WEBVTT - Tyler Cowen on Why AI Hasn't Changed the World Yet 0:00:02.720 --> 0:00:16.720 Bloomberg Audio Studios, Podcasts, radio News. 0:00:18.200 --> 0:00:21.880 Hello and welcome to another episode of the Odd Lots podcast. 0:00:21.960 --> 0:00:24.400 I'm Jolle Wisenthal and I'm Tracy Alloway. 0:00:25.000 --> 0:00:27.880 You know, not a novel observation here, but I do 0:00:27.960 --> 0:00:31.440 think that if, like in early twenty twenty two or 0:00:31.560 --> 0:00:34.960 late twenty twenty one, you had someone had revealed to 0:00:35.040 --> 0:00:37.400 you all of the amazing things that we could do 0:00:37.440 --> 0:00:41.239 with AI these days, I think you would have expected 0:00:41.280 --> 0:00:44.600 that either the broader economy or society would be more 0:00:44.640 --> 0:00:48.640 different than it has been. Like I am large, you know, 0:00:49.320 --> 0:00:52.640 I think most people's jobs are done roughly the same way. 0:00:53.400 --> 0:00:56.000 Sort of society still seems to operate, although a little 0:00:56.040 --> 0:00:58.640 maybe a little bit worse every day. I don't know, 0:00:58.800 --> 0:01:01.400 But like I just don't think like it's mind blowing 0:01:01.400 --> 0:01:04.000 technology and get by and large, like it hasn't had 0:01:04.040 --> 0:01:07.040 the economic disruption that I think many people would have guessed. 0:01:07.080 --> 0:01:09.280 Maybe I'm cynical on the subject, but I always say, 0:01:09.280 --> 0:01:13.360 never underestimate the human capacity for stasis, I guess, and 0:01:13.440 --> 0:01:16.440 making things much more difficult than they actually need to be, 0:01:17.000 --> 0:01:20.840 and putting up you know, bureaucratic barriers, regulatory barriers and 0:01:20.880 --> 0:01:23.720 things like that. So it's not it's not that surprising 0:01:23.760 --> 0:01:25.360 to me, but it is true. You have a lot 0:01:25.400 --> 0:01:27.720 of economists out there who think there was going to 0:01:27.760 --> 0:01:31.080 be this massive productivity boost, right well, it's certainly. 0:01:30.720 --> 0:01:33.039 Tech people like I don't know, like economists. You know, 0:01:33.120 --> 0:01:35.200 they're like, oh, there were some we're going to have 0:01:35.240 --> 0:01:38.279 a productivity boom. We're raising our estimates from two percent 0:01:38.319 --> 0:01:41.520 too and a quarter percent, and then you have all 0:01:41.560 --> 0:01:43.959 things relative to and then you know, we talked to 0:01:44.000 --> 0:01:45.960 Kathy Wood one time, and I think she what did 0:01:46.000 --> 0:01:49.760 she predict twenty percent real GDP growth for twenty years? 0:01:49.880 --> 0:01:54.400 Or you certainly have these people in Silicon Valley deflationary boom, 0:01:54.480 --> 0:01:58.880 post scarcity any any minute, right now, some real gaps 0:01:58.960 --> 0:02:01.480 I would say, between how lot of economists think about 0:02:01.480 --> 0:02:03.760 some of these things and the numbers, then economists would 0:02:03.800 --> 0:02:07.560 be comfortable using versus maybe literally. 0:02:07.360 --> 0:02:10.600 Everyone else like Kathy would certainly many others. Yeah, you 0:02:10.639 --> 0:02:14.280 know what I think about is how different would your 0:02:14.320 --> 0:02:18.680 career in my career have been had AI existed in 0:02:18.800 --> 0:02:21.200 like two thousand and eight, two thousand and six when 0:02:21.240 --> 0:02:22.799 we were starting to blog. 0:02:22.880 --> 0:02:25.760 Basically, it's a really good question, and I don't know 0:02:25.919 --> 0:02:28.480 the answer because part of me thinks, well, you know, 0:02:28.560 --> 0:02:31.240 that whole path that was like defined my career like 0:02:31.280 --> 0:02:34.520 would not have been there, would not have existed or no. Maybe, 0:02:34.560 --> 0:02:37.320 But on the other hand, maybe in two thousand and 0:02:37.320 --> 0:02:40.640 six I would have been like just like I was 0:02:40.680 --> 0:02:45.440 super early into blogging, super early into experimenting with AI news, 0:02:45.480 --> 0:02:47.520 and it would look different, but maybe I would have 0:02:48.000 --> 0:02:49.079 rowed some different. 0:02:48.880 --> 0:02:49.720 Different but the same. 0:02:49.840 --> 0:02:51.799 Yeah, different but the same, you know, Like I think 0:02:51.800 --> 0:02:53.200 it's a it's pretty hard to tell. 0:02:53.440 --> 0:02:56.360 One thing I know is like back then we were 0:02:56.400 --> 0:03:00.240 writing to optimize Google results, right, that was basic or know, 0:03:00.360 --> 0:03:02.919 like I know social media was disseminating stuff as well, 0:03:02.919 --> 0:03:05.400 but like Google was still the main platform for a 0:03:05.400 --> 0:03:08.280 lot of stuff. I'm sure if we were doing it now, 0:03:08.400 --> 0:03:12.840 we would be optimizing for chat, GPT or perplexity or 0:03:12.840 --> 0:03:15.839 someone like that to actually pick up the content. I mean, 0:03:16.000 --> 0:03:17.240 I think that's the audience. 0:03:17.480 --> 0:03:19.600 Yeah it is. And I think like any especially any 0:03:19.680 --> 0:03:22.320 like commercial publisher, right that's trying to do a scale. 0:03:22.360 --> 0:03:26.040 I think like a sort of niche voice expert could 0:03:26.080 --> 0:03:28.240 still have like their audience that comes to them directly, 0:03:28.280 --> 0:03:31.320 of course, but like it's scale scale for sure every 0:03:31.360 --> 0:03:34.400 publisher is trying to figure this out. Anyway, We really 0:03:34.400 --> 0:03:36.839 do have the perfect guest, someone we should have had 0:03:36.840 --> 0:03:39.800 on the podcast years and years ago. It's almost surprising 0:03:39.880 --> 0:03:41.960 that it's just the first time we've had him on. 0:03:42.160 --> 0:03:45.080 It is crazy, It is crazy, but it is someone 0:03:45.160 --> 0:03:47.640 who I think is like at the intersection of everything 0:03:47.680 --> 0:03:51.080 that we're talking about. Because he's an economist. He knows 0:03:51.160 --> 0:03:53.440 the tech really well. The tech people know him, the 0:03:53.480 --> 0:03:55.520 tech people love him. He may even be one of 0:03:55.560 --> 0:03:59.160 the preferred economists for this. I feel he's like the 0:03:59.200 --> 0:04:01.280 economist that all people in the AI world want to 0:04:01.280 --> 0:04:04.880 get his take. A long time blogger, someone I, you know, 0:04:04.960 --> 0:04:06.920 like both of us sort of gave up on mike blogging. 0:04:06.960 --> 0:04:08.840 Although we have our newsletter. It's like close enough, but 0:04:08.880 --> 0:04:11.440 it's different than blogging. Someone who's stuck with the medium 0:04:11.480 --> 0:04:14.040 for a long time. One of the true original econ 0:04:14.120 --> 0:04:17.000 bloggers that I've been reading for over twenty years. We're 0:04:17.040 --> 0:04:19.239 going to be speaking with Tyler Cowen. Here's the host 0:04:19.240 --> 0:04:22.760 of the Conversations with Tyler podcast. He also, obviously is 0:04:22.760 --> 0:04:26.360 one of the two bloggers that the famous Marginal Revolution blog. 0:04:26.880 --> 0:04:33.279 Economics professor gmu appreciator of ethnic foods all around the 0:04:33.720 --> 0:04:38.320 DC Northern Virginia, Maryland area. Someone known on the internet 0:04:38.360 --> 0:04:40.200 for a long time, Tyler, thank you so much for 0:04:40.520 --> 0:04:43.440 coming on odd Laws. Really thrilled to have you here. 0:04:44.040 --> 0:04:45.400 Hello, happy to be here. 0:04:45.800 --> 0:04:49.560 Amazing. What do you think about my initial assessment? How 0:04:49.600 --> 0:04:51.719 fair do you think it is that, Like, had you 0:04:51.880 --> 0:04:55.400 known in twenty twenty one how powerful these tools would be, 0:04:55.880 --> 0:04:59.200 maybe we'd be a bit surprised that by and large, 0:04:59.279 --> 0:05:01.560 like busines, and this seems to more or less run 0:05:01.600 --> 0:05:03.479 the same, I'm. 0:05:03.320 --> 0:05:06.040 Not surprised at all. So what I see right now 0:05:06.160 --> 0:05:09.080 is people using AI as an add on to their 0:05:09.120 --> 0:05:12.240 pre existing work routines. Oh, you need to write a memo, 0:05:12.480 --> 0:05:14.520 you ask AI how to do it. You need to 0:05:14.520 --> 0:05:17.000 write a column, you ask AI to proofread it or 0:05:17.040 --> 0:05:20.200 fact check it. And that works great, But those are 0:05:20.279 --> 0:05:23.239 marginal gains. What we really need to see a major 0:05:23.320 --> 0:05:27.320 impact is new organizations built around AI, and those will 0:05:27.360 --> 0:05:30.839 be startups. They will come only slowly. It will take 0:05:31.400 --> 0:05:34.720 twenty or more years before they really transform the economy. 0:05:35.440 --> 0:05:37.559 And in the meantime, it's a whole bunch of add ons, 0:05:37.600 --> 0:05:40.159 which are fun and fine, but that's why I think 0:05:40.160 --> 0:05:40.640 it's slow. 0:05:40.960 --> 0:05:43.560 What can you tell us about the history of technology 0:05:44.080 --> 0:05:49.040 such that legacy organizations that existed prior to the invention 0:05:49.440 --> 0:05:54.040 of maybe potentially revolutionary new technology have a hard time 0:05:54.760 --> 0:05:57.320 massively changing their workflows. 0:05:58.360 --> 0:06:01.640 Well, you can take even very simple examples. So Toyota 0:06:01.680 --> 0:06:05.560 starts competing with General Motors in the nineteen seventies, General 0:06:05.600 --> 0:06:08.919 Motors is paralyzed. It cannot really come back and adopt 0:06:08.920 --> 0:06:11.800 the new and superior Toyota methods, and those are not 0:06:12.120 --> 0:06:15.960 really that different compared to say AI. So there's just plenty, 0:06:16.240 --> 0:06:20.320 plenty examples. Old mainstream media could not cope very well 0:06:20.360 --> 0:06:23.080 with the Internet. There are exceptions like The New York Times. 0:06:23.800 --> 0:06:27.600 Odd Lot's podcast would be another thank you, But it's 0:06:27.640 --> 0:06:30.680 the norm. So we're seeing it again with AI. And 0:06:30.720 --> 0:06:33.400 again you need a complete turnover of who and what 0:06:33.560 --> 0:06:35.600 is doing business for it to really matter on a 0:06:35.640 --> 0:06:36.200 big scale. 0:06:37.560 --> 0:06:40.520 If you need that complete turnover, and you need some 0:06:40.640 --> 0:06:43.920 time for AI to become fully embedded in a business 0:06:43.960 --> 0:06:47.280 model or for a business model to form around AI, 0:06:48.120 --> 0:06:50.640 what industry or what part of the economy would you 0:06:50.720 --> 0:06:53.600 expect it, I guess to show up first in that 0:06:53.680 --> 0:06:55.280 sort of revolutionary way. 0:06:56.680 --> 0:06:59.160 Well, we have obvious data on this, and it's programming. 0:06:59.520 --> 0:07:02.640 You will hear people who do programming claim that say 0:07:02.680 --> 0:07:05.320 eighty percent of the work is now done by AIS. 0:07:05.520 --> 0:07:08.520 I suspect that's an overstatement, but there's no doubt at 0:07:08.560 --> 0:07:11.680 all that there's simply a lot of programming already done 0:07:11.680 --> 0:07:16.360 by AIS. When you have low fixed costs, a competitive sector, 0:07:16.640 --> 0:07:21.320 immediate feedback, you know, the revenue has to flow programming. 0:07:21.400 --> 0:07:24.760 And also New York City finance, there's been quantston finance 0:07:24.840 --> 0:07:28.400 for a long time. Those quants are now you could say, 0:07:28.440 --> 0:07:30.880 more AI equipped than they used to be, and those 0:07:30.960 --> 0:07:33.040 areas already are being revolutionized. 0:07:33.560 --> 0:07:37.080 I've heard them and got some pitches to do episodes, 0:07:37.120 --> 0:07:39.240 which we should do at some point. But I've heard 0:07:39.280 --> 0:07:43.600 about like some law firms that are being like new 0:07:43.720 --> 0:07:46.720 AI law firms where there are lawyers, et cetera. But 0:07:46.800 --> 0:07:49.920 from the very ground up, the idea is perhaps there 0:07:50.000 --> 0:07:54.280 is some way to just get way more productivity if 0:07:54.320 --> 0:07:57.400 they start from the very beginning with some combination of 0:07:57.680 --> 0:08:00.840 lawyers plus AI models. It seems like that could be 0:08:00.880 --> 0:08:03.440 the kind of thing where maybe the legacy law firms 0:08:03.640 --> 0:08:07.320 are seeing some productivity gains from AI. There's probably some 0:08:07.360 --> 0:08:10.280 evidence you could find but that a new one with 0:08:10.320 --> 0:08:14.560 a totally different approach could deliver that big productivity boost 0:08:14.560 --> 0:08:16.280 that actually ends up changing the industry. 0:08:17.160 --> 0:08:20.000 It's already the case, say, mid tier associates are much 0:08:20.080 --> 0:08:23.360 less needed. But there's one big problem with law in particular, 0:08:23.840 --> 0:08:27.080 and that is the way large language models work. Now 0:08:27.160 --> 0:08:30.040 you have to send your queries somewhere else. You can't 0:08:30.080 --> 0:08:32.440 just own and control and hold the whole thing on 0:08:32.480 --> 0:08:34.960 your hard drive. Now. I think within a few years 0:08:35.040 --> 0:08:38.600 time that will be very different. But until then, major 0:08:38.720 --> 0:08:42.480 law firms are extremely skittish about just typing in their 0:08:42.559 --> 0:08:45.240 questions and sending it, you know, to San Francisco. I 0:08:45.280 --> 0:08:47.480 don't actually think there's a risk, but when you think 0:08:47.480 --> 0:08:50.680 about how fiduciary responsibility works, they just don't want to 0:08:50.720 --> 0:08:51.000 do it. 0:09:07.559 --> 0:09:11.280 Talk more about I guess privacy concerns and regulation, because 0:09:11.320 --> 0:09:13.880 this seems to be an area where if you are 0:09:13.920 --> 0:09:16.280 in a heavily regulated industry, or if you're just in 0:09:16.320 --> 0:09:18.680 an industry that tends to be full of paranoid people 0:09:18.880 --> 0:09:22.120 like lawyers, it does seem like there's going to be 0:09:22.120 --> 0:09:25.800 a natural tendency to be very very cautious. When it 0:09:25.840 --> 0:09:28.000 comes to sharing data with AI, you're going to be 0:09:28.040 --> 0:09:30.760 worried about actual data ownership the queries that you're sending 0:09:30.840 --> 0:09:34.240 to San Francisco, as you say, are those industries just 0:09:34.320 --> 0:09:36.240 inevitably going to be slow to adapt. 0:09:37.880 --> 0:09:40.880 They'll be slow to adapt again until the point where 0:09:40.920 --> 0:09:44.400 they just control their own model and they hold it 0:09:44.440 --> 0:09:46.640 within the firm and no one else really can get 0:09:46.679 --> 0:09:52.160 at it. So what you need is cheaper models where 0:09:52.280 --> 0:09:54.840 a law firm can afford to have its own model. 0:09:54.880 --> 0:09:56.920 And I think that's a few years away. It's not 0:09:57.000 --> 0:09:59.400 a very long time away, but it won't come in 0:09:59.440 --> 0:10:02.960 six months. Sam Altman I just did a podcast with 0:10:03.080 --> 0:10:07.240 him and he said a privacy problem is AI queries 0:10:07.360 --> 0:10:09.920 or subject to subpoena, and he thinks they should have 0:10:09.960 --> 0:10:13.200 as much protection as say, your conversation with your lawyer 0:10:13.320 --> 0:10:15.920 or your doctor or your therapist. I think that's a 0:10:15.920 --> 0:10:19.800 good idea, but that hasn't happened yet, and until it happens, 0:10:19.920 --> 0:10:21.880 or you get the whole thing on your own hard drive, 0:10:22.280 --> 0:10:24.560 progress in law is going to be slow. But once 0:10:24.600 --> 0:10:27.480 the progress comes, that's one of the areas where I 0:10:27.480 --> 0:10:30.600 think AI has the most promise it's just very very 0:10:30.640 --> 0:10:34.040 good at mastering a large corpus of text and organizing 0:10:34.040 --> 0:10:34.480 it for you. 0:10:35.240 --> 0:10:37.880 It is interesting, isn't it. Where it's like, Okay, if 0:10:37.880 --> 0:10:40.359 you are my lawyer, you and I could have a 0:10:40.400 --> 0:10:44.280 conversation and it would be not it would not be uh. 0:10:44.240 --> 0:10:44.960 It'd be privileged. 0:10:45.040 --> 0:10:48.960 Yeah, it would be privileged. Or you know, one can 0:10:49.000 --> 0:10:52.640 have conversations with anyone, and as long as you're doing 0:10:52.679 --> 0:10:55.160 it on the phone, or if you're doing it person 0:10:55.200 --> 0:10:59.079 to person, it's much more difficult to get that, you know, 0:10:59.200 --> 0:11:02.720 in discover or I always think about this with you know, 0:11:02.800 --> 0:11:06.360 public public records of employees that sometimes you can get 0:11:06.400 --> 0:11:09.240 their emails through Floyers, et cetera, but you can't get 0:11:09.240 --> 0:11:13.040 the cons content of conversations. It does feel I hadn't 0:11:13.040 --> 0:11:16.520 really thought about this dynamic though, that when you're using 0:11:16.559 --> 0:11:19.200 an AI it is sort of like a conversation and 0:11:19.280 --> 0:11:22.880 yet sort of from an evidentiary basis, it would be 0:11:23.000 --> 0:11:24.080 much more like an email. 0:11:25.200 --> 0:11:27.320 I think when it comes to medical issues, there are 0:11:27.360 --> 0:11:30.560 many more people willing to share their data. Not everyone 0:11:30.640 --> 0:11:34.040 by any means. Some medical conditions are secretor people just 0:11:34.080 --> 0:11:36.960 don't want others to know. But I see many, many 0:11:37.000 --> 0:11:39.240 people I know typing in all kinds of things about 0:11:39.240 --> 0:11:42.480 their medical history to say GPT five and getting what 0:11:42.559 --> 0:11:45.360 are on the whole very good answers. It's like medical 0:11:45.360 --> 0:11:49.199 diagnosis for free spreading now to the whole world. A 0:11:49.240 --> 0:11:51.840 lot of countries where people just don't even have access 0:11:51.880 --> 0:11:54.920 to good doctors at all, and I think that will 0:11:55.080 --> 0:11:58.400 be important more quickly than the law innovations. 0:11:59.600 --> 0:12:01.719 Just as a thought experiment, what does all of this 0:12:01.880 --> 0:12:04.120 mean for insurers? Because I kind of think, you know, 0:12:04.160 --> 0:12:05.920 I think about a bunch of people typing in their 0:12:05.960 --> 0:12:09.600 medical information. I think about basically the explosion in data 0:12:09.640 --> 0:12:11.959 that we have nowadays, and it seems like the insurance 0:12:12.000 --> 0:12:15.920 industry would be one place that would really benefit from 0:12:15.920 --> 0:12:19.480 all this trove of additional information if they could access 0:12:19.520 --> 0:12:21.120 it well. 0:12:21.120 --> 0:12:23.480 This is one of my worries about AI in general. 0:12:23.679 --> 0:12:27.240 I'm quite positive on what's happening, But as insurers get 0:12:27.280 --> 0:12:30.720 better and better information on their customers, this is just 0:12:30.760 --> 0:12:32.880 through big data more generally, it doesn't have to be 0:12:32.960 --> 0:12:36.520 current large language models. They know exactly how to write 0:12:36.520 --> 0:12:38.920 the risk and how to price the premium, and in 0:12:38.960 --> 0:12:41.880 a sense for the buyer, it's not insurance anymore. So, 0:12:41.920 --> 0:12:43.760 if we know your house is going to burn down 0:12:43.800 --> 0:12:46.200 with high probability and you have to pay the super 0:12:46.280 --> 0:12:49.240 high premium, you don't really have the benefits of the insurance. 0:12:49.600 --> 0:12:54.000 So some insurance markets might unravel if through big data 0:12:54.480 --> 0:12:57.120 the insurers learn too much about what's likely to happen. 0:12:57.320 --> 0:13:02.760 Economists seem to be very consistent about the effects of 0:13:02.800 --> 0:13:06.560 technology on labor demand, which is that in the end 0:13:06.600 --> 0:13:09.240 it washes out right, some people, Okay, there's a disruption, 0:13:09.400 --> 0:13:11.520 but I'm gonna save money because I use the I. 0:13:11.720 --> 0:13:13.679 But that means I have more spending power, and then 0:13:13.720 --> 0:13:16.199 I'm gonna buy something else that I wouldn't have bought 0:13:16.280 --> 0:13:20.120 had it not been for paying wages, and then that'll 0:13:20.160 --> 0:13:22.880 create demand for labor elsewhere. And so in the end, 0:13:22.880 --> 0:13:27.160 the idea that you could really have tech driven unemployment 0:13:27.240 --> 0:13:31.600 at scale that is not transitory or not temporary is hard. 0:13:31.760 --> 0:13:35.680 A lot of economists seem to be intuitively skeptical of 0:13:35.720 --> 0:13:39.080 this idea. Whereas you have people in the ifields, fifty 0:13:39.080 --> 0:13:40.880 percent of the people aren't gonna have jobs we need 0:13:41.000 --> 0:13:44.480 ubi otherwise there's gonna be a permanent underclass. Could there 0:13:44.520 --> 0:13:48.560 be something different about AI such that it doesn't have 0:13:48.720 --> 0:13:51.760 the same labor market effects that past technologies have had. 0:13:52.720 --> 0:13:56.000 I would say, you understand me. Well, So the energy sector, 0:13:56.080 --> 0:13:59.000 there's going to be a lot of new jobs taking 0:13:59.040 --> 0:14:03.319 care of older people. I think as AI produces more potential, 0:14:03.360 --> 0:14:06.640 medical innovations will need to test them, So the biomedical 0:14:06.679 --> 0:14:09.960 sector testing clinical trials, there'll be a lot new jobs. 0:14:10.760 --> 0:14:14.000 I'm not worried about mass unemployment, and most economists are not, 0:14:14.440 --> 0:14:16.640 and I agree with their perspectives, which I think you 0:14:16.679 --> 0:14:17.800 outlined pretty clearly. 0:14:18.400 --> 0:14:21.000 What do you say, though, because I have a feeling 0:14:21.120 --> 0:14:23.920 that when you're out in San Francisco, they don't see 0:14:23.960 --> 0:14:27.160 it that way, and they talk a lot, and some 0:14:27.240 --> 0:14:29.600 of them are more in you know, UBI talk and 0:14:29.880 --> 0:14:32.560 permanent underclass talk and all of this stuff. Do they 0:14:32.640 --> 0:14:35.080 see your perspective when you make this case? What do 0:14:35.120 --> 0:14:37.360 they say or what are they What are a lot 0:14:37.400 --> 0:14:39.360 of them seem to be missing about the logic that 0:14:39.400 --> 0:14:40.400 you spell out. 0:14:40.880 --> 0:14:43.200 Well, I think more and more they're coming around to 0:14:43.240 --> 0:14:46.560 the economist's point of view. So Andre Carpathi, who was 0:14:46.600 --> 0:14:49.720 at open AI, you know, in the most important years, 0:14:50.080 --> 0:14:53.400 he just did a podcast saying he thinks adjustment will 0:14:53.400 --> 0:14:56.120 be slow. Things will be fine, will grow at two 0:14:56.200 --> 0:14:59.600 points something percent. There won't be mass unemployment, and you 0:14:59.640 --> 0:15:02.200 wouldn't I have heard that, say two years ago. But 0:15:02.240 --> 0:15:04.560 I think as people see the models rolling out, and 0:15:04.600 --> 0:15:07.160 as you mentioned well, the real world impact it is 0:15:07.200 --> 0:15:12.160 stretched out in time, right, it's not all immediate. Earlier 0:15:12.200 --> 0:15:14.280 on people had more the sense that AI was a 0:15:14.360 --> 0:15:16.720 kind of god box that you just talked to it 0:15:16.840 --> 0:15:19.800 and it can magically do anything and convert that into 0:15:19.840 --> 0:15:22.360 results in the real world. But if you think about 0:15:22.360 --> 0:15:25.480 your actual job, even if it's a highly intellectual job, 0:15:25.800 --> 0:15:28.560 so much of what you do is the interaction between 0:15:28.560 --> 0:15:32.080 your intellect, your physical presence, your interactions with others, your travel, 0:15:32.640 --> 0:15:36.320 many other things. And until we get to some far, 0:15:36.480 --> 0:15:40.120 far off world where the robots are perfect copies of you, 0:15:40.360 --> 0:15:43.800 which I don't think ever will come, jobs will be fine, 0:15:43.840 --> 0:15:46.840 but they will change a lot, and I'm actually worried 0:15:46.840 --> 0:15:50.240 about who will be the biggest losers. I think poor 0:15:50.280 --> 0:15:54.040 people will do great, the very wealthy will do well, 0:15:54.480 --> 0:15:56.840 but people who are sort of upper upper middle class 0:15:57.400 --> 0:16:00.800 will find this automatic ticket to a law or consulting 0:16:00.920 --> 0:16:03.920 job that assured they would be upper upper middle class 0:16:03.960 --> 0:16:06.120 for the rest of their lives. I think a lot 0:16:06.160 --> 0:16:07.600 of that is going away already. 0:16:08.400 --> 0:16:11.480 I think also no one would have expected plastic surgery, 0:16:11.520 --> 0:16:13.800 I guess, to be a beneficiary of the AI revolution. 0:16:13.960 --> 0:16:15.800 But if you think that what's going to matter in 0:16:15.840 --> 0:16:19.720 the future is like your personal presence and your network 0:16:19.760 --> 0:16:21.960 of social contacts, then I guess we should all be 0:16:22.040 --> 0:16:29.160 working on on our looks Maxing, Yeah, absolutely right, charisma. Okay, noted, 0:16:29.320 --> 0:16:32.480 everyone work on their respective charisma. One thing I was 0:16:32.520 --> 0:16:36.160 wondering is the impact of AI on public finances. I 0:16:36.160 --> 0:16:39.200 am not very good at tax policy. Joe knows this 0:16:39.240 --> 0:16:43.840 because I've complained about taxes to him repeatedly. But I'm thinking, 0:16:46.160 --> 0:16:49.880 but if you're thinking about where the value add of 0:16:49.960 --> 0:16:52.360 AI actually shows up in the economy, So you know, 0:16:52.440 --> 0:16:55.240 presumably you got a productivity boost. We're not entirely sure 0:16:55.240 --> 0:16:57.280 how much that's going to be. But where does that 0:16:57.320 --> 0:17:01.880 additional output actually show up in terms of revenue for governments? 0:17:01.880 --> 0:17:04.520 How is that collected? And how would you expect the 0:17:04.600 --> 0:17:06.439 distribution to vary across the world. 0:17:07.520 --> 0:17:10.440 Well, I think in the United States medium term, they'll 0:17:10.480 --> 0:17:14.400 just be much much more. Healthcare will be new drugs, 0:17:14.480 --> 0:17:18.040 new medical devices, will have to test all these things, 0:17:18.080 --> 0:17:20.800 will have to produce them, and that will be It 0:17:20.840 --> 0:17:23.239 was already the case that those sectors were growing, but 0:17:23.320 --> 0:17:26.959 that growth will be accelerated. So that's where I think, God, 0:17:27.000 --> 0:17:29.760 they'll be the biggest difference longer term, and people will 0:17:29.760 --> 0:17:33.480 live longer because we'll fix at least partially various diseases 0:17:33.480 --> 0:17:36.280 and maladies. So if you live to be ninety four 0:17:36.800 --> 0:17:39.280 across a lifetime, you spend way more on healthcare than 0:17:39.320 --> 0:17:41.640 if you live to be seventy seven. And that's yet 0:17:41.680 --> 0:17:45.080 further growth for the healthcare sector. But some things like 0:17:45.200 --> 0:17:50.040 medical diagnosis that's already very cheap, like you know, a 0:17:50.080 --> 0:17:54.520 good large language model probably outperforms your current doctor at 0:17:54.600 --> 0:17:56.800 least if you type in what's wrong with you properly. 0:17:57.480 --> 0:18:01.399 But does that additional productivity or or the output generated 0:18:01.440 --> 0:18:04.600 by AI, does that actually show up an additional taxation 0:18:04.680 --> 0:18:05.320 for the government. 0:18:06.520 --> 0:18:11.119 Well, the healthcare sector generates an enormous amount of taxation revenue. 0:18:11.280 --> 0:18:13.639 I do think we'll have some sectors that maybe just 0:18:13.680 --> 0:18:16.639 become free in the same way that Wikipedia is free. 0:18:17.200 --> 0:18:19.800 So I could imagine say ten or twenty percent of 0:18:19.840 --> 0:18:23.440 the music sector is music you create at home using 0:18:23.480 --> 0:18:26.240 your own AI and it's a customized song for you, 0:18:26.600 --> 0:18:29.400 and maybe you paid a subscription for the service. But 0:18:29.480 --> 0:18:33.520 rather than spending more money on Spotify or a streaming service, 0:18:33.760 --> 0:18:36.960 you just build the music and that's a partial substitute 0:18:37.119 --> 0:18:40.000 for some human created music. I don't think human created 0:18:40.040 --> 0:18:42.800 music will go away at all. People want to enjoy 0:18:42.920 --> 0:18:45.560 the human touch, the feeling that you're a fan of Taller, 0:18:45.560 --> 0:18:48.160 Swift or whatever. But there's going to be a lot 0:18:48.200 --> 0:18:51.440 of AI generated music and art and many other areas, 0:18:51.680 --> 0:18:54.760 and some of it will be free. But that's not 0:18:54.840 --> 0:18:58.600 a problem from a revenue standpoint. So instead of spending 0:18:58.640 --> 0:19:02.600 money on safe buying a picture, you create one digitally 0:19:02.640 --> 0:19:05.040 at home with your AI. You'll spend that money on 0:19:05.080 --> 0:19:05.640 something else. 0:19:06.480 --> 0:19:09.119 I got to ask, now, are you a swiftie. 0:19:09.440 --> 0:19:13.959 No, it's worrying for me. It's a little too predictable, 0:19:14.400 --> 0:19:17.520 so I have to say I'm not, but I'm glad 0:19:17.600 --> 0:19:19.440 other people are. Let's put it that way. 0:19:19.600 --> 0:19:21.880 Do you have a theory let's talk, let's just talk 0:19:21.920 --> 0:19:24.560 about music. Do you have like a theory of the 0:19:24.600 --> 0:19:29.159 swift phenomenon? Like, is there a reason? What is it? 0:19:29.200 --> 0:19:32.000 Because it's just so, doesn't I don't know what's your 0:19:32.359 --> 0:19:36.159 what's your meta take on the tailor, the swifties and 0:19:36.280 --> 0:19:37.440 culture and society. 0:19:38.600 --> 0:19:41.520 Well, it's super polished and because of the Internet, the 0:19:41.640 --> 0:19:44.640 very biggest of celebrities can be much bigger than before, 0:19:45.160 --> 0:19:47.960 and someone will fill a few of those slots. But 0:19:48.000 --> 0:19:52.159 I think also how she presents herself. She has the 0:19:52.240 --> 0:19:57.520 guise of being attractive without feeling threatening to other women. Uh, 0:19:57.560 --> 0:20:00.560 and there's something all American about her and quite generic. 0:20:00.840 --> 0:20:03.600 She doesn't rule out the fandom of that many people 0:20:03.800 --> 0:20:06.520 m hm. And she's the one who's filled that slot. 0:20:06.600 --> 0:20:10.360 She's been brilliant at managing her career and seems to 0:20:10.359 --> 0:20:12.520 just stick at it and has an incredible work ethic. 0:20:12.560 --> 0:20:15.679 People say the shows are incredible. As more and more 0:20:15.760 --> 0:20:18.640 life goes online, who can give a good show? It's 0:20:18.640 --> 0:20:21.560 the charisma and looks. Point Well, she seems a pleasant 0:20:21.600 --> 0:20:23.600 that I've never been to one, but I hear plenty 0:20:23.600 --> 0:20:26.960 of reports. And you put all that together and she's, 0:20:27.119 --> 0:20:28.919 you know, the megastar of the music world. 0:20:29.359 --> 0:20:32.800 Do you think like culture, Like there's this popular idea 0:20:32.800 --> 0:20:34.800 that culture is sort of dead, and I do think 0:20:34.800 --> 0:20:38.520 that is probably overstated, But you know, you look at 0:20:38.560 --> 0:20:41.760 movies and people have people have observed this for a 0:20:41.800 --> 0:20:45.040 long time. It's just rehashes of franchises that have been 0:20:45.080 --> 0:20:47.679 around for thirty years. And I think if you look 0:20:47.720 --> 0:20:52.000 at Spotify's streaming, there's still this overwhelming tyranny of the 0:20:52.000 --> 0:20:56.560 boomer rock et cetera, and this feeling that culture in 0:20:56.680 --> 0:20:59.280 many respects is that's rehash that is very hard for 0:20:59.359 --> 0:21:02.160 new things to to break out. I mean, Taylor Swift 0:21:02.200 --> 0:21:07.000 at this point is decades old phenomenon. Is that real 0:21:07.160 --> 0:21:09.520 in your view? Or is that just people in the 0:21:09.560 --> 0:21:13.479 pundit class who've been lazy and not discovering new things 0:21:13.520 --> 0:21:15.560 and not actually putting in the effort because they're not 0:21:15.600 --> 0:21:17.320 young anymore and they're not going out and they say 0:21:17.320 --> 0:21:18.639