WEBVTT - AI Economic Impact Overstated, Says MIT's Acemoglu 0:00:02.520 --> 0:00:06.680 You're listening to Asia Centric from Bloomberg Intelligence, the podcast 0:00:06.720 --> 0:00:09.560 that pulls back the curtain on global business so you 0:00:09.600 --> 0:00:13.119 can invest better across the Asia Pacific rim. I'm John 0:00:13.200 --> 0:00:15.360 Lee in Hong Kong, and I'm. 0:00:15.280 --> 0:00:19.800 Kaite Dmitrieva with Bloomberg News, also in Hong Kong. We're 0:00:19.840 --> 0:00:23.560 looking at AI this week and specifically how it might 0:00:23.600 --> 0:00:26.759 not be this big growth driver that markets and some 0:00:26.920 --> 0:00:30.000 analysts seem to think it might be. On the high end, 0:00:30.280 --> 0:00:34.240 McKenzie sees up to a twenty six trillion dollar boost 0:00:34.320 --> 0:00:37.919 to the global economy. The top economist at Goldman Sachs 0:00:38.080 --> 0:00:41.040 sees it adding nine percent to US productivity. 0:00:41.760 --> 0:00:46.000 Yes, but not everyone is this optimistic. According to Sequoia, 0:00:46.560 --> 0:00:49.960 AI companies need to earn at least six hundred billion 0:00:50.159 --> 0:00:54.720 in revenue to justify current AI infrastructure spend. Seems like 0:00:54.760 --> 0:00:57.480 we're going nowhere near this figure, and our guest is 0:00:57.480 --> 0:00:58.279 also critical. 0:00:59.120 --> 0:01:02.440 I'm excited about our guest this week, John, because he's 0:01:02.480 --> 0:01:05.360 one of the big thinkers on AI, both in terms 0:01:05.360 --> 0:01:09.600 of the social and economic impacts. So to walk us 0:01:09.600 --> 0:01:13.960 through this brave new world is darron a Smoglu, Institute 0:01:14.000 --> 0:01:18.720 Professor at MIT and author of books, including Why Nations Fail. 0:01:19.240 --> 0:01:23.880 He's joining us from Cambridge, Massachusetts. Welcome Deren, Hi Katia, 0:01:23.920 --> 0:01:24.360 Hi John. 0:01:24.400 --> 0:01:27.200 It's great pleasure to be with you to talk about AI. 0:01:27.640 --> 0:01:31.000 We're glad to have you. Darren. You found in a 0:01:31.040 --> 0:01:35.560 recent paper that AI will impact something like less than 0:01:35.760 --> 0:01:39.560 five percent of all tasks. And you pulled back and 0:01:39.600 --> 0:01:42.920 you looked at the effects on the US economy, finding 0:01:42.920 --> 0:01:46.400 that US productivity would rise by only about zero point 0:01:46.440 --> 0:01:50.280 five percent and GDP growth by zero point nine percent 0:01:50.400 --> 0:01:52.520 over the next decade. So it's a lot of numbers there, 0:01:52.560 --> 0:01:56.080 but it's overall just quite a bit lower than estimates 0:01:56.160 --> 0:01:59.680 from many of your colleagues and economists tell us about that. 0:01:59.680 --> 0:02:02.440 Why do you think the impact will be less than 0:02:02.600 --> 0:02:03.600 what people think? 0:02:04.680 --> 0:02:06.559 Yeah, I mean, I think the bottom line is exactly. 0:02:06.640 --> 0:02:09.040 It's a bit, quite a bit lower than other numbers 0:02:09.040 --> 0:02:12.680 that are being floated around, and I think it reflects 0:02:12.760 --> 0:02:17.800 a deeper issue, which is that we are very much 0:02:18.040 --> 0:02:23.120 at the beginning of the AI technological journey, wherever that 0:02:23.240 --> 0:02:27.919 might be, wherever that might go, and there aren't many 0:02:28.360 --> 0:02:33.680 applications that can be transformative. Yet if you look at 0:02:33.720 --> 0:02:36.720 AI what it can do right now, it can provide 0:02:36.720 --> 0:02:40.440 a little bit of better information to a few decision makers, 0:02:41.000 --> 0:02:43.840 and it can perform a few tasks, but it does 0:02:43.880 --> 0:02:47.959 not have the capability to do much in any task. 0:02:48.040 --> 0:02:52.240 In any occupation that involves a central element of interacting 0:02:52.280 --> 0:02:55.840 with the real world, such as those in construction or manufacturing, 0:02:56.680 --> 0:03:00.680 or those that involve moving things around the real world, 0:03:00.680 --> 0:03:04.280 it cannot at the moment have a big input into 0:03:04.320 --> 0:03:09.360 things that are social in nature, for example, psychiatry, entertainment, 0:03:09.960 --> 0:03:13.520 things that involve many individuals coming together and using their 0:03:13.600 --> 0:03:17.720 judgment or team interactions. So once you exclude all of 0:03:17.760 --> 0:03:21.560 these occupations, there are a bunch of white colored things 0:03:21.600 --> 0:03:25.160 that people do in their offices that require better information, 0:03:25.400 --> 0:03:29.440 that can benefit from better processing of language or data, 0:03:29.639 --> 0:03:31.840 and those are the things that AI can at the 0:03:31.880 --> 0:03:36.480 moment have a moderate, non trivial, but a moderate effect 0:03:36.960 --> 0:03:39.960 in helping us do better. And that's at the root 0:03:40.080 --> 0:03:42.000 of both my numbers and the fact that we don't 0:03:42.040 --> 0:03:43.520 have the killer apps for AI. 0:03:43.360 --> 0:03:46.560 Yet, Darren. If you listen to a lot of people 0:03:46.600 --> 0:03:49.360 on AI, they say it's going to be transformative for 0:03:49.640 --> 0:03:52.400 most industries, but you seem to take a different view, 0:03:52.840 --> 0:03:56.480 like what percentage of jobs do you think AI will 0:03:56.520 --> 0:03:57.120 have an impact? 0:03:58.040 --> 0:04:00.920 Well, I think that's a very difficult question to answer, John, 0:04:01.520 --> 0:04:05.240 because it depends on what horizon we're talking about, and 0:04:05.280 --> 0:04:08.680 what types of investments and what type of direction of 0:04:08.720 --> 0:04:12.800 AI we choose for the future. So when people talk 0:04:12.960 --> 0:04:17.360 of absolutely transformative effects of AI, I think they are 0:04:17.760 --> 0:04:22.120 mixing two things. One is the capabilities of AI as 0:04:22.120 --> 0:04:26.000 we have it today, and the second is where AI 0:04:26.320 --> 0:04:31.320 might go, for example, something close to superintelligence. It's a 0:04:31.360 --> 0:04:34.400 hope or nightmare of some people. And of course, if 0:04:34.440 --> 0:04:37.680 AI is so smart that it can do many many things, 0:04:37.680 --> 0:04:39.599 and it can be integrated with robots, it can start 0:04:39.680 --> 0:04:42.880 driving cars and airplanes, that is a very different story 0:04:42.920 --> 0:04:44.080 than the kind of AI. 0:04:43.800 --> 0:04:44.440 We have today. 0:04:45.200 --> 0:04:49.320 The other difficulty in answering that question is that there 0:04:49.360 --> 0:04:53.520 is sometimes a presumption that somehow there's a single direction 0:04:53.600 --> 0:04:56.720 of AI, that we are just moving towards that direction, 0:04:56.800 --> 0:04:58.720 and the only choice we have is whether it's going 0:04:58.760 --> 0:05:00.760 to be China or the US would gets there and 0:05:01.440 --> 0:05:03.960 how quickly we're going to get there. The truth couldn't 0:05:03.960 --> 0:05:06.960 be farther from that. There are so many different things 0:05:06.960 --> 0:05:09.080 we can use AI, like technologies. At the end of 0:05:09.120 --> 0:05:11.440 the day, what we're calling AI today is machine learning 0:05:12.279 --> 0:05:17.599 large capacity for effectively processing data. We can use that 0:05:17.760 --> 0:05:21.320 for automation, we can use that for science, we can 0:05:21.440 --> 0:05:24.200 use that for creating better information for a variety of 0:05:24.279 --> 0:05:28.720 different occupations. So depending on which direction we take and 0:05:28.800 --> 0:05:33.320 how quickly corporate investors are convinced to jump on the 0:05:33.320 --> 0:05:36.560 AI bandwagon, its effects are going to be different than 0:05:36.560 --> 0:05:38.480 it's effects that are going to be more or less pervasive. 0:05:40.360 --> 0:05:43.200 I mean, there's certainly investors who have jumped on the 0:05:43.240 --> 0:05:48.240 AI bandwagon. I guess the question is have they done 0:05:48.279 --> 0:05:50.560 it too soon? Yeah? 0:05:50.600 --> 0:05:53.599 Absolutely, I think, Katya, that's the key question. The reason 0:05:53.600 --> 0:05:57.440 why I wrote the paper that you mentioned, Katya, is 0:05:57.480 --> 0:06:01.760 because I think the AI hype is very counterproductive for 0:06:01.800 --> 0:06:06.680 two reasons. First, if you look at the narrative in 0:06:06.680 --> 0:06:09.880 the United States in the second half of twenty twenty 0:06:09.880 --> 0:06:12.000 three and in the first half of twenty twenty four, 0:06:12.760 --> 0:06:16.719 you will see that managers in many companies, not just 0:06:16.720 --> 0:06:20.200 publicly traded companies, but in many companies are under tremendous 0:06:20.200 --> 0:06:25.760 pressure to invest in AI ads. New journalists, their colleagues, 0:06:26.320 --> 0:06:29.800 management consultants are continuously on their next saying what have 0:06:29.839 --> 0:06:32.520 you done on AI, are you falling back behind your competitors, 0:06:33.160 --> 0:06:38.360 And the outcome is inevitably people throwing money into AI 0:06:38.440 --> 0:06:41.360 without knowing what they're going to use it for, nor 0:06:41.560 --> 0:06:44.960 having the technology be ready for doing useful things for 0:06:45.040 --> 0:06:49.719 most companies. The second is that the AI hype, fueled 0:06:49.760 --> 0:06:52.440 by a few tech companies and a few tech journalists 0:06:52.920 --> 0:06:56.599 is actually cementing the current direction of AI and the 0:06:56.640 --> 0:07:02.800 current structure of the industry. Open AI wants trillions of 0:07:02.800 --> 0:07:06.880 dollars of investment. Why because they believe they have the 0:07:06.960 --> 0:07:09.120 right business model and they want to be the leader. 0:07:09.200 --> 0:07:12.040 They don't want anybody else to catch up, and the 0:07:12.080 --> 0:07:15.400 best way of doing that is actually exciting investors to 0:07:15.640 --> 0:07:19.240 want to invest more in AI. But first of all, 0:07:19.480 --> 0:07:23.080 it may well be that global society is going to 0:07:23.080 --> 0:07:25.400 be better served by a different technology, more open source, 0:07:25.480 --> 0:07:27.720 or a different approach to AI, or even if we're 0:07:27.720 --> 0:07:30.080 going to adopt the same approach, we may want more competition. 0:07:30.440 --> 0:07:33.400 We may want smaller startups to sort of be the 0:07:33.440 --> 0:07:37.640 ones that grow. So those are all things that investors 0:07:38.280 --> 0:07:42.040 attention and focus are going to influence majorly. But even 0:07:42.040 --> 0:07:44.880 more importantly from the point of view of my research, 0:07:45.120 --> 0:07:47.560 is the direction of AI. I think that the current 0:07:47.600 --> 0:07:52.760 emphasis on automation and using AI for manipulating users on 0:07:52.800 --> 0:07:56.920 social media or better ads and better sort of ways 0:07:57.040 --> 0:08:00.680 of capturing consumers is not the most socially active one. 0:08:00.880 --> 0:08:03.160 So if we need to change the direction of AI 0:08:03.160 --> 0:08:05.840 in a more socially productive direction, we may again need 0:08:05.880 --> 0:08:08.960 to go a little bit slower and more contemplatively about 0:08:09.200 --> 0:08:11.040 what is it that we want to do. What are 0:08:11.080 --> 0:08:14.400 the places where AI can have the biggest social impacts? 0:08:14.440 --> 0:08:15.880 And again the hype doesn't help that. 0:08:16.960 --> 0:08:18.960 Can we pull back? I'm curious, you know, we're talking 0:08:19.000 --> 0:08:23.480 about what investors expect from AI and how much companies 0:08:23.520 --> 0:08:27.400 are investing in it. What are we talking about? Is 0:08:27.440 --> 0:08:30.880 it just automation? What is AI in business today? And 0:08:30.880 --> 0:08:32.960 it seems like a simple question, Oh. 0:08:32.920 --> 0:08:36.080 My amazing question, because it's very difficult to get a 0:08:36.080 --> 0:08:39.920 straight answer to that. If you talk to some business leaders, 0:08:40.800 --> 0:08:44.480 especially in private, they'll say, of course, we're going to 0:08:44.480 --> 0:08:46.319 automate everything. That's what we want to do, and that's 0:08:46.320 --> 0:08:50.080 what AI is for Others, Sometimes the same people more 0:08:50.120 --> 0:08:52.000 publicly say no, no, we don't want to automate. We 0:08:52.000 --> 0:08:54.520 want to use AI for other things. If you look 0:08:54.559 --> 0:08:59.720 at Microsoft, which is the big partner to open Ai, 0:09:00.960 --> 0:09:04.200 they brand everything Core Pilot, and I think it's a 0:09:04.280 --> 0:09:08.000 genuine effort for them to try to find a way 0:09:08.080 --> 0:09:11.560 to use AI to help managers or to help decision makers. 0:09:11.840 --> 0:09:13.760 But on the other hand, if you look at what 0:09:13.840 --> 0:09:16.720 open AI's leaders say, they say, we're going to automate 0:09:16.760 --> 0:09:19.760 to stuff. So you know, it's a real confusion, and 0:09:19.840 --> 0:09:22.959 I think that confusion reflects exactly the same forces that 0:09:23.000 --> 0:09:25.599 we've been talking about Katya, which is that a we 0:09:25.800 --> 0:09:28.520 don't know where the future lies. There are many directions, 0:09:28.679 --> 0:09:31.720 the technology is malleable, we can make different choices, and 0:09:32.080 --> 0:09:34.000 right now we are very much at the beginning, so 0:09:34.040 --> 0:09:37.360 we can dream. I personally think, for example, that all 0:09:37.400 --> 0:09:41.400 of this talk of superintelligence and you know, AI doing 0:09:41.440 --> 0:09:44.560 everything that humans do better than humans within twenty years, etc. 0:09:45.200 --> 0:09:48.680 It's all misleading, but there's no way we can know 0:09:48.720 --> 0:09:50.800 whether it is or not, because we're talking about twenty 0:09:50.880 --> 0:09:55.640 years in a very fast changing field, and people can dream. 0:09:55.920 --> 0:09:59.760 Asia Centric is produced by Bloomberg Intelligence. We're more than 0:09:59.760 --> 0:10:03.679 fi five hundred experienced analysts and strategists work around the 0:10:03.720 --> 0:10:07.640 clock to bring you timely, world class research. Our coverage 0:10:07.679 --> 0:10:12.719 spans two hundred market indices, currencies, commodities, and industries, as 0:10:12.720 --> 0:10:16.000 well as over two thousand equities and credits. If you 0:10:16.080 --> 0:10:20.560 like what you hear, don't forget to subscribe and chairm Yeah, Darren, 0:10:20.559 --> 0:10:24.959 there's a lot of people like an ai to building railroads. 0:10:25.320 --> 0:10:28.120 There's a view that, like nineteenth century America, if you 0:10:28.120 --> 0:10:30.840 build the railroads, the trains will come. Sounds like you 0:10:30.840 --> 0:10:32.319 don't buy into that. 0:10:33.240 --> 0:10:36.760 Well, look, that's a very interesting debate. John perfect analogy 0:10:37.320 --> 0:10:43.080 because I actually personally think railways were critical for US development, 0:10:43.120 --> 0:10:46.400 and they were pretty pretty important for British development. But 0:10:46.840 --> 0:10:50.640 you know, some of the most respectable economic historians who 0:10:50.720 --> 0:10:53.920 work on this topic, for example, Robert Fogel, who want 0:10:54.080 --> 0:10:57.040 the Nobel Prize for his work on railways, made his 0:10:57.120 --> 0:11:00.240 reputation by arguing exactly the opposite. He said, Look, we 0:11:00.360 --> 0:11:05.040 had other ways of shipping goods between places and transporting 0:11:05.120 --> 0:11:09.559 other stuff and moving people around, such as canals, and 0:11:09.640 --> 0:11:12.080 if you look at railways, that improve things. But the 0:11:12.120 --> 0:11:16.120 marginal improvement wasn't all that big. So he argued that 0:11:16.240 --> 0:11:20.280 railways weren't such a big deal. Now, I think that 0:11:20.440 --> 0:11:26.280 railways really did change inputs into manufacturing and change the 0:11:26.320 --> 0:11:29.360 nature of technology. So I'm not agreeing with Fogel, but 0:11:29.400 --> 0:11:33.840 I'm just pointing out that even on something so important 0:11:33.920 --> 0:11:38.280 that happened one hundred and fifty years ago, there is disagreement. 0:11:38.600 --> 0:11:40.960 So of course there's going to be some disagreement on 0:11:41.120 --> 0:11:44.400 AI as to whether it's going to be completely transformative 0:11:44.440 --> 0:11:46.800 as a hub that changes everything, or it's going to 0:11:46.880 --> 0:11:48.559 enable us to do a few things that we were 0:11:48.559 --> 0:11:51.440 doing a little better look at search. Of course, I 0:11:51.480 --> 0:11:55.240 think AI can help search. You said just the ideal 0:11:55.360 --> 0:11:57.400 kind of thing, which is, you know, it's an information 0:11:57.559 --> 0:12:01.400 processing type of technology. So when I typed something into 0:12:01.440 --> 0:12:05.280 my browser, I can now get help from AI. But 0:12:05.480 --> 0:12:07.880 is that going to be transformative? If I get a 0:12:07.920 --> 0:12:10.760 little bit better suggestions about where to go for on vacation, 0:12:11.320 --> 0:12:15.240 or whether I get pointed to the right journal article, 0:12:15.920 --> 0:12:19.079 is that really going to change the nature of truism? 0:12:19.160 --> 0:12:21.040 Is that really going to change the nature of science? 0:12:21.160 --> 0:12:23.000 So those are the questions that I think we need 0:12:23.040 --> 0:12:23.640 to grapple with. 0:12:25.080 --> 0:12:29.080 And what would you call transformative change? Like you sell 0:12:29.080 --> 0:12:33.000 you some examples where you know, having this this app 0:12:33.400 --> 0:12:37.200 maybe suggest a different kind of cover letter or you know, 0:12:37.360 --> 0:12:40.800 suggests a different The different introduction to a paper is 0:12:40.880 --> 0:12:44.479 one thing, but transformative changes something else transformative. 0:12:44.520 --> 0:12:47.040 So let me give you two examples from my vantage 0:12:47.080 --> 0:12:50.760 point of transformative technology that I think most people will 0:12:50.800 --> 0:12:53.080 be familiar with, and then I'll tell you what my 0:12:53.679 --> 0:12:56.439 direction for AI that will be transformative. I think one 0:12:56.480 --> 0:13:02.319 transformative technology was Henry Ford's car factories, because they completely 0:13:02.320 --> 0:13:08.120 reorganized manufacturing and made it possible to first of all, 0:13:08.200 --> 0:13:11.720 use electric power in a much more efficient way, at 0:13:11.760 --> 0:13:15.120 the same time automating work and introducing very new range 0:13:15.240 --> 0:13:18.840 of new skills and new tasks. And that really spearheaded 0:13:18.960 --> 0:13:21.719 transformative growth in the auto industry and also in other 0:13:21.800 --> 0:13:25.199 manufacturing industries that copied it. Another one is the Internet. 0:13:25.520 --> 0:13:27.160 Sure there was a hypen, there was a boom for 0:13:27.200 --> 0:13:28.880 the Internet, but if you look at the Internet, it 0:13:28.920 --> 0:13:33.280 was a rather new way of bringing people together. It 0:13:33.360 --> 0:13:37.320 was at the same time an amazing new technology for 0:13:37.440 --> 0:13:42.040 sharing information. But it also enabled companies to often completely 0:13:42.080 --> 0:13:45.520 new services, create new platforms, So that I think, in 0:13:45.559 --> 0:13:48.760 my mind is transformative. So in the same way, I 0:13:48.800 --> 0:13:52.720 think AI as an information technology can be transformative if 0:13:52.720 --> 0:13:55.800 it does two things. One is just like the Internet. 0:13:56.320 --> 0:13:59.240 It helps us create new platforms to bring people together, 0:13:59.400 --> 0:14:04.360 to exchange formation, to exchange labor, to find better ways 0:14:04.480 --> 0:14:09.040 of actualizing their potential. Second, it actually helps us deal 0:14:09.120 --> 0:14:13.679 with fundamental shortages in the economy. So today we have 0:14:14.360 --> 0:14:19.640 a great shortage of skilled crafts people, electricians, plumbers, better 0:14:19.840 --> 0:14:24.000 manufacturing workers, more skilled teachers. I think these are all 0:14:24.080 --> 0:14:26.880 things that can be helped with AI. Because why Because 0:14:26.920 --> 0:14:30.640 AI can act as a tool that provides information to 0:14:30.680 --> 0:14:34.520 people that's relevant, real time, context specific. So you can 0:14:34.560 --> 0:14:37.000 be a better electrician with AI helping you. You can 0:14:37.040 --> 0:14:39.160 be a better teacher with AI helping you. And the 0:14:39.200 --> 0:14:41.920 critical thing here is that if you really want to 0:14:41.960 --> 0:14:45.520 realize it's not automation. You're not replacing the electrician or 0:14:45.560 --> 0:14:48.160 the teacher, but you're trying to enable them to do 0:14:48.360 --> 0:14:51.520 better in their task and to enable them to perform new, 0:14:51.600 --> 0:14:54.440 more sophisticated tasks. I think those are the paths for AI. 0:14:54.480 --> 0:14:57.480 And you see something in common between these two directions, 0:14:57.520 --> 0:15:00.800 I pointed out they're both about amplifying human capabilities, not 0:15:00.920 --> 0:15:05.160 sidelining humans, but augmenting humans, creating better tasks, new things 0:15:05.160 --> 0:15:07.200 for humans to do. 0:15:07.200 --> 0:15:09.120 Do you use AI? 0:15:09.520 --> 0:15:13.320 I used it. Yeah, I used early on chat Gipt, 0:15:14.080 --> 0:15:16.320 and I would be lying if I said the first 0:15:16.360 --> 0:15:19.800 time I used chat Gipt if I wasn't pretty impressed 0:15:19.840 --> 0:15:23.800 by the way that it converses with you. But soon 0:15:23.840 --> 0:15:27.800 I discovered that I could do most things that chat 0:15:27.840 --> 0:15:30.560 Gipt three point five or chat Gypt four was capable 0:15:30.560 --> 0:15:34.280 of in other ways that I felt more comfortable, like, 0:15:34.400 --> 0:15:38.840 for example, if I went to sources, I could check 0:15:38.880 --> 0:15:42.360 these things myself faster rather than you know, converse with AI, 0:15:42.560 --> 0:15:45.760 get some suggestions, and then follow the leads. So at 0:15:45.800 --> 0:15:48.440 the end of the day, I am not currently directly 0:15:48.520 --> 0:15:50.760 using AI, but I'm of course aware that in some 0:15:50.920 --> 0:15:54.120 other platforms that I use, AI is in the background. 0:15:54.200 --> 0:15:57.360 So I use Google for search, and I use an 0:15:57.400 --> 0:15:59.920 Apple phone and there is some AI in the background. 0:16:00.960 --> 0:16:03.520 They're an outside of chatjpt. There seems to be a 0:16:03.640 --> 0:16:08.680 dath of any killer AI apps. Have you seen anything 0:16:08.960 --> 0:16:12.360 getting excited over and how much patience do you think 0:16:12.840 --> 0:16:16.160 investors will have finding the new killer app for AI? 0:16:17.240 --> 0:16:19.040 No, I have not seen anything that I would say 0:16:19.080 --> 0:16:21.520 it's a great app, And I don't actually think that 0:16:21.640 --> 0:16:26.080 chat Gipt itself isn't that great. Again, I think the 0:16:26.120 --> 0:16:30.520 capability is there. If it was used differently, the architecture 0:16:30.600 --> 0:16:33.560 of AI and the vast amount of processing power and 0:16:33.640 --> 0:16:36.240 data that it has could be more useful. But I 0:16:36.280 --> 0:16:40.760 don't find it so impressive. If you know, chat gipt 0:16:40.880 --> 0:16:44.480 can write a shakespeareance on it, or can sideline teachers 0:16:44.560 --> 0:16:47.720 and pretend that you know, all the students need to 0:16:47.760 --> 0:16:50.440 do is actually go to chat gipt and ask questions. 0:16:50.440 --> 0:16:53.400 Those are not to me killer apps. So I think 0:16:53.800 --> 0:16:57.920 that's the next big task, next big challenge for the industry. 0:16:58.400 --> 0:17:02.120 Let's find something that's both useful for businesses and actually 0:17:02.200 --> 0:17:05.240 socially beneficial. That would be such a good challenge for 0:17:05.320 --> 0:17:05.880 the industry. 0:17:06.520 --> 0:17:09.159 Yeah, the social benefit of it's sort of the idea 0:17:09.200 --> 0:17:12.680 that I want AI to do my dishes and fold 0:17:12.720 --> 0:17:14.520 my laundry and not take my job away. 0:17:14.560 --> 0:17:15.320 That would be great too. 0:17:15.400 --> 0:17:21.040 Yes, I wonder if we could move outside of the US. 0:17:21.160 --> 0:17:23.359 I know you focused on the US and your paper, 0:17:23.680 --> 0:17:26.560 but you do have a global outlook, and I wonder 0:17:26.640 --> 0:17:30.480 if the impact of AI might be felt differently in 0:17:30.560 --> 0:17:34.760 different countries, depending on where they're starting from and the 0:17:34.800 --> 0:17:37.919 penetration of technology. What are your thoughts? 0:17:38.520 --> 0:17:40.400 Absolutely, I think there are a couple of issues there 0:17:40.400 --> 0:17:43.760 we have to watch out for. The First one is 0:17:43.920 --> 0:17:46.720 that what happens in the US and other leading AI 0:17:46.840 --> 0:17:49.400 powers is going to influence what happens in the rest 0:17:49.400 --> 0:17:51.000 of the world. The rest of the world is not 0:17:51.040 --> 0:17:55.040 ready for AI. They're politicians, their thought leaders are not 0:17:55.119 --> 0:17:58.480 focusing on AI, so by and large, AI is going 0:17:58.520 --> 0:18:01.760 to be something that's done to them rather than something 0:18:01.800 --> 0:18:04.680 in which they have agency. Still, it can turn out 0:18:04.680 --> 0:18:07.159 to be good for them because if it goes in 0:18:07.200 --> 0:18:10.879 a direction that really makes you know, a worker's more productive, 0:18:11.320 --> 0:18:14.600 then it will spread to say India and Indonesia, and 0:18:14.640 --> 0:18:18.720 it's going to make workers productive there. But I think 0:18:19.200 --> 0:18:22.600 if it goes in an automation direction, things are reversed. 0:18:23.720 --> 0:18:32.560 Many developing countries critically depend on their human resources semi skilled, cheap, active, 0:18:32.640 --> 0:18:36.680 flexible labor. If AI starts replacing things that this word 0:18:36.760 --> 0:18:40.640 labor does, it's going to have, you know, disruptive effects. 0:18:41.200 --> 0:18:43.560 If AI changes the global division of labor is going 0:18:43.640 --> 0:18:47.879 to have disruptive effects. And moreover, many new technologies create 0:18:48.520 --> 0:18:52.600 a pattern of winner takes all, meaning that some countries 0:18:52.640 --> 0:18:55.840 that are first may leave the rest behind. And that's 0:18:55.880 --> 0:18:59.359 another concern and then the final wildcard, well, which is 0:18:59.520 --> 0:19:02.639 we're already seeing it realize in the developing world, is 0:19:02.680 --> 0:19:06.119 that AI is also a phenomenal technology for surveillance. You know, 0:19:06.560 --> 0:19:09.800 a lot of the AI energy in China is going 0:19:09.840 --> 0:19:13.800 to surveillance activities. That's been very well documented. And also 0:19:14.440 --> 0:19:18.199 many of the surveillance related technologies, ranging from facial recognition, 0:19:18.480 --> 0:19:22.520 things that can sweep the Internet, et cetera, are just 0:19:22.600 --> 0:19:25.359 are not staying just in China. They're being exported to 0:19:26.000 --> 0:19:28.840 dozens of countries around the world, and most of them authoritarian, 0:19:29.160 --> 0:19:33.160 and they're using these technologies with great enthusiasm. And it's 0:19:33.200 --> 0:19:36.280 not just authoritarian countries. I mean, you know, we can 0:19:36.359 --> 0:19:39.800 debate where US should be on the spectrum, but US 0:19:39.840 --> 0:19:44.360 invests a lot in AI related to national security apparatus, 0:19:44.400 --> 0:19:48.800 and you know, privacy from the government has become less 0:19:48.840 --> 0:19:51.680 secure in the United States and in some other Western 0:19:51.760 --> 0:19:54.440 countries as well. So there are going to be other 0:19:54.840 --> 0:19:58.120 politically first order implications of AI as well, and those 0:19:58.200 --> 0:20:00.720 might matter more for the developing world. 0:20:01.600 --> 0:20:03.600 Okay, but it sounds like a reading between the lines, 0:20:03.680 --> 0:20:07.760 sounds like developed, rich countries would have a bigger benefit 0:20:07.800 --> 0:20:09.480 from AI than developing countries. 0:20:10.840 --> 0:20:14.040 I think right now that would be my guess. But 0:20:14.280 --> 0:20:19.080 if I had one recommendation to developing country leaders, I 0:20:19.080 --> 0:20:21.879 would say, this is the right time to form a 0:20:21.880 --> 0:20:28.159 new consortium, you know, especially with leading countries such as India, Indonesia, Brazil, Turkey, 0:20:28.240 --> 0:20:32.119 Mexico playing a leading role which is about AI, and 0:20:32.600 --> 0:20:35.359 there needs to be a perspective from the developing world 0:20:35.840 --> 0:20:38.199 and a voice from developing world, because you know, we 0:20:38.320 --> 0:20:42.640 hear from businesses in the US, we hear from increasingly now, 0:20:42.640 --> 0:20:44.600 which is a good thing, from some worker groups in 0:20:44.600 --> 0:20:46.600 the US or in the Western world, but we don't 0:20:46.680 --> 0:20:51.000 hear the perspective, the interests, the sort of the things 0:20:51.000 --> 0:20:54.960 that the developing emerging world once and is concerned about AI. 0:20:56.480 --> 0:21:00.800 I wonder if if we see that happening yet, is 0:21:00.840 --> 0:21:01.440 that happening? 0:21:01.680 --> 0:21:04.720 No? No, I mean, if you go to many developing countries, 0:21:04.800 --> 0:21:08.919 there is interest within the public and some businesses, but 0:21:09.000 --> 0:21:12.800 there's no policy. Policy makers are not focused on that. 0:21:12.960 --> 0:21:16.439 And unfortunately, you know, we live in a polarized world 0:21:16.880 --> 0:21:19.600 that's divided, not just within countries. It's not just within 0:21:19.640 --> 0:21:22.200 the US or within France that there are big divisions. 0:21:22.960 --> 0:21:25.840 There are big divisions between countries as well, and so 0:21:25.920 --> 0:21:30.160 it's becoming increasingly difficult for different country leaders to work 0:21:30.200 --> 0:21:31.120 together with each other. 0:21:32.480 --> 0:21:35.720 Yeah, there's you mentioned the US in China earlier. It 0:21:35.840 --> 0:21:40.400 seems like there's this new I mean, not a new analogy, 0:21:40.400 --> 0:21:42.959 but you know, the space race, right except take it 0:21:43.000 --> 0:21:47.760 to AI between China and the US. Is there a 0:21:48.040 --> 0:21:51.120 sense that you know, given the minimal economic impact that 0:21:51.359 --> 0:21:53.360 you've computed so far, I mean, is there a risk 0:21:53.480 --> 0:21:56.040 that they're spending on the wrong things. 0:21:56.680 --> 0:21:58.760 Well, they're spending on the wrong things, and it's maybe 0:21:58.800 --> 0:22:02.399 actually fueling the wrong narry. You know, two arguments that 0:22:02.440 --> 0:22:07.240 are very common in the US are First, we cannot 0:22:07.280 --> 0:22:09.720 regulate AI because if we do, China is going to 0:22:09.720 --> 0:22:13.000 take the leadership. And two, we have to be tougher 0:22:13.119 --> 0:22:15.639 on China otherwise they will become the AI leader, and 0:22:15.680 --> 0:22:17.440 we want to be the AI leader. And whoever becomes 0:22:17.480 --> 0:22:20.400 the AI leader controls the world. Now, what you see 0:22:20.440 --> 0:22:23.080 in both of these narratives is not just you know, 0:22:23.160 --> 0:22:28.680 hyping up AI, but also deepening the tensions between US 0:22:28.680 --> 0:22:31.160 and China, which is the last thing we want, Which 0:22:31.200 --> 0:22:33.000 is the last thing we want in a world without AI. 0:22:33.080 --> 0:22:35.240 And if we're gonna have AI, we want it even 0:22:35.359 --> 0:22:38.159 less because why because AI is a global technology, So 0:22:38.240 --> 0:22:40.520 if you're gonna regulate it, you need US and China 0:22:40.560 --> 0:22:41.240 to work together. 0:22:43.000 --> 0:22:45.479 So if you're an investor right now, like we've been 0:22:45.520 --> 0:22:48.760 talking about some pretty big themes, if you're an investor 0:22:49.840 --> 0:22:53.000 and you are putting money into a company on a 0:22:53.160 --> 0:22:57.160 bet on an AI bet, are there risks to that? 0:22:57.560 --> 0:22:58.719 Should they be doing that? 0:23:00.160 --> 0:23:02.639 Investor? And of course there are risks, and it's a 0:23:02.720 --> 0:23:05.720 very difficult thing to do because even imagine everything I 0:23:05.760 --> 0:23:09.120 said here is right, and it's even more severe than 0:23:09.160 --> 0:23:13.439 I am saying. But as long as investors keep on 0:23:13.560 --> 0:23:16.280 investing in AI and Vidia, stocks are going to do 0:23:16.359 --> 0:23:19.920