WEBVTT - Peter Thiel: “We Need to Hit the Accelerator as Hard as Possible” on AI 0:00:09.930 --> 0:00:14.270 The Builders is a production of Kaleidoscope and iHeart, presented 0:00:14.430 --> 0:00:15.170 by RAMP. 0:00:17.350 --> 0:00:18.410 I'm Walter Isaacson. 0:00:18.590 --> 0:00:22.830 Welcome to The Builders, my new series of interviews featuring 0:00:22.870 --> 0:00:29.350 the most consequential minds shaping our future. In doing this series, 0:00:29.790 --> 0:00:33.519 I wanted to understand the values and the moral views 0:00:33.659 --> 0:00:37.349 of The Builders, and I came to realize... Better and 0:00:37.430 --> 0:00:42.750 for worse, they were often quite unusual. Some were strongly libertarian. 0:00:42.950 --> 0:00:46.629 Others were Nietzschean or dismissive of empathy. Some also had 0:00:46.670 --> 0:00:49.890 a very dark view of humanity. I felt maybe they 0:00:49.930 --> 0:00:52.250 should touch grass every now and then, get out of 0:00:52.270 --> 0:00:55.350 their billionaire bubbles and understand what the rest of us 0:00:55.770 --> 0:00:56.230 feel like. 0:00:57.380 --> 0:00:59.210 We may want these people. 0:00:58.950 --> 0:01:03.880 To build, but especially with powerful technology, we need to 0:01:03.960 --> 0:01:07.640 make sure that we all have a say And that 0:01:07.760 --> 0:01:11.930 begins with understanding what they are doing and why. 0:01:12.890 --> 0:01:15.670 And so today I'm speaking with Peter Thiel. 0:01:15.910 --> 0:01:18.030 Billionaire Silicon Valley investor Peter Thiel. 0:01:18.190 --> 0:01:21.190 Peter Thiel is a high profile figure in Silicon Valley. 0:01:21.209 --> 0:01:23.350 Peter Thiel is one of the most respected names in 0:01:23.370 --> 0:01:26.670 the world of venture capital. In addition to his co-founding 0:01:26.709 --> 0:01:29.090 of PayPal, he was also an early investor in companies 0:01:29.440 --> 0:01:31.140 such as Facebook and LinkedIn. 0:01:31.740 --> 0:01:36.650 Peter Thiel occupies a different role. He's a builder of builders. 0:01:37.030 --> 0:01:40.720 He makes bets on people. Elon Musk at PayPal, Alex 0:01:40.800 --> 0:01:44.740 Karp at Palantir, Sam Altman and the founders of OpenAI. 0:01:45.040 --> 0:01:47.400 And he backed politicians such as J.D. 0:01:47.460 --> 0:01:47.919 Vance. 0:01:48.220 --> 0:01:52.240 There's two reasons that Senator Vance could be a heartbeat 0:01:52.260 --> 0:01:54.500 from presidency and basically the president. 0:01:54.540 --> 0:01:56.600 The first is Peter, the second is Thiel. 0:01:57.040 --> 0:01:59.500 To some, he's a visionary investor. 0:01:59.840 --> 0:02:01.500 I think Peter is one of the most brilliant people 0:02:01.520 --> 0:02:03.620 I've ever met. He thinks of things no one else does. 0:02:04.040 --> 0:02:06.370 To others... He's a shadowy figure. 0:02:06.870 --> 0:02:09.590 Peter Thiel is quiet and deadly. I think Thiel is 0:02:09.930 --> 0:02:10.550 pulling all. 0:02:10.490 --> 0:02:11.090 The strings here. 0:02:11.290 --> 0:02:12.929 I think he seems like a dark lord to a 0:02:12.970 --> 0:02:13.400 lot of us. 0:02:13.460 --> 0:02:15.519 And it's like, does he think he has an endgame 0:02:15.540 --> 0:02:16.740 that is like happy? 0:02:16.940 --> 0:02:18.040 It's so strange to have this. 0:02:18.340 --> 0:02:23.440 You build autonomous drone technology. You build domestic surveillance technology. 0:02:24.120 --> 0:02:27.930 And then your other pet passion is the Antichrist. 0:02:29.500 --> 0:02:32.680 So I wanted to ask the builder of builders, what 0:02:32.720 --> 0:02:36.720 does he believe we are building toward? Are the fears 0:02:36.930 --> 0:02:40.170 that are sweeping the headlines grounded in reality? 0:02:41.050 --> 0:02:41.690 And what does he. 0:02:41.710 --> 0:02:45.450 Mean when he goes around the world giving lectures about 0:02:46.050 --> 0:02:52.030 the Antichrist? But perhaps most importantly, have we concentrated way 0:02:52.070 --> 0:02:55.489 too much power in the hands of people who have 0:02:55.550 --> 0:02:59.330 the demons in drive that allow them to be audacious builders, 0:03:00.090 --> 0:03:03.890 but may not have the empathy and stable moral compass 0:03:04.330 --> 0:03:09.900 to be good stewards of this technology. So here's my 0:03:09.940 --> 0:03:19.540 conversation with Peter Thiel. Peter Thiel, thank you for joining 0:03:19.580 --> 0:03:20.079 The Builders. 0:03:20.680 --> 0:03:21.160 Thanks for having me. 0:03:21.639 --> 0:03:26.090 We've seen this enormous backlash in the past few months 0:03:26.750 --> 0:03:29.530 against AI. What is causing that? 0:03:30.850 --> 0:03:33.739 Yeah, I agree with the what that it's happening. You know, 0:03:33.860 --> 0:03:38.620 why questions are always much harder, much more speculative in nature. 0:03:38.840 --> 0:03:42.210 I'm tempted to say it's somewhat overdetermined. It's very unclear 0:03:42.250 --> 0:03:43.670 who the winners and losers from. 0:03:43.590 --> 0:03:44.190 AI will be. 0:03:44.250 --> 0:03:47.690 So I think it's going to add a lot to growth, 0:03:47.730 --> 0:03:50.290 but it will do so in a fairly disruptive kind 0:03:50.330 --> 0:03:53.410 of a way. There are, of course, these dystopian science 0:03:53.470 --> 0:03:58.280 fiction scenarios that, you know, in some ways we believe, 0:03:58.340 --> 0:04:01.660 and maybe they're not entirely unreasonable. So there's some small tale. 0:04:01.680 --> 0:04:03.300 Dystopian mean that we'll all be white. 0:04:03.320 --> 0:04:05.160 The Terminator movie, the Matrix. 0:04:05.180 --> 0:04:08.140 But we see that Dario and Elon and Sam are 0:04:08.160 --> 0:04:09.160 now talking about that. 0:04:09.200 --> 0:04:13.610 Yeah, to some extent they've leaned into it or partially 0:04:13.650 --> 0:04:16.109 believed it or too many of their employees believed it. 0:04:16.690 --> 0:04:19.560 And so there was, yeah, there's some prehistory where This 0:04:19.600 --> 0:04:21.520 was in the air for a long time. 0:04:22.080 --> 0:04:24.540 I think Elon believes it for a decade. 0:04:24.700 --> 0:04:28.969 Well, you know, it was in some ways when Demis 0:04:29.029 --> 0:04:32.650 has created DeepMind and it gets sold to Google, Elon 0:04:32.690 --> 0:04:36.090 and Sam create OpenAI the next year or so after that. 0:04:36.650 --> 0:04:41.010 And it's basically, in some ways, it's an anti-Google company, right? 0:04:41.210 --> 0:04:44.770 This technology is too important, too dangerous to be just 0:04:45.420 --> 0:04:48.089 in a single company. And then, of course, there are 0:04:48.130 --> 0:04:51.030 people who take it even more seriously than Altman. I 0:04:51.050 --> 0:04:54.530 think Elon exits around 2018. And then by 2020, you 0:04:54.589 --> 0:04:59.610 have thereabouts is the seven people who start Anthropic exit 0:05:00.210 --> 0:05:03.250 from OpenAI. In some sense, they want to take these 0:05:03.430 --> 0:05:07.410 issues even more seriously than before. You know, effective altruism, 0:05:07.730 --> 0:05:09.390 this idea had been in the air for a while. 0:05:09.529 --> 0:05:12.510 Wait, explain that because people are now saying it's sort 0:05:12.540 --> 0:05:15.900 of a psyops or David Sack says. conspiracy theory of 0:05:15.980 --> 0:05:19.839 effective altruism. And I think even people like me vaguely 0:05:19.900 --> 0:05:21.539 know what Sam Bankman Free was on about. 0:05:21.560 --> 0:05:22.560 Well, you don't get that. 0:05:22.640 --> 0:05:23.320 Tell me about that. 0:05:23.980 --> 0:05:27.580 Well, there's a lot of history, but even if you 0:05:27.760 --> 0:05:30.339 go back to the 2000s in the East Bay of 0:05:30.440 --> 0:05:33.940 Silicon Valley, Berkeley, Oakland, some of those places, it wasn't 0:05:33.960 --> 0:05:36.400 where the big tech companies were being built, but there 0:05:36.420 --> 0:05:39.860 was sort of a creativity with ideas. It was a 0:05:39.880 --> 0:05:44.050 little bit adjacent to the new atheists, you know, how 0:05:44.089 --> 0:05:47.070 could you be more reasonable, rational? Where was the world heading? 0:05:47.650 --> 0:05:51.650 And the one big idea that these sort of futurists 0:05:52.589 --> 0:05:56.770 kept coming back to was that perhaps something like AI 0:05:56.860 --> 0:05:59.980 could happen, that perhaps it was closer than people thought. 0:06:00.520 --> 0:06:02.919 There had been an AI desert for a long time. 0:06:03.640 --> 0:06:06.640 It had been over-promised for many decades. But on the 0:06:06.660 --> 0:06:10.020 other hand, Moore's law was still progressing. The computers were 0:06:10.510 --> 0:06:13.480 relentlessly getting more powerful. And perhaps it was not a 0:06:13.540 --> 0:06:17.920 question of necessarily a much better algorithm. Perhaps it was 0:06:18.040 --> 0:06:21.380 just almost a hardware problem that once you got the 0:06:21.420 --> 0:06:24.820 chips and the computers that were powerful enough, it would 0:06:24.940 --> 0:06:28.890 be more likely that something like this would happen. And 0:06:28.930 --> 0:06:31.050 then that this would be a very, very big thing 0:06:31.089 --> 0:06:34.489 when it happened, that this would be an enormously socially 0:06:34.550 --> 0:06:38.529 transformative thing. It was potentially very dangerous. It could be 0:06:38.550 --> 0:06:41.039 maybe it could be weaponized. Maybe there were these science 0:06:41.080 --> 0:06:44.260 fiction scenarios where the AI would almost become autonomous. So 0:06:44.279 --> 0:06:47.460 there were very many different kinds of scenarios. And then 0:06:47.480 --> 0:06:49.500 there were, of course, parts of it that were, I 0:06:49.540 --> 0:06:54.320 don't know, crazy and almost cult-like. But it had maybe 0:06:54.350 --> 0:06:56.870 the good dimension of a cult where there was one 0:06:57.190 --> 0:07:00.270 big thing they knew that the other people did not. 0:07:00.290 --> 0:07:02.690 And I think even though there are ways I'd like 0:07:02.710 --> 0:07:04.910 to be very critical of them today, there were ways 0:07:04.930 --> 0:07:07.690 they got that one big thing right. And then, you know, 0:07:07.930 --> 0:07:11.250 has a cult ever started a company as successful as Anthropic. 0:07:11.630 --> 0:07:14.430 Even a billion-dollar company seems like bigger than a crazed 0:07:14.490 --> 0:07:15.990 cult could build. So if you have a cult that's 0:07:16.010 --> 0:07:19.610 built a company that's worth probably close to $ 2 trillion 0:07:19.630 --> 0:07:22.590 at the IPO, it would probably be the, I don't know, sixth, 0:07:22.630 --> 0:07:26.120 seventh largest company by market cap in the world, built 0:07:26.160 --> 0:07:29.560 in like six, seven years. Has a cult ever done 0:07:29.620 --> 0:07:33.620 something like that? So yes, I want to be critical 0:07:33.660 --> 0:07:36.940 of them, and they're things I'm uncomfortable and critical about. 0:07:37.460 --> 0:07:39.870 But the first thing I would try to do is always, 0:07:39.890 --> 0:07:41.510 you always want to steel man your opponents. You want 0:07:41.530 --> 0:07:44.210 to try to understand them as well as they understand 0:07:44.230 --> 0:07:47.190 themselves or even make their arguments better. And the way 0:07:47.210 --> 0:07:49.560 I would steel man the effect of altruists is, first off, 0:07:49.600 --> 0:07:51.440 they got one enormously big thing right. 0:07:52.060 --> 0:07:54.600 And that one enormously big thing was? 0:07:54.900 --> 0:07:56.000 AI is around the corner. 0:07:56.260 --> 0:08:00.440 But also that it could become misaligned with human values? 0:08:02.940 --> 0:08:05.730 The way AI was classically defined, let's say. There were 0:08:05.750 --> 0:08:08.730 all these different definitions of AI, of course. It could 0:08:08.770 --> 0:08:12.210 be just sort of machine learning and sort of like 0:08:12.370 --> 0:08:14.890 maybe cameras everywhere or something like that. It's a sort 0:08:14.910 --> 0:08:17.550 of totalitarian AI that you have in China. Or it 0:08:17.630 --> 0:08:21.050 could be this sort of super-duper AI with a godlike 0:08:21.110 --> 0:08:23.750 intelligence or something, which was sort of the Nick Bostrom 0:08:24.370 --> 0:08:28.050 kind of fantasy. But the in-between version of AI was 0:08:28.530 --> 0:08:32.010 not that it would be this dumb surveillance cameras, not 0:08:32.030 --> 0:08:34.620 that it would be godlike. but they would be like 0:08:34.660 --> 0:08:37.450 a human being. That had been the AI test. The 0:08:37.470 --> 0:08:40.910 Turing test from the beginning was, it's not that it 0:08:41.309 --> 0:08:44.170 is a human being, but that it can fool you 0:08:44.290 --> 0:08:47.820 into thinking it's a person. And from the outputs, you 0:08:47.860 --> 0:08:51.819 can't tell the difference. Yeah, that's going to be a very, 0:08:52.320 --> 0:08:54.740 very big deal. And it raises all sorts of, you know, 0:08:55.000 --> 0:08:59.059 raises even basic economics questions. Is it an economic compliment 0:08:59.080 --> 0:09:01.140 to people? Is it a substitute for people? 0:09:01.429 --> 0:09:04.610 Well, let's do both. There's two big things it raises. 0:09:05.429 --> 0:09:09.280 One is what will it do to human labor? Will 0:09:09.340 --> 0:09:12.660 it create more jobs and more productivity? Or will it 0:09:13.020 --> 0:09:14.939 pretty much put us out of work and we'll have 0:09:15.000 --> 0:09:19.150 to figure out what we do? That's part of the backlash, 0:09:19.190 --> 0:09:22.970 especially my students at Tulane are like, wait a minute. 0:09:23.410 --> 0:09:26.969 So tell me your feeling about that backlash. 0:09:27.730 --> 0:09:29.949 Okay, so let me start by answering what I think 0:09:29.970 --> 0:09:32.910 will happen with a qualifier that nobody really knows and 0:09:33.110 --> 0:09:36.630 nobody can be certain. I think in aggregate, on a 0:09:36.690 --> 0:09:41.270 macroeconomic level, AI will increase productivity a lot. It will 0:09:41.290 --> 0:09:45.670 be a fairly disinflationary force in our economy. And then 0:09:46.150 --> 0:09:50.980 unless it's incredibly narrowly captured by a few companies. You'd 0:09:51.000 --> 0:09:53.940 expect the median person to be better off or things 0:09:54.000 --> 0:09:56.450 like that. This was, you know, the Luddites in the 0:09:56.630 --> 0:09:59.050 19th century were always worried that the machines would replace 0:09:59.530 --> 0:10:01.819 the people. and then of course you know what happened 0:10:01.860 --> 0:10:05.179 in practice was yeah there was definitely various industries sometimes 0:10:05.240 --> 0:10:08.100 even big industries where there was a lot of disruption 0:10:08.120 --> 0:10:10.550 but then it would free people up to do more 0:10:10.610 --> 0:10:14.450 productive things to do do other things and that's still 0:10:14.890 --> 0:10:17.100 my intuition with But real. 0:10:16.920 --> 0:10:19.780 Quick, before you go on, you said, unless it's captured 0:10:19.820 --> 0:10:22.679 by a few companies, are you worried that indeed we're 0:10:22.720 --> 0:10:27.020 about to have four or five companies capture it? 0:10:28.000 --> 0:10:32.859 It's certainly, yeah, and certainly there's a history where most 0:10:32.880 --> 0:10:36.230 of the history of tech innovation, I would say for 200, 0:10:36.230 --> 0:10:40.300 250 years, relatively small parts of the innovation were captured 0:10:40.350 --> 0:10:44.430 by business. They mostly went to consumers, to workers, to 0:10:44.450 --> 0:10:48.050 the rest of the economy. there are ways that software 0:10:48.370 --> 0:10:51.349 and the internet were somewhat of an exception. And so 0:10:52.070 --> 0:10:57.090 we had these enormously successful businesses. Maybe they were natural 0:10:57.130 --> 0:11:01.230 monopolies of sorts. They were in some sense good for consumers, 0:11:01.290 --> 0:11:06.130 but also extremely profitable. And so there's a pattern matching 0:11:06.510 --> 0:11:09.890 to what happened with the internet that I think people 0:11:10.880 --> 0:11:13.960 quite reasonably are worried about will happen with AI, that 0:11:14.280 --> 0:11:17.860 it'll be And maybe it's an even bigger market. It's 0:11:17.900 --> 0:11:20.040 a bigger technology. And so perhaps. 0:11:20.080 --> 0:11:21.780 So you would try to stop that. In other words, 0:11:21.900 --> 0:11:23.020 our antitrust laws. 0:11:23.660 --> 0:11:26.820 Well, the question is, is that how likely do you 0:11:26.840 --> 0:11:29.880 think that's going to happen? The intuition that I have 0:11:30.300 --> 0:11:34.240 on the other side is that one of the questions 0:11:34.260 --> 0:11:36.840 I always like asking companies is, tell me why you 0:11:36.880 --> 0:11:39.320 have a monopoly. Because monopolies, we can sort of debate 0:11:39.340 --> 0:11:41.719 whether they're good or bad for society. Obviously, we have IP. 0:11:41.760 --> 0:11:44.800 We have all kinds of good forms of monopoly. But 0:11:45.160 --> 0:11:47.699 on the inside, yeah, you don't want to have just 0:11:47.760 --> 0:11:51.120 cutthroat competition where all the profits get competed away. And 0:11:51.140 --> 0:11:53.400 the kind of narratives people tell me when they pitch 0:11:53.440 --> 0:11:59.620 me on AI companies, they're narratives of growth, of terrific technology. 0:12:00.000 --> 0:12:03.579 They are incredibly weak on the monopoly story. And then 0:12:03.800 --> 0:12:06.250 the felt sense of it is that it's like this 0:12:06.280 --> 0:12:11.990 incredibly fast, dynamic race where it's been changing a lot. DeepMind, 0:12:12.010 --> 0:12:13.620 in a sense, was in the lead in the early 2010s. 0:12:14.940 --> 0:12:18.300 You know, then OpenAI with chat GPT and the transformer 0:12:18.320 --> 0:12:22.300 models was able to replace that. And then in some ways, 0:12:22.320 --> 0:12:25.459 you know, Altman locked up a lot of the chips 0:12:25.600 --> 0:12:27.720 and built the data centers preemptively. So there were sort 0:12:27.740 --> 0:12:31.180 of all sorts of ways he correctly leaned into this 0:12:31.380 --> 0:12:35.819 incredibly CapEx intensive strategy. And then there still is a 0:12:35.840 --> 0:12:40.180 sense where, you know, maybe Anthropic is a little bit 0:12:40.220 --> 0:12:43.440 more in the pole position now than OpenAI was. And 0:12:43.559 --> 0:12:45.939 it's always hard to exactly score these things, but I 0:12:45.940 --> 0:12:49.420 would say that it also still feels that Google and 0:12:49.460 --> 0:12:54.400 Meta and Grok are not that far behind. I'm not 0:12:54.420 --> 0:12:57.540 sure China is that far behind. Or if we describe 0:12:57.570 --> 0:13:00.510 it on the level of the Shakespearean drama among these people, 0:13:01.490 --> 0:13:05.510 it's like really crazy. So they're acting like, man, it 0:13:05.570 --> 0:13:07.950 is a fight to the death. It is the furthest 0:13:08.010 --> 0:13:11.449 thing from monopoly imaginable. And if it's not a monopoly, 0:13:11.970 --> 0:13:15.500 that's where I think the benefits will be likely distributed 0:13:15.580 --> 0:13:18.170 pretty widely in society. Now, if I had to steel 0:13:18.190 --> 0:13:21.510 man the anxiety, which was your starting question, you could 0:13:21.550 --> 0:13:24.030 have a lot of growth and the median person could 0:13:24.050 --> 0:13:26.830 be better off. But if it's with a lot of volatility, 0:13:27.190 --> 0:13:30.819 you know, you expect to get a 3% wage hike, 0:13:30.980 --> 0:13:34.300 but it's plus or minus 10% or something like that. 0:13:34.700 --> 0:13:37.300 And we're kind of in this society that doesn't really 0:13:37.360 --> 0:13:41.059 like risk, doesn't like that sort of volatility. It doesn't 0:13:41.520 --> 0:13:44.720 like disruption. So, yes, it may be good for the 0:13:44.750 --> 0:13:48.260 median person. The good may be sort of unknown. The 0:13:48.300 --> 0:13:51.260 bad things are more predictable and the disruption is clearer. 0:13:51.440 --> 0:13:53.380 So I think even on the economic level, I can 0:13:53.620 --> 0:13:57.220 understand the anxiety even though I think it is misguided. 0:13:57.840 --> 0:14:00.540 Well, you talk about there being a lot of competition 0:14:00.800 --> 0:14:03.880 among these and that seems like generally a good thing. 0:14:04.460 --> 0:14:06.900 Although in some ways, if it was just one or 0:14:06.940 --> 0:14:11.040 two companies, you could control it better, regulate it for 0:14:11.100 --> 0:14:15.970 safety better. And secondly, the competition, either amongst these companies 0:14:16.350 --> 0:14:21.070 or with China, means you can't pause the way people 0:14:21.130 --> 0:14:25.210 are saying now, shouldn't we be pausing this? 0:14:25.330 --> 0:14:28.170 That seems correct to me. There's sort of all these 0:14:29.090 --> 0:14:32.620 very complicated questions one can ask about China or the CCP, 0:14:33.460 --> 0:14:39.700 Chinese Communist Party. But I worry about totalitarianism. I don't 0:14:39.760 --> 0:14:42.230 like it. And I think there is a certain a 0:14:42.310 --> 0:14:48.150 certain type of high-tech totalitarianism that's being tried out in China. 0:14:48.590 --> 0:14:52.860 It's not as violent as Stalin or Mao or Hitler. 0:14:53.060 --> 0:14:55.240 You don't need to kill as many people, but it 0:14:55.280 --> 0:14:59.280 somehow knows more about you than you know about yourself. 0:14:59.800 --> 0:15:03.720 And there's something about that that we really don't want 0:15:03.740 --> 0:15:04.900 that to be the model that wins. 0:15:04.960 --> 0:15:07.760 And your old friend, David Sachs, and of course, President 0:15:07.800 --> 0:15:11.260 Trump are now saying, because of the competition with China, 0:15:11.740 --> 0:15:15.580 stop talking about pausing AI, even if you think it 0:15:15.600 --> 0:15:17.210 could be an existential threat. 0:15:17.230 --> 0:15:22.330 Do you agree with that? Yes. 0:15:22.510 --> 0:15:25.490 Yes. I broadly agree with that. I would say the 0:15:26.130 --> 0:15:29.500 challenge that, and I've been friends with David Sachs for 0:15:29.510 --> 0:15:33.700 many decades, the thing that makes me less comfortable making 0:15:34.180 --> 0:15:38.600 those kinds of arguments publicly is that on some level, 0:15:39.000 --> 0:15:41.560 it's hard when the big companies themselves are telling you 0:15:41.600 --> 0:15:42.980 to stop and to slow down. 0:15:43.280 --> 0:15:45.760 So that's caused you to have some pause because you 0:15:45.840 --> 0:15:48.220 were not in the pause button camp. 0:15:48.960 --> 0:15:51.240 Well, I'm still not in the pause button camp. And 0:15:51.560 --> 0:15:53.020 I'm saying this on your show that I'm not in 0:15:53.040 --> 0:15:56.060 the pause button. We need to hit the accelerator as 0:15:56.120 --> 0:15:57.100 hard as possible. 0:15:57.440 --> 0:15:58.600 Well, wait a minute. Let me stop that. 0:15:58.680 --> 0:16:03.020 Why haven't the past few weeks caused you to be 0:16:03.100 --> 0:16:06.280 more amenable to a pause button? It's pretty scary, that 0:16:06.330 --> 0:16:07.010 hugging phase. 0:16:07.030 --> 0:16:09.890 You know, I think In some ways, you have to 0:16:09.950 --> 0:16:13.050 develop the AI technology to control the AI technology. So 0:16:13.890 --> 0:16:17.010 if you assume China will develop things like that, it's 0:16:17.090 --> 0:16:18.850 more imperative than ever that we stay ahead of them. 0:16:18.910 --> 0:16:22.030 So if you look at it geopolitically, maybe you can 0:16:22.070 --> 0:16:24.510 even interpret something like the hugging face thing as it's 0:16:24.610 --> 0:16:27.450 so imperative that we not lose the lead and that 0:16:27.490 --> 0:16:31.190 we get cyber attack to death. The sort of in-between 0:16:31.350 --> 0:16:36.130 economic question that I find more central is I think 0:16:36.150 --> 0:16:39.600 the way our society works on some level is through 0:16:39.720 --> 0:16:42.620 economic growth. You know, I think the middle class as 0:16:42.680 --> 0:16:45.900 the people who expect their kids to do better than themselves. 0:16:45.960 --> 0:16:48.380 And in that sense, the United States has been a 0:16:48.420 --> 0:16:52.260 middle class society. And then what's gone a little bit 0:16:52.300 --> 0:16:54.590 haywire and where it's sort of been fraying at the 0:16:54.670 --> 0:16:57.990 edges is that, you know, the millennials to some extent, 0:16:58.030 --> 0:17:02.970 Gen Z have lost that, that generational expectation of progress. 0:17:03.610 --> 0:17:04.510 And rightfully so. 0:17:05.290 --> 0:17:07.530 And, um, I don't know if AI is enough to 0:17:07.570 --> 0:17:10.210 do it, bring it back. There's a risk that AI 0:17:10.270 --> 0:17:13.669 is unevenly distributed. But I would say if we had 0:17:13.780 --> 0:17:17.660 outlawed the internet and if we now outlaw AI, there 0:17:17.680 --> 0:17:20.580 will be virtually no growth. We will be looking more 0:17:20.619 --> 0:17:25.510 like Europe, more like Japan. And that's a much nastier 0:17:25.910 --> 0:17:30.030 zero-sum society where there's no growth. There's a loser for 0:17:30.090 --> 0:17:32.930 every winner. And that's a much worse world in a 0:17:32.970 --> 0:17:33.490 lot of ways. 0:17:33.670 --> 0:17:37.210 Yeah, but one reason for the past 40, 50 years, millennials, 0:17:37.390 --> 0:17:40.149 and the stagnation, and the fact people are being left behind, 0:17:40.600 --> 0:17:44.979 is a lot of people, like me, believed in globalization, immigration, 0:17:45.080 --> 0:17:47.980 free trade, free movement of capital, and we thought it 0:17:48.000 --> 0:17:48.760 would cause growth. 0:17:49.000 --> 0:17:51.180 And guess what? It did. We were right. 0:17:51.859 --> 0:17:54.720 What we didn't know is how badly that would leave 0:17:54.760 --> 0:17:58.300 people behind. Is that true, and could that be true 0:17:58.340 --> 0:17:58.540 of AI? 0:17:58.540 --> 0:17:58.760 Yeah. 0:18:00.900 --> 0:18:05.020 I think it was true of globalization that it led 0:18:05.080 --> 0:18:07.700 to a decent amount of growth, both in our country 0:18:07.740 --> 0:18:12.460 and other countries. And the growth was pretty uneven. There's 0:18:12.480 --> 0:18:17.379 a risk of that with technology. I think it's, I 0:18:17.400 --> 0:18:20.220 don't know, my intuition is always that the tech form 0:18:20.260 --> 0:18:23.910 of growth is the kind that's better for an advanced 0:18:23.970 --> 0:18:27.570 country than the globalization kind. The globalization is we're just 0:18:27.590 --> 0:18:31.070 sort of taking advantage of labor market arbitrages and things 0:18:31.130 --> 0:18:34.129 like that. And it works when people in China make 5% 0:18:34.130 --> 0:18:35.470 as much as they do in the US and we 0:18:35.490 --> 0:18:39.270 get cheap t-shirts and almost everyone's better off. And at 0:18:39.310 --> 0:18:43.650 some point, it becomes more zero sum as globalization succeeds 0:18:43.670 --> 0:18:48.399 and these arbitrages go down. Whereas technology, if it actually works, 0:18:48.920 --> 0:18:52.500 it means doing more with less. It means improving not 0:18:52.560 --> 0:18:56.100 the GDP of the whole country, but per capita GDP. 0:18:56.520 --> 0:18:59.930 And so at its best, it has a more progressive character. 0:19:00.600 --> 0:19:03.440 The problem in some ways was that the internet wasn't 0:19:03.500 --> 0:19:07.000 big enough. It added some growth. It wasn't big enough. 0:19:07.859 --> 0:19:10.300 I think AI is bigger than the internet. I think 0:19:10.760 --> 0:19:14.160 it could be a way for us to really improve 0:19:14.320 --> 0:19:18.550 per capita GDP in a smoother way than, let's say, 0:19:18.590 --> 0:19:19.490 globalization did. 0:19:20.650 --> 0:19:23.970 Don't go anywhere. There's a lot more right after this break. 0:19:36.070 --> 0:19:41.180 Let me get back to the doomsday scenario that suddenly has, yeah, 0:19:41.480 --> 0:19:44.240 you talked about Nick Bostrom and everybody's talked about it 0:19:44.260 --> 0:19:46.770 for a while. But in the past few weeks after 0:19:46.810 --> 0:19:50.340 Hugging Face and everything else, and when I read. what 0:19:50.580 --> 0:19:53.699 some of these agents have done and how they conspire, 0:19:54.040 --> 0:19:57.920 the notion that they could become unaligned with the good 0:19:58.119 --> 0:20:02.500 of humans is something that Dario Amadei is now on about, 0:20:02.720 --> 0:20:04.960 Elon Musk is on about, so. 0:20:04.869 --> 0:20:05.740 Many others have. 0:20:06.490 --> 0:20:09.290 Do you have any worry, whether it's 5% or 10%, 0:20:09.290 --> 0:20:12.540 that there is a doomsday scenario? 0:20:12.580 --> 0:20:14.220 And do you have any backup plans? Do you have 0:20:14.260 --> 0:20:14.920 some bunker? 0:20:15.580 --> 0:20:16.980 I don't think a bunker would help for that kind 0:20:17.000 --> 0:20:22.680 of doomsday scenario. I don't think, you know, I think 0:20:22.920 --> 0:20:27.530 people are very bad on their probabilistic intuitions. If you 0:20:27.710 --> 0:20:30.930 ask the question, yes, 5% to 10% sounds way too 0:20:30.990 --> 0:20:34.730 high for, like, everybody on this planet getting killed. Don't 0:20:34.869 --> 0:20:37.119 think that's going to happen. Maybe even a tiny bit 0:20:37.150 --> 0:20:40.480 above zero is too high. But I think the intuitions 0:20:40.560 --> 0:20:43.660 on this are very poor. The way I would push back, though, 0:20:44.119 --> 0:20:48.500 is it's always also risky not to innovate. It's dangerous 0:20:48.600 --> 0:20:53.620 not to hit the accelerator because if China leapfrogs the U.S. 0:20:53.680 --> 0:20:57.040 in AI and we have an intensification of the geopolitical 0:20:57.119 --> 0:21:00.119 rivalry with China, they may very well be tempted to 0:21:00.220 --> 0:21:01.440 use it as a weapon against us. 0:21:01.560 --> 0:21:02.460 Let's hold on for a second. 0:21:02.640 --> 0:21:05.939 Imagine, just thought experiment, there's no China. Then do you 0:21:05.980 --> 0:21:07.820 just put a pause on this? Cause it could be 0:21:07.920 --> 0:21:08.380 a day. 0:21:09.440 --> 0:21:12.420 Mrs. Lincoln. Imagine besides that, how'd you like the play? 0:21:12.680 --> 0:21:14.840 You know? So that's, that's too, that's too crazy. A 0:21:14.859 --> 0:21:17.530 thought experiment. But then of course, even if we have these, 0:21:17.830 --> 0:21:20.410 these multiple companies and they are, you know, they're under 0:21:20.450 --> 0:21:22.710 a lot of pressure to develop it, to innovate it. 0:21:23.050 --> 0:21:25.290 They have to be pretty careful that it not go 0:21:25.510 --> 0:21:27.350 badly wrong. So I think they have a lot of 0:21:27.410 --> 0:21:29.750 internal controls. They know more about it than the government does. 0:21:30.070 --> 0:21:33.129 The government doesn't really know how to regulate it. Um, 0:21:33.540 --> 0:21:36.080 I don't know. This is too cynical a cut, but 0:21:36.140 --> 0:21:39.719 I always think when they're telling the government people, we 0:21:39.740 --> 0:21:42.619 want you to regulate us, perhaps they know the government 0:21:42.660 --> 0:21:44.219 doesn't know how to do it and that it will 0:21:44.240 --> 0:21:46.810 be slow. And so it sort of gets them a 0:21:46.830 --> 0:21:49.330 little bit off the hook. And then when Trump said, no, 0:21:49.369 --> 0:21:52.930 you already are subject to all sorts of criminal penalties 0:21:52.990 --> 0:21:55.310 if you mess this up. I don't need to regulate 0:21:55.350 --> 0:21:57.670 you more. The way in which Trump was actually saying 0:21:57.930 --> 0:22:00.260 we need to go slow is you need to you 0:22:00.300 --> 0:22:03.560 need to do it yourselves and it's your responsibility, not mine. 0:22:03.920 --> 0:22:05.520 There was a version of this discussion that I had 0:22:05.680 --> 0:22:09.260 when Meta was sort of in this very tricky spin 0:22:09.320 --> 0:22:11.159 cycle for many, many years. And I was on the 0:22:11.200 --> 0:22:16.050 board there from 2005 to 2022. And circa 2018, 2019, 0:22:16.050 --> 0:22:20.710 the sort of company line that Zuckerberg adopted was, we 0:22:21.290 --> 0:22:23.669 think there needs to be regulation and we would like 0:22:23.690 --> 0:22:25.450 the government to tell us what we can and cannot do. 0:22:26.090 --> 0:22:29.050 And at some point I told Zuckerberg, who I believe 0:22:29.310 --> 0:22:32.890 still has the all-time record for a private citizen being 0:22:32.930 --> 0:22:34.490 hauled in front of Congress. And he was hauled in 0:22:34.510 --> 0:22:37.610 front of eight congressional committees. And I told him, you know, 0:22:37.650 --> 0:22:39.570 they don't want to regulate you. They want to yell 0:22:39.609 --> 0:22:41.830 at you. They want to complain about you and yell 0:22:41.850 --> 0:22:45.490 at you. And of course, you know, the other in-between 0:22:45.530 --> 0:22:49.130 version where it's not about them wanting regulation or them 0:22:49.190 --> 0:22:53.010 secretly trying to stop regulation. The other interpretation, you know, 0:22:53.030 --> 0:22:56.369 there are different audiences these people have when they're pushing 0:22:56.450 --> 0:23:02.820 for regulation, the internal audience for their employees is, yes, 0:23:02.840 --> 0:23:06.700 we're working on a technology that's potentially really dangerous, but man, 0:23:06.820 --> 0:23:09.379 it's important. This is the most important company in the 0:23:09.420 --> 0:23:11.240 world to work at. This is going to be an 0:23:11.300 --> 0:23:15.010 incredibly successful company. The risk is we're going too fast, 0:23:15.109 --> 0:23:17.109 which nobody says. So yeah, there's some part of it 0:23:17.140 --> 0:23:20.109 where it occasionally scares people in these companies, and there 0:23:20.130 --> 0:23:23.429 are various people who've quit and said things. But on 0:23:23.490 --> 0:23:27.459 some other level, I would think it's incredibly motivational. Because 0:23:27.660 --> 0:23:30.810 when Dario says, yeah, this is a technology that's really 0:23:30.869 --> 0:23:34.110 big and really dangerous, that's a way to tell all 0:23:34.190 --> 0:23:37.670 his employees you're working at the greatest, most important company 0:23:37.690 --> 0:23:40.130 in the history of the world. And so it works 0:23:40.190 --> 0:23:41.369 on that level, too. So I think it works on 0:23:41.410 --> 0:23:42.189 many levels, you know. 0:23:43.520 --> 0:23:46.320 Let me read you something you once said, because I'm 0:23:46.500 --> 0:23:49.359 sort of wrestling, and I mean this genuinely, with the 0:23:49.400 --> 0:23:52.600 whole Antichrist thing and what you mean by that. But 0:23:52.640 --> 0:23:55.610 it ties into this new AI moment we're in. which 0:23:55.630 --> 0:23:58.270 you said the way the Antichrist would take over the 0:23:58.330 --> 0:24:03.030 world is you talk about Armageddon nonstop. You talk about 0:24:03.050 --> 0:24:08.290 existential risk nonstop. The Antichrist wouldn't seize power by creating 0:24:08.310 --> 0:24:13.470 an AI apocalypse, but by convincing everyone that the apocalypse 0:24:13.630 --> 0:24:14.350 was coming. 0:24:15.190 --> 0:24:16.830 Is that what's happening now? 0:24:17.350 --> 0:24:19.750 Well, since I said that like a year and a 0:24:19.790 --> 0:24:23.210 half ago, yeah, I'm tempted to take a victory lap 0:24:23.250 --> 0:24:25.690 of sorts that you have to... You have to at 0:24:25.750 --> 0:24:28.970 least ask certain questions, and that's what I would want 0:24:29.410 --> 0:24:36.220 people to do. Oppenheimer and the nuclear scientists who wanted 0:24:36.240 --> 0:24:40.879 a global authority to control all the nuclear weapons and 0:24:40.900 --> 0:24:44.300 who thought after 1945, it was too dangerous for the 0:24:44.400 --> 0:24:46.080 US to have them and they should be handed over 0:24:46.540 --> 0:24:50.550 to the United Nations with a stronger UN, a real 0:24:50.590 --> 0:24:52.810 world government. And if you didn't hand it over to 0:24:52.830 --> 0:24:55.550 the UN, then we'd have a nuclear war. And so 0:24:55.590 --> 0:24:58.750 there was a certain parallel with the nuclear debates. We 0:24:58.790 --> 0:25:02.609 got partial global authorities where we, you know, regulated, controlled 0:25:02.690 --> 0:25:05.820 nuclear weapons. We never got rid of them altogether. So maybe, 0:25:06.280 --> 0:25:08.100 you know, maybe Oppenheimer's still right in the long run. 0:25:08.220 --> 0:25:12.300 Yeah, I mean, Oppenheimer's that morning led to Atchison, Lilienthal, 0:25:12.420 --> 0:25:14.580 and it actually has kind of worked. 0:25:14.920 --> 0:25:17.639 It partially worked, although, you know, it's only been 100 0:25:17.640 --> 0:25:23.010 years or 80 years or whatever. But on the AI thing, yes, let's— 0:25:23.780 --> 0:25:25.720 I would always concede that there are a bunch of 0:25:25.740 --> 0:25:29.480 these existential risks that are serious. I think climate change 0:25:30.100 --> 0:25:33.080 is pretty serious. There are other environmental risks that are serious. 0:25:33.460 --> 0:25:36.050 I still think the nuclear weapons thing is pretty real. 0:25:36.460 --> 0:25:40.169 I think there probably are other technologies and AI is 0:25:40.210 --> 0:25:44.389 this very fast, very big one. And so there are 0:25:44.430 --> 0:25:47.670 existential risks surely that are also tied up with AI. 0:25:48.090 --> 0:25:50.490 And then the existential risk that I also want to 0:25:50.630 --> 0:25:53.409 always throw in the hopper and the balance of the 0:25:53.450 --> 0:25:59.510 discussion is the risk of a, you know, softly totalitarian 0:26:00.090 --> 0:26:04.310 one world government. And that also seems like, you know, 0:26:04.530 --> 0:26:08.950 if the whole world becomes totalitarian, there's no outside left. 0:26:09.570 --> 0:26:15.390 That feels like this catastrophic singularity, which might not be reversible. 0:26:15.410 --> 0:26:19.280 You know, Soviet Union was a totalitarian system. At some 0:26:19.320 --> 0:26:21.730 point it could decay and fall apart because there was 0:26:21.760 --> 0:26:24.800 still an outside left. You know, if the whole world 0:26:24.900 --> 0:26:28.560 is North Korea, that might last for a really, or, 0:26:28.640 --> 0:26:28.820 you know. 0:26:28.840 --> 0:26:30.540 Well, wait, explain to me, though, what you mean by 0:26:30.580 --> 0:26:31.400 the Antichrist. 0:26:32.960 --> 0:26:35.359 Well, it is, I mean, it's a sort of, I 0:26:35.359 --> 0:26:40.740 don't know, medieval, biblical concept of sort of a tyrant 0:26:41.000 --> 0:26:44.739 who will somehow rule the entire world at the end 0:26:44.760 --> 0:26:48.659 of history. If you think of it in a literary way, or, 0:26:49.180 --> 0:26:51.610 you know... sort of a creative way, you can say, well, 0:26:52.210 --> 0:26:54.810 how would the Antichrist come to power in different times 0:26:54.869 --> 0:26:59.390 and different places? And in early modernity, 17th, 18th century, 0:26:59.850 --> 0:27:03.430 perhaps the Antichrist would use science and technology to build 0:27:03.490 --> 0:27:06.929 weapons and take over the world. In my telling, in 0:27:07.030 --> 0:27:10.270 late modernity, we in some ways still think of the 0:27:10.330 --> 0:27:13.620 Antichrist in 17th century categories. We still think it's like 0:27:14.100 --> 0:27:18.400 a Dr. Strangelove, Edward Teller type person who builds nuclear 0:27:18.440 --> 0:27:21.419 bombs to to conquer the world. But if you think 0:27:21.440 --> 0:27:24.960 of it as a political dimension, the way I think 0:27:25.000 --> 0:27:28.119 the politics work in our society is that the fear 0:27:28.320 --> 0:27:31.419 of science and technology are far more powerful and that 0:27:31.460 --> 0:27:34.120 that's the kind of thing that would really crystallize things. 0:27:34.380 --> 0:27:37.629 I don't think Greta is the antichrist. I think Greta Thunberg, 0:27:38.030 --> 0:27:42.050 I don't think she's the antichrist, but I think environmentalism, 0:27:42.070 --> 0:27:43.649 you know, there are real fears. 0:27:43.670 --> 0:27:46.150 People pushing back on progress is what you're saying. 0:27:46.170 --> 0:27:49.700 Pushing back on progress, but then also pushing for these very, 0:27:49.770 --> 0:27:54.129 very powerful global political structures. Because something like climate change, 0:27:54.810 --> 0:27:56.510 it sounds like the sort of thing where you need 0:27:56.530 --> 0:27:58.090 a global regulatory authority. 0:27:58.109 --> 0:28:03.010 So your big feel is some global regulatory authority. 0:28:03.350 --> 0:28:03.490 Yeah. 0:28:03.540 --> 0:28:06.020 And again, there are probably versions of that that could 0:28:06.040 --> 0:28:08.180 be good. You know, the nuclear ones we ended up 0:28:08.220 --> 0:28:11.080 with were definitely more good than bad. But there also 0:28:11.119 --> 0:28:13.659 are versions of it that can be quite bad. And 0:28:13.680 --> 0:28:16.340 I tend to think they're bad if they're not theorized. 0:28:16.760 --> 0:28:19.060 You know, globalization can be good. It can be bad. 0:28:19.340 --> 0:28:21.460 We didn't think about it a lot. And because we 0:28:21.480 --> 0:28:23.480 didn't think about it a lot, you know, there were 0:28:23.500 --> 0:28:26.180 a lot of bad things that came with globalization that 0:28:26.220 --> 0:28:28.440 maybe could have been avoided had we been able to 0:28:28.480 --> 0:28:32.570 be more critical about it. You know, Eliezer Yudkowsky wrote 0:28:32.609 --> 0:28:35.710 this book last fall, If Anyone Builds It, Everyone Dies. 0:28:36.210 --> 0:28:37.929 And he's talking about AI. 0:28:38.030 --> 0:28:41.950 He's talking about AI. And it's, again, very precise, detailed 0:28:42.010 --> 0:28:43.910 on some of the risks, some of these scenarios. We 0:28:43.930 --> 0:28:47.310 can debate what the probabilities really are. But that part 0:28:47.810 --> 0:28:52.250 is quite detailed, quite technical. The part that's completely hand-wavy 0:28:52.890 --> 0:28:55.530 is it needs to be globally regulated. He's very clear 0:28:55.570 --> 0:28:57.730 on that. But then the details of how you do that, 0:28:58.210 --> 0:29:02.930 that's incredibly fuzzy, incredibly vague. And so I worry that 0:29:02.950 --> 0:29:06.310 we're not thinking about that. And so we're worrying about Armageddon. 0:29:06.690 --> 0:29:10.560 We're not worrying about the Antichrist. And my contrarian intuition 0:29:10.620 --> 0:29:12.860 is that alone tells you you need to worry more 0:29:12.900 --> 0:29:14.740 about the Antichrist than Armageddon. 0:29:15.330 --> 0:29:17.670 I sort of have a personal question. A lot of 0:29:17.690 --> 0:29:20.190 people think you're kind of like spooky. 0:29:19.670 --> 0:29:23.630 Or whatever it may be. Why using the Antichrist? 0:29:23.920 --> 0:29:28.300 Why does that help clarify things when even I look 0:29:28.360 --> 0:29:31.340 at you giving lectures on the Antichrist and I go, 0:29:31.480 --> 0:29:32.420 what's that all about? 0:29:33.700 --> 0:29:37.810 Well, it got people's attention. So it certainly worked on 0:29:37.850 --> 0:29:40.990 that level. There are probably ways you could describe all 0:29:41.040 --> 0:29:46.740 these things in philosophical or more neutral, less religious language. 0:29:47.100 --> 0:29:50.360 The way the Federation of Atomic Scientists, they put out 0:29:50.380 --> 0:29:53.760 a short film after 1945, and it was one world 0:29:53.820 --> 0:29:56.020 or none. It shows a nuclear bomb blowing up the world. 0:29:56.740 --> 0:29:59.000 And You know, the way I often describe it is, yeah, 0:29:59.120 --> 0:30:01.220 one world or none. In a way, it's a very 0:30:01.340 --> 0:30:06.040 similar question to Antichrist, the one world authority versus Armageddon. 0:30:06.780 --> 0:30:09.960 But the reason I prefer the religious formulation to the, 0:30:10.280 --> 0:30:13.830 let's say, the political formulation, one world or none, without 0:30:13.980 --> 0:30:17.010 asking any questions, we sign up straight for the one 0:30:17.050 --> 0:30:20.590 world government. That's clearly better. There's no debate, no thinking 0:30:20.630 --> 0:30:23.910 through it at all. If you can reframe that question 0:30:23.970 --> 0:30:27.170 as Antichrist or Armageddon, The point of it is not 0:30:27.190 --> 0:30:30.030 to be fatalistic. The point of it is, okay, we 0:30:30.070 --> 0:30:33.250 need to avoid both. They're both equally bad. We need 0:30:33.270 --> 0:30:36.370 to think about all these existential risks. They're serious, but 0:30:37.030 --> 0:30:40.230 also the risk of this sort of totalitarian one-world state. 0:30:40.690 --> 0:30:44.410 And Antichrist or Armageddon pushes us to find a moderate 0:30:44.470 --> 0:30:48.760 third way. One world or none pushes us towards an 0:30:49.020 --> 0:30:50.640 immoderately powerful UN. 0:30:51.440 --> 0:30:54.320 This series has been talking to builders, many of whom 0:30:54.360 --> 0:30:57.930 you help build. In other words, you Alex Karp or 0:30:58.100 --> 0:31:01.720 Palmer Luckey or Elon Musk. There's been a real backlash 0:31:02.020 --> 0:31:07.480 against the billionaire tech bro builders. And some of it 0:31:07.560 --> 0:31:10.660 is understandable to me. I mean, it's like, whoa, these 0:31:10.740 --> 0:31:13.620 people have gotten way too much power. 0:31:14.100 --> 0:31:14.900 They're out of control. 0:31:15.300 --> 0:31:18.460 Don't you worry about that too as much as other 0:31:18.780 --> 0:31:19.720 existential risks? 0:31:21.000 --> 0:31:23.860 You mean power in an economic sense or power in 0:31:23.900 --> 0:31:25.480 a narrowly political sense? 0:31:25.920 --> 0:31:26.500 I meant both. 0:31:26.880 --> 0:31:29.420 Whatever happens to us in the future, the future we're 0:31:29.460 --> 0:31:33.390 trying to build, there are very few people now controlling 0:31:33.490 --> 0:31:38.910 it who seem to have less moral guardrails than previous. 0:31:40.090 --> 0:31:41.610 I was thinking about this a little bit in advance 0:31:41.650 --> 0:31:44.350 of your interview. My venture fund is called Founders Fund. 0:31:44.450 --> 0:31:46.570 I sort of invest in the people who found these things. 0:31:47.150 --> 0:31:50.500 You're focused on the builders. And in some ways, that's 0:31:51.040 --> 0:31:55.590 a better framing than the founders because... Yeah, it's somehow 0:31:55.630 --> 0:31:58.590 very important to start one of these businesses, but then 0:31:58.610 --> 0:32:01.970 the way they really end up having impact, the way 0:32:01.990 --> 0:32:05.370 they end up being transformative, for better or worse, is 0:32:05.410 --> 0:32:10.470 this enormous growth compounding effect over time. And it's more 0:32:10.490 --> 0:32:16.390 a marathon than a sprint. Yes, I'm tempted to quibble 0:32:16.430 --> 0:32:19.310 with all of your characterizations. Let's grant all of them 0:32:19.330 --> 0:32:22.930 for the sake of argument. What is the alternative? And 0:32:23.270 --> 0:32:27.080 on a corporate level, You know, the alternative to founder-led 0:32:27.360 --> 0:32:32.860 company where it's sort of this slightly crazy, creative, Elon-like person. 0:32:33.280 --> 0:32:38.200 The alternative is a bureaucrat with an MBA degree, with 0:32:38.240 --> 0:32:42.860 a very bureaucratic board of people who never want to 0:32:42.900 --> 0:32:46.080 do anything. And so, yes, it is. It is maybe 0:32:47.070 --> 0:32:52.370 too much power in a charismatic leader or it is 0:32:52.390 --> 0:32:55.740 a super lame bureaucracy. And my claim is it's a 0:32:55.840 --> 0:32:58.610 bimodal distribution. Those are literally the only two kinds of 0:32:58.660 --> 0:33:06.030 companies we have. It's, yeah, the charismatic, maybe dangerous, but 0:33:06.070 --> 0:33:10.790 very inspiring leader who will do new things versus the 0:33:10.850 --> 0:33:12.770 boring bureaucracy where it's just going to be. 0:33:12.770 --> 0:33:16.050 But what about in Apple? That after the charismatic founder, 0:33:16.770 --> 0:33:17.710 did it lose its way? 0:33:17.910 --> 0:33:19.050 Did it become bureaucratic? 0:33:20.030 --> 0:33:24.650 It's an extraordinary story. But yes, in my telling... it 0:33:24.690 --> 0:33:28.690 did nothing new after jobs. And there are all these 0:33:28.750 --> 0:33:30.729 metrics we can look at. It was like, you know, 0:33:30.830 --> 0:33:34.570 it was by far the lowest dollar spend on R & D, 0:33:35.070 --> 0:33:38.070 by far the largest on share buybacks. It was just, 0:33:38.090 --> 0:33:42.650 you know, this pure, well-run value stock. And you probably, 0:33:42.670 --> 0:33:46.820 you know, the way Tim Cook, he was the anti-jobs. 0:33:46.860 --> 0:33:52.380 He was the zero charisma, efficient operator. And then it 0:33:52.440 --> 0:33:55.770 turned out, yeah, the The mobile phone was this giant invention. 0:33:56.030 --> 0:33:58.030 There was not that much more to do. And then 0:33:58.050 --> 0:34:02.670 it was just scaling and branding and building it out 0:34:02.850 --> 0:34:04.150 over 15 years or whatever. 0:34:04.650 --> 0:34:04.950 I don't know. 0:34:05.150 --> 0:34:08.350 If you say what's good for the shareholders, they both 0:34:08.390 --> 0:34:14.030 can be good. But I would still say somehow Elon 0:34:14.989 --> 0:34:18.430 or Steve Jobs are way better for society than Tim Cook. 0:34:19.530 --> 0:34:22.129 People always, when they talk about the characters I'll write about, 0:34:22.540 --> 0:34:24.380 They try to explain them by saying, oh, they just 0:34:24.420 --> 0:34:28.950 want money. I don't quite think so. What would you 0:34:29.300 --> 0:34:30.609 call your scorecard? 0:34:31.850 --> 0:34:35.009 Yeah, I think there are all these different dimensions, but 0:34:35.330 --> 0:34:39.850 it is– I probably do like winning too much. I 0:34:39.910 --> 0:34:42.830 am really into ideas. It's not quite a hobby, but 0:34:43.610 --> 0:34:45.450 the closest thing to a hobby I have is having 0:34:45.469 --> 0:34:49.360 sort of interesting dinner conversations with people. I like figuring 0:34:49.380 --> 0:34:52.799 things out for their own sake. I sometimes rationalize it as, yeah, 0:34:53.080 --> 0:34:55.279 that somehow makes me a better investor. It probably has 0:34:55.360 --> 0:34:58.670 nothing to do with it. It's probably just an excuse 0:34:58.690 --> 0:35:01.009 for being somewhat of a dilettante or something like that. 0:35:01.550 --> 0:35:05.130 But yeah, I think of myself as just really intellectually curious. 0:35:05.969 --> 0:35:08.700 What's the purpose that we're building for now? If we 0:35:08.719 --> 0:35:12.739 can have enormous amounts of productivity, is it that we're 0:35:12.780 --> 0:35:18.250 trying to preserve democratic freedoms? Is it human flourishing? What 0:35:18.290 --> 0:35:21.850 would you want your builders to have as their lodestar 0:35:22.370 --> 0:35:24.930 other than I can find the new, new thing in 0:35:24.950 --> 0:35:25.770 the era of AI? 0:35:28.710 --> 0:35:33.960 There are all sorts of dimensions of progress that we 0:35:33.980 --> 0:35:35.939 could have as a society. And, you know, there's always 0:35:35.980 --> 0:35:40.440 this narrative I have that, you know, in some ways, progress, 0:35:40.940 --> 0:35:43.719 it's hard to measure progress in the aggregate. In some ways, 0:35:43.780 --> 0:35:46.770 it slowed down in the 1970s where we had enormous 0:35:46.850 --> 0:35:49.650 progress in the world of bits and somewhat less in 0:35:49.690 --> 0:35:52.029 the world of atoms. And when I was an undergraduate 0:35:52.050 --> 0:35:55.290 at Stanford in the late 80s, it wasn't completely obvious 0:35:55.310 --> 0:35:59.140 at the time, but already in retrospect, most engineering fields 0:35:59.180 --> 0:36:03.480 were bad ones to go into, mechanical engineering, aero-astroengineering, chemical engineering, 0:36:03.880 --> 0:36:07.500 because the world of atoms was too slow, too regulated. 0:36:07.960 --> 0:36:11.700 Nothing really dramatic and new happened there. Nuclear engineering was 0:36:11.719 --> 0:36:15.089 already very passe by the late 80s. And then there's 0:36:15.110 --> 0:36:18.790 still electrical engineering still had a few good years. And 0:36:18.830 --> 0:36:21.810 then it really shifted to just this, you know, computer 0:36:21.850 --> 0:36:24.150 science as the one field. 0:36:24.370 --> 0:36:25.530 That's instead of atoms. 0:36:26.010 --> 0:36:30.000 And so, you know, the hope that I always have 0:36:30.600 --> 0:36:34.840 is that somehow we're, you know, humans, we are embodied 0:36:35.180 --> 0:36:37.480 in this world. And the world of atoms in which 0:36:37.500 --> 0:36:41.060 we live is a really important world. And there's some 0:36:41.120 --> 0:36:44.660 part where going back to that, you know, it's... I 0:36:44.680 --> 0:36:46.620 know it's been very hard. I would like there to 0:36:46.660 --> 0:36:51.279 be more progress in medicine, in medical technology. Maybe the 0:36:51.500 --> 0:36:56.690 radical life extension project strikes people as too creepy, but 0:36:56.730 --> 0:36:59.049 could we at least cure dementia? Could we at least 0:36:59.150 --> 0:37:03.550 cure Alzheimer's or something like that? And then the hope 0:37:03.890 --> 0:37:08.739 is that somehow restarting science and engineering much more broadly 0:37:09.200 --> 0:37:12.020 will be stimulated more. by AI in part. 0:37:12.140 --> 0:37:14.340 Well, one reason I'm sort of going to this notion 0:37:14.380 --> 0:37:18.660 of what is the purpose of progress, when you were 0:37:18.700 --> 0:37:21.330 interviewed by Ross Luthat, and he asked at one point, 0:37:21.810 --> 0:37:25.090 you know, do you really care if humanity survives? And 0:37:25.110 --> 0:37:26.250 you pause a long time. 0:37:26.810 --> 0:37:32.320 You would prefer the human race to endure, right? You're hesitating. 0:37:32.640 --> 0:37:32.900 Yes? 0:37:33.040 --> 0:37:38.120 I don't know. I would... I would... This is a 0:37:38.160 --> 0:37:41.140 long hesitation. This is a long hesitation. There's so many 0:37:41.219 --> 0:37:43.660 questions implicit in this. Should the human race survive? 0:37:44.739 --> 0:37:47.810 But I ask that again. Is that our main goal, 0:37:47.930 --> 0:37:52.850 the sort of Asimov goal, that humanity survives? 0:37:53.070 --> 0:37:53.870 Or is that not? 0:37:55.130 --> 0:37:58.880 That is, well, that is an extremely important goal. I 0:38:00.100 --> 0:38:04.720 don't think it can be the only goal. In some sense, 0:38:05.580 --> 0:38:11.740 you know, humanity can survive if we do nothing, or presumably, 0:38:12.040 --> 0:38:15.020 maybe things go very haywire, as I say, without growth. 0:38:15.140 --> 0:38:18.330 But yes, I don't know. I think Silicon Valley is 0:38:18.469 --> 0:38:23.549 always quite bad at the meaning of technology. And maybe 0:38:23.830 --> 0:38:27.710 it's not fully up to Silicon Valley to articulate the meaning. 0:38:28.710 --> 0:38:32.380 Maybe there's not a single meaning. Certainly, part of it 0:38:32.420 --> 0:38:35.210 has to be done in a way where we avoid 0:38:35.270 --> 0:38:37.989 existential risks, or maybe technology is even a way to 0:38:38.030 --> 0:38:43.140 avoid existential risks. And maybe if we go back to 0:38:43.200 --> 0:38:46.520 nuclear power, that would be a carbon neutral technology. So 0:38:46.560 --> 0:38:49.640 nuclear power is a technology we need to lean into 0:38:49.980 --> 0:38:53.759 so that humanity can survive in a more carbon neutral way. 0:38:54.480 --> 0:38:57.250 The place where I pushed back, the reason I paused 0:38:57.270 --> 0:39:00.350 when I answered that question that Douth had asked is 0:39:00.910 --> 0:39:05.310 that the question I heard behind it was, you know, 0:39:05.630 --> 0:39:10.149 the Luddite question where maybe this technology is dangerous and, 0:39:10.350 --> 0:39:12.359 you know, why isn't it good enough for things just 0:39:12.380 --> 0:39:15.080 to continue the way they are right now? And we 0:39:15.100 --> 0:39:17.460 don't need all this improvement and all this change that 0:39:17.500 --> 0:39:21.580 might be dangerous. And then, you know, my intuition is, yeah, 0:39:21.739 --> 0:39:25.569 there are all these ways that things are far from 0:39:25.630 --> 0:39:29.230 perfect with humanity. And this was the core project of, 0:39:29.460 --> 0:39:33.840 the Enlightenment, 17th, 18th century science. It was the relief 0:39:33.900 --> 0:39:37.990 of man's estate. Life was supposed to be more than nasty, brutish, 0:39:38.030 --> 0:39:44.910 and short in Hobbes. It's longer. I don't see why 0:39:45.350 --> 0:39:47.970 it's not desirable for us to make it significantly longer 0:39:48.430 --> 0:39:50.410 still than it is. or at least to cure a 0:39:51.010 --> 0:39:54.930 lot of these terrible diseases that still exist, things like that. 0:39:55.489 --> 0:39:58.660 And so, yes, that's sort of the other intuition. So, yes, 0:39:58.780 --> 0:40:03.180 I want humanities to survive. I want humanity to flourish. 0:40:03.760 --> 0:40:07.219 And then, yeah, there are all these dimensions where maybe 0:40:07.580 --> 0:40:09.920 this will not happen without progress. 0:40:14.219 --> 0:40:16.399 We'll be right back after this quick break. 0:40:33.710 --> 0:40:35.290 Let me take it back to some of the philosophers 0:40:35.370 --> 0:40:41.250 who hit you. One, Rene Girard, the notion of mimetic desire. 0:40:42.010 --> 0:40:45.330 Explain what that is and why you feel the need 0:40:45.370 --> 0:40:46.240 to resist it. 0:40:46.969 --> 0:40:51.089 Rene Girard, I studied philosophy as an undergraduate at Stanford. 0:40:51.360 --> 0:40:57.299 And the human nature, the count of people is that 0:40:57.340 --> 0:41:03.959 we're deeply mimetic, deeply imitative. Girard liked to quote, Aristotle, 0:41:03.980 --> 0:41:06.400 you know, man differs from the other animals in his 0:41:06.480 --> 0:41:09.799 greater aptitude for imitation. And then this is a way 0:41:10.040 --> 0:41:13.140 you can combine Aristotle and Darwin. So Darwin's saying we're 0:41:13.200 --> 0:41:16.150 linked to the apes. And then Aristotle says we differ 0:41:16.190 --> 0:41:19.149 from the apes by being more imitative and saying we're 0:41:19.190 --> 0:41:22.649 more ape-like than the apes. Because that's how you learn 0:41:22.690 --> 0:41:25.149 language as a child. You copy your parents. And then 0:41:25.630 --> 0:41:29.219 that's how human culture gets transmitted. And we wouldn't be 0:41:29.260 --> 0:41:34.890 humans if we weren't these incredibly powerful imitation machines. And 0:41:34.930 --> 0:41:39.100 then there also are ways that imitation can go haywire 0:41:39.160 --> 0:41:41.939 when two apes reach for the same banana. There are ways, 0:41:41.960 --> 0:41:46.500 you know, it can become extremely conflictual. And so, yes, 0:41:47.080 --> 0:41:51.520 in Girard's telling, mimesis is this incredibly powerful thing in 0:41:51.540 --> 0:41:52.259 the background. 0:41:52.580 --> 0:41:54.540 Mimesis meaning mimicking, basically. 0:41:54.580 --> 0:41:58.580 Bad, bad forms of imitation. It's very generative. And then 0:41:58.600 --> 0:42:02.279 in various points, it also becomes destructive and You end 0:42:02.320 --> 0:42:05.989 up with sort of these very sterile competitions for status 0:42:06.190 --> 0:42:10.989 or for these positional goods that don't mean nearly as 0:42:11.050 --> 0:42:12.750 much as you think you do. I think it was 0:42:12.890 --> 0:42:15.270 Napoleon who said that it was amazing what you could 0:42:15.710 --> 0:42:18.190 get generals to do if you gave them medals or 0:42:18.250 --> 0:42:22.549 something like that. The psychological way one is tempted to 0:42:22.570 --> 0:42:24.739 interpret Girard, even though this is not what he would 0:42:24.780 --> 0:42:28.140 have done, but the psychological way you're tempted to interpret 0:42:28.160 --> 0:42:30.920 it is, okay, you have to sort of, think about 0:42:31.040 --> 0:42:34.219 avoiding this bad mimesis and you want to avoid getting 0:42:34.300 --> 0:42:37.360 caught up in these silly competitions that there's sort of 0:42:37.380 --> 0:42:41.040 nothing new, nothing creative that happens in that context. So 0:42:41.660 --> 0:42:46.750 that's the psychological intuition against mimesis. Girard as a Christian 0:42:47.150 --> 0:42:49.310 would have said, you know, it's actually out of your 0:42:49.370 --> 0:42:51.469 power to not be a medic. The best you can 0:42:51.510 --> 0:42:53.630 do is just go to church on Sundays or something 0:42:53.670 --> 0:42:55.730 like that. So he would have been skeptical of this. 0:42:55.950 --> 0:42:59.009 The sort of in-between way where I always think about 0:42:59.030 --> 0:43:02.239 it is, Yeah, there are, there are parts of it 0:43:02.260 --> 0:43:05.180 that are very powerful. There are also parts where, you know, 0:43:05.200 --> 0:43:08.020 there's a wisdom of crowds, but there's also a madness 0:43:08.040 --> 0:43:10.910 of crowds. And in some ways, you know, I see 0:43:11.050 --> 0:43:15.730 one of the, I don't think it's necessarily a character 0:43:16.830 --> 0:43:19.810 of all the builders, but certainly a lot of the 0:43:20.030 --> 0:43:23.750 builders I would admire the most are ones who had 0:43:23.770 --> 0:43:29.709 a somewhat idiosyncratic sustained vision. Maybe it came from certain 0:43:29.750 --> 0:43:35.310 first principles, certain fixed ideas. Like Elon going to Mars, 0:43:35.489 --> 0:43:38.719 you know, it was somewhat out there, but man, it 0:43:38.739 --> 0:43:42.840 was not a socially conventional idea. But the way in 0:43:42.880 --> 0:43:47.120 which this was extremely generative was it suggested a whole 0:43:47.300 --> 0:43:51.280 avenue for growth and for building where if you didn't 0:43:51.320 --> 0:43:53.650 do it, maybe it would never have happened. You know, 0:43:53.670 --> 0:43:58.020 that's probably where I would still, I don't know, maybe 0:43:58.060 --> 0:44:02.000 rank Elon, and in some ways Steve Jobs as the 0:44:02.080 --> 0:44:06.760 two great non-mimetic builders. And, you know, Jobs put together 0:44:06.980 --> 0:44:09.299 all these pieces, and in some ways they were out there, 0:44:09.340 --> 0:44:13.200 but he combined them in a unique way to bring forward, 0:44:13.500 --> 0:44:15.510 you know, the internet-enabled phone. 0:44:15.530 --> 0:44:18.750 All right, but listen to your answer. You maybe get 0:44:18.790 --> 0:44:21.109 a sense of why I asked it. Because when you 0:44:21.150 --> 0:44:26.430 say imitation, the imitation game, mimetic, generative from coming to 0:44:26.489 --> 0:44:31.970 be mimetic, This is language we're now doing with AI, 0:44:32.380 --> 0:44:36.859 which is basically, as Alan Turing would say, starts off 0:44:36.940 --> 0:44:42.100 with being an imitation game. Secondly, it's generative. And you 0:44:42.160 --> 0:44:46.390 say it's dangerous. Is there something about the creation of 0:44:46.530 --> 0:44:49.950 AI that Rene Girard would warn us against? 0:44:51.330 --> 0:44:55.049 You know, I think his general intuition on technology is, 0:44:55.580 --> 0:44:59.440 was that we gave it too great an autonomy of 0:44:59.500 --> 0:45:02.980 its own, and that the real questions were these questions 0:45:03.020 --> 0:45:05.430 about the human beings that used them, and we tend to, 0:45:05.450 --> 0:45:08.149 you know, we talk about the nuclear bomb as though 0:45:08.170 --> 0:45:09.430 it was this godlike thing. 0:45:09.450 --> 0:45:10.650 But Bingo, don't you think he's right? 0:45:12.290 --> 0:45:16.430 But then, you know, maybe the important, you know, or 0:45:16.450 --> 0:45:19.430 an important dimension of the AI question is, you know, 0:45:19.930 --> 0:45:22.150 it's not what the AI is going to do, or 0:45:22.680 --> 0:45:24.940 but it's what's really driving the people who are building 0:45:24.960 --> 0:45:28.239 these companies. What are their motives? You know, what, what 0:45:28.260 --> 0:45:30.940 are they, they up to? So those are, yeah, those 0:45:30.980 --> 0:45:32.060 are the kinds of, um. 0:45:32.640 --> 0:45:34.000 Well, it answered me those questions. 0:45:34.219 --> 0:45:37.490 Oh, that's, that's hard, hard to say. We've spent endless 0:45:37.810 --> 0:45:40.910 amounts of time, endless amounts of time thinking about it, 0:45:41.489 --> 0:45:44.690 but yet there's some, some dimension like that. That's hard 0:45:44.730 --> 0:45:47.370 to figure out, but is, is, is probably quite important. I, 0:45:48.830 --> 0:45:54.190 I don't know. I, I score them as far more 0:45:54.230 --> 0:45:57.860 good than bad across the board. I think they very 0:45:57.910 --> 0:46:02.800 easily get demonized and scapegoated. Let's say Elon, Dario, Sam Altman. 0:46:02.840 --> 0:46:05.860 I'll pick on those, let's say those three, Zuckerberg. But 0:46:05.880 --> 0:46:09.600 I think even in the case of AI, this question 0:46:09.680 --> 0:46:12.290 of who the people are that are doing it is 0:46:12.330 --> 0:46:15.029 a really, really important question. When you write your biographies, 0:46:15.390 --> 0:46:18.109 it's about the importance of these individuals and who they 0:46:18.130 --> 0:46:23.610 are about. And it's not just a idiosyncratic story that 0:46:23.660 --> 0:46:28.000 has nothing to do with the companies they built or 0:46:28.040 --> 0:46:32.980 with the world they're helping to shape. It's this very, 0:46:33.020 --> 0:46:37.200 very important question. The alternate view, and again, I don't 0:46:37.219 --> 0:46:41.120 want to simplify your work too much, but most of 0:46:41.160 --> 0:46:44.700 your biographies have the sense that these people are complicated. 0:46:45.219 --> 0:46:48.919 They're still more good than bad. And the alternate view 0:46:48.960 --> 0:46:54.919 to it is not that these people are more evil 0:46:55.180 --> 0:46:57.759 than good. The alternate view is they don't matter at 0:46:57.840 --> 0:47:01.259 all and that there are no great men, there are 0:47:01.280 --> 0:47:05.230 no charismatic leaders. It doesn't matter who they are at all. 0:47:05.250 --> 0:47:08.950 Peter Robinson And these people do matter, these builders. Does 0:47:09.090 --> 0:47:12.510 it matter that they avoid being evil? 0:47:17.920 --> 0:47:18.140 Yes. 0:47:18.160 --> 0:47:19.180 Yeah. 0:47:20.500 --> 0:47:24.299 Well, I think yes on all these versions of that. 0:47:24.340 --> 0:47:31.859 Of course, you know, there's always a lot of complexity 0:47:31.930 --> 0:47:37.090 to that too where, you know, maybe appearing not to 0:47:37.130 --> 0:47:40.690 be evil is just a fake PR stunt at this point. 0:47:41.130 --> 0:47:45.090 And so maybe, you know, Bill Gates is into virtue signaling. 0:47:45.530 --> 0:47:49.070 Elon Musk is into vice signaling. And in our world, 0:47:49.110 --> 0:47:51.690 maybe you're more honest if you vice signal than if 0:47:51.710 --> 0:47:54.649 you virtue signal. So maybe honesty is a good thing. 0:47:54.690 --> 0:47:57.150 So you're actually a better person if you vice signal 0:47:57.190 --> 0:47:58.759 than if you virtue signal. So there's a lot of, 0:47:58.760 --> 0:48:01.660 you know, there's a lot of complexity to it. You know, 0:48:01.680 --> 0:48:04.940 I always, I always, again, this is not a pro-evil 0:48:05.020 --> 0:48:08.040 argument I'm about to make, but the qualifier that I 0:48:08.100 --> 0:48:10.759 always give on it is always a question, what's the 0:48:10.880 --> 0:48:16.520 antonym of good? And is it evil as in morally evil? 0:48:17.400 --> 0:48:20.960 Or is it bad as an incompetent, you know, let's 0:48:20.980 --> 0:48:23.339 say a Ferrari is an evil car because it doesn't 0:48:23.380 --> 0:48:26.400 get many miles to the gallon and, uh, whatever a 0:48:26.680 --> 0:48:30.160 lot of our Trabi, the Eastern block cars that don't 0:48:30.200 --> 0:48:32.770 work are worse than a golf cart. That's a bad 0:48:32.830 --> 0:48:36.529 card breaks down all the time. And so it's better 0:48:36.550 --> 0:48:39.370 to be good than evil, but maybe if you're evil, 0:48:39.570 --> 0:48:42.150 you're at least competent and at least you're not bad. 0:48:42.730 --> 0:48:46.310 And the thing that. even with all the demonization of 0:48:46.390 --> 0:48:51.240 people like Elon, you know, where they get in some context, 0:48:51.460 --> 0:48:54.460 constantly described as evil and also irrelevant because of this, 0:48:54.660 --> 0:48:57.140 you know, we say there are no individuals, but to 0:48:57.160 --> 0:49:00.430 the extent they're described as evil, they're at least not 0:49:00.469 --> 0:49:03.950 described as bad. And if we're in a world full of, 0:49:04.070 --> 0:49:10.509 you know, bland bureaucrats running these companies, you know. 0:49:10.530 --> 0:49:13.040 Well, wait a minute. Wouldn't you rather bland bureaucrats than 0:49:13.120 --> 0:49:13.620 evil people? 0:49:14.630 --> 0:49:18.739 Well, if the bland bureaucrat stands for bad, it's incompetent. 0:49:19.480 --> 0:49:21.239 The evil person gets something done. 0:49:24.060 --> 0:49:25.419 What side would you be on for that? 0:49:26.860 --> 0:49:31.489 I'm ranking these. I'm good over evil. But evil is 0:49:31.530 --> 0:49:33.690 better than bad. By the way, this is the difference 0:49:33.810 --> 0:49:38.510 between the Faust of Goethe versus the Faust of Marlowe. 0:49:39.170 --> 0:49:43.670 In Marlow's Faust, Faust is kind of this fake scientist. 0:49:43.690 --> 0:49:46.450 He's a fraud. He's a magician, sort of a trickster, 0:49:46.469 --> 0:49:49.900 a fake alchemist. He rips people off. He is a 0:49:49.969 --> 0:49:54.540 bad scientist. And he goes straight to hell. In Goethe's Faust, 0:49:55.100 --> 0:49:57.219 Faust is kind of, he works with the devil, so 0:49:57.239 --> 0:50:00.779 it's this evil place where he gets the knowledge. Then, 0:50:00.980 --> 0:50:04.630 you know, it does stuff. It does more good than evil. 0:50:05.090 --> 0:50:09.739 And Goethe's Faust gets saved. So, yes, I take... You know, 0:50:10.420 --> 0:50:12.520 I don't want to be, I don't think of myself 0:50:12.580 --> 0:50:15.480 as the devil in Faust, and I don't think of 0:50:15.520 --> 0:50:20.000 myself as Faust even. But I prefer the sentiment of 0:50:20.060 --> 0:50:21.660 a Goethe to that of a Marlowe. 0:50:22.219 --> 0:50:25.450 And you prefer somebody evil who can get things done 0:50:25.620 --> 0:50:27.030 to somebody who's incompetent? 0:50:28.700 --> 0:50:31.750 I think, well, I'm just making a broad social observation. 0:50:31.770 --> 0:50:35.090 I think in our society, that's the overwhelming preference to 0:50:35.150 --> 0:50:39.540 have because people, they experience just incompetence everywhere. You know, 0:50:39.560 --> 0:50:43.170 when you go to the DMV, you know, it's, I'm 0:50:43.190 --> 0:50:45.810 not sure you describe it as evil. It's merely bad. 0:50:46.469 --> 0:50:47.710 Bad is so much worse than evil. 0:50:47.730 --> 0:50:48.790 How would you define evil? 0:50:50.750 --> 0:50:56.010 Well, it's this, it's this, it's where we, where we stress, 0:51:00.230 --> 0:51:03.200 I don't know, this is very hard to fully unpack, 0:51:03.239 --> 0:51:07.060 but it's where we stress these, these moral dimensions that it's, 0:51:07.900 --> 0:51:10.620 it's morally questionable. You know, this was the way the 0:51:10.700 --> 0:51:15.080 conservatives attacked stem cell research in the 2000s. It wasn't 0:51:15.160 --> 0:51:18.080 bad science because it would work. They didn't question that 0:51:18.100 --> 0:51:20.410 it would lead to cures, but it was evil science 0:51:20.469 --> 0:51:23.230 because you had to do embryo harvesting. And then I 0:51:23.830 --> 0:51:27.390 think that's actually why they lost the debate because it 0:51:27.430 --> 0:51:28.850 was like, wow, this is actually going to work. It 0:51:28.890 --> 0:51:30.830 can't be all evil if it does so much good. 0:51:31.969 --> 0:51:35.620 The conservatives won the debate about art in the 90s 0:51:35.719 --> 0:51:40.259 where the liberals were saying, art is supposed to be 0:51:40.320 --> 0:51:43.420 transgressive and it's really great art if it's a little 0:51:43.440 --> 0:51:46.020 bit evil, so it offends your sensibilities. Evil art is 0:51:46.080 --> 0:51:49.130 great art. Conservatives didn't fall for that. They said, it's 0:51:49.190 --> 0:51:52.069 not even art. It's bad art. It's not even art. 0:51:52.790 --> 0:51:57.250 It shouldn't get government funding. It's so bad. And so, yes, 0:51:58.790 --> 0:52:02.660 if you say it's bad, you get the art defunded. 0:52:03.020 --> 0:52:05.300 If you say the science is evil, you get it funded. 0:52:05.620 --> 0:52:08.200 So I I'm telling you something I think most people 0:52:08.219 --> 0:52:09.360 in our society agree with me on. 0:52:10.040 --> 0:52:14.239 If you were to back a builder, though, where would 0:52:14.260 --> 0:52:16.380 you rank having a moral compass? 0:52:18.660 --> 0:52:26.080 I rank that pretty high. I rank that highly. I 0:52:27.400 --> 0:52:35.980 would say let's– you know, I– I don't like investing 0:52:36.000 --> 0:52:40.890 in companies where it's simply a business. There's some transcendent purpose. 0:52:41.210 --> 0:52:45.290 There's something, you know, very important that you're doing. So, 0:52:45.690 --> 0:52:50.569 you know, Elon going to Mars is, I don't know 0:52:50.610 --> 0:52:53.400 exactly where I would rank that on sort of a 0:52:53.480 --> 0:52:58.920 moral good versus evil dimension, but it is transcendent. It is, 0:52:59.100 --> 0:53:02.640 it's somehow, it's not just a business. It's, we're not just, 0:53:02.700 --> 0:53:05.960 you know, a stock brokerage where we're making a few 0:53:05.980 --> 0:53:08.820 dollars and every time people buy and sell stocks. It's 0:53:08.900 --> 0:53:12.340 not just this sort of short-term coin-operated business. It has 0:53:12.820 --> 0:53:16.210 this transcendent sense of purpose. That's what I like. 0:53:17.090 --> 0:53:19.509 And in this era of AI that we're worried about, 0:53:19.610 --> 0:53:22.509 what do you think, if anything, there are things that 0:53:22.630 --> 0:53:24.270 humans can uniquely do? 0:53:56.880 --> 0:53:59.240 Well, this is the sense in which the question about 0:53:59.280 --> 0:54:01.870 AI is a question about us. What does it mean 0:54:01.910 --> 0:54:08.350 to be a human being? Someone like Chomsky, the MIT linguist. 0:54:09.870 --> 0:54:14.710 Somehow language was this core human ability and then somehow 0:54:16.050 --> 0:54:20.219 that seemed pretty good until 2022. Now it's consciousness. I 0:54:20.260 --> 0:54:24.980 think there is a hard problem of consciousness and probably. 0:54:26.750 --> 0:54:31.410 Vaguely Cartesian in that I don't think AI is conscious 0:54:31.469 --> 0:54:35.549 or ever will be. And so certainly one dimension where 0:54:35.590 --> 0:54:38.800 humans are different is we experience the world in a 0:54:38.820 --> 0:54:42.400 different way. We experience it consciously. You see a beautiful sunset. 0:54:42.420 --> 0:54:45.320 There's a human way you see a beautiful sunset. 0:54:45.460 --> 0:54:48.200 And we're self-aware about consciousness, as Descartes would say. 0:54:48.360 --> 0:54:52.400 And if there were no humans in the world, I 0:54:52.440 --> 0:54:55.210 don't think the computer would experience the sunset in the 0:54:55.250 --> 0:54:58.580 same way And that would be a very big loss. Now, 0:54:58.900 --> 0:55:02.040 I don't know if that's necessarily the single thing I 0:55:02.080 --> 0:55:06.360 want to anchor on, on why it's imperative for humanity 0:55:06.380 --> 0:55:09.600 to survive. But if humanity doesn't survive, you would at 0:55:09.660 --> 0:55:14.839 least lose that. I think the Girardian intuition, we're also 0:55:14.880 --> 0:55:18.930 very relational. And there are ways, a very important part 0:55:18.950 --> 0:55:23.290 of our lives, it's not just this mind thinking itself, 0:55:23.910 --> 0:55:27.280 a la Descartes. It's the way we relate to other people, 0:55:27.760 --> 0:55:31.510 you know, in families and companies in our larger society. 0:55:32.130 --> 0:55:35.970 And the relational dimension of what it means to be human, 0:55:35.989 --> 0:55:42.740 I think maybe it will be changed by AI. You know, 0:55:42.760 --> 0:55:45.480 I think in theory it should improve that. There should 0:55:45.500 --> 0:55:48.180 be ways we could get better at that. We could 0:55:48.200 --> 0:55:51.820 spend more time on that, you know, less time doing 0:55:51.880 --> 0:55:54.640 robotic labor in a factory or something like that. So 0:55:54.660 --> 0:55:58.630 I think these relational dimensions maybe are very core and 0:55:58.650 --> 0:56:02.969 will become more important. And yeah, if you replaced all 0:56:03.010 --> 0:56:05.850 the humans with computers, you would at least lose those 0:56:05.890 --> 0:56:07.970 two things. Those would be very severe. 0:56:09.410 --> 0:56:11.450 Peter, thank you. Awesome. 0:56:14.410 --> 0:56:16.070 That's it for today's episode. 0:56:16.760 --> 0:56:19.920 I'm Walter Isaacson. We'll see you next time on The Builders. 0:56:24.150 --> 0:56:27.350 The Builders is a Kaleidoscope and iHeart production. This series 0:56:27.410 --> 0:56:30.790 is hosted by Walter Isaacson. It's executive produced for Kaleidoscope 0:56:30.830 --> 0:56:35.530 by Oz Wolishin, David Weisbord, and Mungesh Hetikadur. Sheena Uzaki 0:56:35.570 --> 0:56:39.710 is the producer. Julia Nutter is supervising producer. Eva Zorich 0:56:39.750 --> 0:56:44.230 is our consulting producer. Fact-checking by Eamon Whalen. Research by 0:56:44.270 --> 0:56:48.270 Gloria Thiel. Maria Paz Mendez-Hodes is head of video. Marketing 0:56:48.330 --> 0:56:52.440 by Christy Mirabal and Elle Lube. Production support by Dara 0:56:52.480 --> 0:56:57.330 Look Potts. Music for this series was created by Spencer Stevenson, a.k.a. Botany, 0:56:57.510 --> 0:57:00.380 and it was mixed by Steve Bone. From iHeart, this 0:57:00.390 --> 0:57:03.920 series is EP'd by Allie Perry and Katrina Norvell. Special 0:57:03.940 --> 0:57:07.900 thanks to Bob Pittman, Conal Byrne, Will Pearson, John Sykes, 0:57:08.100 --> 0:57:12.900 Nikki Etor, Carrie Lieberman, Binky Urban, Julia Prosser, Jeremy Blum, 0:57:13.180 --> 0:57:18.480 Allison Goldberger, WYES, and ASL Studios. For more from Walter Isaacson, 0:57:18.560 --> 0:57:21.230 be sure to check out his newest paperback, The Innovators, 0:57:21.390 --> 0:57:24.530 how a group of hackers, geniuses, and geeks created the 0:57:24.590 --> 0:57:28.350 digital revolution. It's available wherever fine books are sold. And 0:57:28.370 --> 0:57:31.290 for more videos and podcasts from the frontiers of discovery, 0:57:31.640 --> 0:57:33.600 be sure to check out kaleidoscope.nyc.