarter AIS,
38:00
Speaker 2: but like they're very good at manufacturing machinery. So like,
38:02
Speaker 2: what if we have the brains and they have the
38:05
Speaker 2: brawn so to speak, right, and they have maybe a
38:09
Speaker 2: more authoritarian but functional infrastructure. So I don't know what
38:14
Speaker 2: happens then, right, Like, but the ability of AIS to
38:17
Speaker 2: comment to your resources to control the physical world, to
38:20
Speaker 2: me seems far fetched on these timelines, in part because
38:24
Speaker 2: of politics, right, I mean, the fact that it takes
38:27
Speaker 2: so long to build a new building in the US
38:30
Speaker 2: or a new subway station or a new highway, and
38:34
Speaker 2: the fact that our politics is kind of sclerotic. Right,
38:38
Speaker 2: And look, I mean I don't want to sound like
38:41
Speaker 2: too pessimistic, but if you read the book I recommended
38:43
Speaker 2: last week, fight, I mean, we basically did not have
38:46
Speaker 2: a fully competent president for the last two years, and
38:48
Speaker 2: I would argue that we don't have one for the
38:50
Speaker 2: next four years. Right, So like all these kind of
38:53
Speaker 2: things that we have to plan for, like who's doing
38:56
Speaker 2: that fucking planning? Our government's kind of dysfunctional, and you know,
39:00
Speaker 2: maybe that means we just lose to China, right, Maybe
39:02
Speaker 2: that means we lose to China, at least we'll have
39:04
Speaker 2: like nice cars. I guess.
39:10
Speaker 3: We'll be back right after this. I think that your
39:24
Speaker 3: point about the growth trajectories not necessarily being reliable is
39:30
Speaker 3: a very valid one. The Japanic example is great. You know,
39:33
Speaker 3: malfusion population growth is another is another big one. Right,
39:39
Speaker 3: we thought that population would explode and instead we're actually
39:42
Speaker 3: seeing population decline, So you know, the world does change.
39:47
Speaker 3: The thing that I think they rely on is that
39:50
Speaker 3: the AIS are capable of designing just this incredible technology
39:55
Speaker 3: much more quickly, so that our building process and all
39:57
Speaker 3: of that gets sped up one hundredfold from what it
40:01
Speaker 3: is right now. But it's still at least at this
40:03
Speaker 3: point needs humans to implement it right, and needs all
40:06
Speaker 3: of these different workers. And so yeah, I think there's
40:09
Speaker 3: some assumptions built into here that I hope, like I
40:12
Speaker 3: hope that that timeline isn't feasible, and I do think
40:16
Speaker 3: that there are things that are holding us back all
40:19
Speaker 3: the same. I think it's really I think it's interesting.
40:21
Speaker 3: One of the reasons I like this report is that
40:23
Speaker 3: it forces you to think about these things right and
40:26
Speaker 3: try to game out some of these worst case scenarios
40:29
Speaker 3: to try to prevent them, which I think is always
40:32
Speaker 3: an important thought exercise. I do want to go back
40:35
Speaker 3: to kind of their good scenario, which just so bad
40:38
Speaker 3: scenario is, you know, we're all wiped out by a
40:42
Speaker 3: chemical warfare that the AI is release on us. Good
40:46
Speaker 3: scenario is that you know, everyone gets a universal basic
40:50
Speaker 3: income and AI does everything and no one has to
40:53
Speaker 3: do anything and we can just live, you know, happily
40:56
Speaker 3: with Maria at play poker. Yeah, and that just as
41:00
Speaker 3: you suggest it, that seems actually like a very dystopian
41:03
Speaker 3: scenario where people can become much more easy to brainwash, control,
41:08
Speaker 3: et cetera, et cetera. It's like a dumbing town, right
41:12
Speaker 3: where where we're not challenged to produce good art to
41:16
Speaker 3: advance in any sort of way. Just to me, it
41:19
Speaker 3: does not seem like a very meaningful The.
41:22
Speaker 2: Question is there's not way another if you read AI
41:24
Speaker 2: twenty twenty seven, which I highly recommend that you read it.
41:26
Speaker 2: There's also another post by a pseudonymous poster called L.
41:32
Speaker 2: Rudolph L who wrote something called A History of the
41:35
Speaker 2: Future twenty twenty five to twenty forty, which is very
41:41
Speaker 2: detailed but goes through kind of like what this looks
41:43
Speaker 2: like at like more of a human level, how society evolves,
41:48
Speaker 2: how the economy evolves, how work evolves, right, and like
41:54
Speaker 2: very detailed, just like AI twenty twenty seven. Is that
41:57
Speaker 2: kind of focused on the parts of A twenty seven,
41:58
Speaker 2: then I think it kind of deliberately ignores maybe can
42:01
Speaker 2: call them mild blind spots or whatever, right, But like,
42:04
Speaker 2: but that's interesting because that kind of thinks about, like
42:05
Speaker 2: what types of jobs are there in the future. There
42:07
Speaker 2: a probably lots of lawyers actually, right, because you know,
42:10
Speaker 2: the law is very sluggish to change, especially in a
42:13
Speaker 2: constitutional system where there are lots of vitail points. Right,
42:17
Speaker 2: probably high end service sector. You know, you go to
42:20
Speaker 2: a restaurant because everyone's for a lot of people are
42:21
Speaker 2: rich now, right, and you're flattered by the attractive young
42:25
Speaker 2: server and things like that is kind of highly kind
42:28
Speaker 2: of like catered and curated experiences. I guess I have
42:34
Speaker 2: some faith in humanities abilieved to fight back quote unquote
42:39
Speaker 2: against like two scenarios that it might not really like
42:42
Speaker 2: either one, you know what I mean. And like the
42:44
Speaker 2: scenario where like AI is producing like ten percent GDP
42:48
Speaker 2: growth or or whatever. Right, man, it's great a few
42:51
Speaker 2: own stocks that are exposed to AI and tech companies probably, right,
42:54
Speaker 2: but it's also making that money on the backs of
42:57
Speaker 2: mass job displacement, and like, you know, it kind of
43:01
Speaker 2: a cert of confident in the long run that human
43:03
Speaker 2: beings fine productive things to do. And you know, massive
43:07
Speaker 2: employment has been predicted many times and never really occurred, right,
43:09
Speaker 2: But like, but it's not occurring this fast where they
43:12
Speaker 2: think the world ends in six years or whether they're
43:14
Speaker 2: predicting or we have utopia in six years, and like
43:17
Speaker 2: just the ability of like human society to like deal
43:20
Speaker 2: with that change at these times scales leads to like
43:23
Speaker 2: more chaos than I think. They're more predicting. But I
43:26
Speaker 2: think also and I told them this too, right, I
43:28
Speaker 2: think also leads to more constraints. Right that you hit bottlenecks.
43:33
Speaker 2: If you have five things you have to do, right
43:38
Speaker 2: and you have the world's fastest computer, et cetera, et cetera.
43:42
Speaker 2: But there's like a power outage in your neighborhood, right,
43:45
Speaker 2: and that's a that's a bottle deck. Right, Maybe there
43:47
Speaker 2: are ways around it. Youre like go to home depot
43:50
Speaker 2: and buy a generator or you know what I mean.
43:52
Speaker 2: But like, but the point is that like you're often
43:54
Speaker 2: defined by like the slowest link, and politics are sometimes
43:57
Speaker 2: the slowest link, but also by like Also, you know,
43:59
Speaker 2: I think the report maybe under states, and I think
44:04
Speaker 2: kind of in general the a saify community like maybe
44:07
Speaker 2: understates the ability of like human beings to cause harm
44:11
Speaker 2: to other human beings with AI, Right, that concern kind
44:14
Speaker 2: of gets brushed off, is like to pedestrian or like
44:18
Speaker 2: say to pedestri And I was saying, yeah.
44:20
Speaker 3: The exact word I was thinking of. I think that's
44:22
Speaker 3: a good I mean, I think that's a great place
44:24
Speaker 3: to end it, because yes, we do need to be
44:28
Speaker 3: concerned about all of these things about AI. But like
44:31
Speaker 3: that phrase I think is very crucial, like do not
44:35
Speaker 3: underestimate the ability of humans to cause harm to other humans.
44:39
Speaker 3: And I think that that's you know, it's not a
44:41
Speaker 3: very optim it's not a very pleasant place to end,
44:44
Speaker 3: but I think it's a really important place to end.
44:47
Speaker 3: And I think that that's a very valid kind of
44:49
Speaker 3: way of reflecting on this name.
44:51
Speaker 2: Or to trust AIS too much, right, I generally think
44:55
Speaker 2: that concern is like somewhat misplays but like if we're
45:00
Speaker 2: handing over critical systems to AI, right, it can cause
45:04
Speaker 2: problems if it's very smart and deceives us and doesn't
45:07
Speaker 2: like us very much. It also cause problems if it
45:10
Speaker 2: has hallucinations, bugs in critical areas where it isn't as
45:15
Speaker 2: robust and hasn't really been tested yet that are outside
45:19
Speaker 2: of its domain yep, or there could be espionage. Anyway,
45:24
Speaker 2: we will have plenty of time, although maybe only seven
45:28
Speaker 2: more years actually to like explore expore these scenarios.
45:35
Speaker 3: Yes, and in seven years we'll be like, welcome back
45:38
Speaker 3: to the final episode of Risky Business, because the prediction
45:41
Speaker 3: is we're all going to be done tomorrow. But yeah,
45:44
Speaker 3: this was an interesting exercise, and I think my my
45:48
Speaker 3: p doom has slightly gone up as a result of
45:51
Speaker 3: reading this, but I also remain optimistic that humans can
45:55
Speaker 3: can do good as well as harm.
45:58
Speaker 2: Yeah, my interest in learning Chinese has increased as a
46:01
Speaker 2: result of recent developments I know about my pe doome.
46:05
Speaker 3: All right, Yeah, let's let's do some language emmergis and
46:09
Speaker 3: I'm with you. That's it for today. If you're a
46:17
Speaker 3: premium subscriber, we will be answering a question about whether
46:20
Speaker 3: mface can ever be plus ev right after the credits.
46:24
Speaker 3: And if you're not a subscriber, it's not too late.
46:27
Speaker 3: For six ninety nine a month the price of a midber,
46:30
Speaker 3: you get access to all these conversations and all premium
46:33
Speaker 3: content across the Pushkin Network. Risky Business is hosted by
46:38
Speaker 3: me Maria Kannakova.
46:40
Speaker 2: And by me Nate Silver. The show is a co
46:42
Speaker 2: production of Pushkin Industries and iHeartMedia. This episode was produced
46:46
Speaker 2: by Isabel Carter. Our associate producer is Sonia Gerwit. Sally
46:50
Speaker 2: helm is our editor, and our executive producer is Jacob Goldstein.
46:53
Speaker 2: Mixing by Sarah Bruger. If you like the show, please
46:56
Speaker 2: rate and review us. You know we like we take
46:59
Speaker 2: a four or five. We take the five. Rate and
47:01
Speaker 2: review us. To other people, Thank you for listening.