WEBVTT - AI's Biggest Problem May Be Its Pace | The Professor Is In 0:00:00.000 --> 0:00:04.240 I'm saying, we have an extraordinary new technology that can 0:00:04.320 --> 0:00:09.480 create greater abundance, and we should stop using it feels 0:00:09.760 --> 0:00:11.080 totally upside down. 0:00:17.640 --> 0:00:21.120 So you released a really fun diving in video this 0:00:21.160 --> 0:00:23.920 week in which you actually looked back in time at 0:00:23.960 --> 0:00:27.120 the Victorian boot making industry. 0:00:28.320 --> 0:00:32.239 Yeah, fascinating period. You might ask, why the heck do 0:00:32.280 --> 0:00:34.519 we care about making boots in the eight and fifties, 0:00:34.560 --> 0:00:39.479 And that's because it had an amazing technological transformation that 0:00:39.640 --> 0:00:42.960 meant the average worker could produce four times as much 0:00:43.159 --> 0:00:45.640 as they could before. Hey, I don't know if that 0:00:45.720 --> 0:00:48.479 reminds you of anything that you might be feeling a 0:00:48.479 --> 0:00:50.080 little bit of anxiety about right now. 0:00:50.280 --> 0:00:52.400 We had some great questions that came out of this 0:00:52.479 --> 0:00:55.560 from our audience, and we're going to dig into those today. 0:00:55.880 --> 0:00:56.880 I'm Megan Connors. 0:00:57.520 --> 0:01:00.520 I'm Justin Wolfers. This is the professor is in. Think 0:01:00.560 --> 0:01:04.679 about this as office hours. Meghan is out there reading 0:01:04.720 --> 0:01:07.080 your comments so she can bring your comments to me. 0:01:07.560 --> 0:01:08.720 The Professor's in. Let's go. 0:01:09.840 --> 0:01:14.360 Yeah, so you've what we found, or what you found, 0:01:14.440 --> 0:01:16.560 is that the number of jobs, total number of jobs 0:01:16.600 --> 0:01:20.160 really didn't budge that much, even though nearly every other 0:01:20.200 --> 0:01:22.240 aspect of boot. 0:01:22.000 --> 0:01:24.240 Making in England at this time did change. 0:01:24.720 --> 0:01:27.480 And when we pick up there, which is, I didn't 0:01:27.480 --> 0:01:31.080 find it. Hillary Verpond a brilliant young economic history and 0:01:31.120 --> 0:01:33.840 has done this great deep dive into boot making in 0:01:33.880 --> 0:01:36.319 the eighteen fifties. If you had told me eight days 0:01:36.319 --> 0:01:39.080 ago I cared at all about boot making England or 0:01:39.120 --> 0:01:42.920 the eighteen fifties, I would have said no. But she 0:01:43.080 --> 0:01:46.360 showed this is I think is where you're going. The 0:01:46.400 --> 0:01:49.240 total amount of employment didn't change even as this incredible 0:01:49.320 --> 0:01:52.880 labor saving technology came along. That's the good news, Megan. 0:01:52.920 --> 0:01:54.200 Do you want to tell people the other side of 0:01:54.200 --> 0:01:54.760 the coin. 0:01:55.040 --> 0:01:58.840 The I mean the tapes of jobs totally changed, So 0:01:59.000 --> 0:02:01.760 you know, certain elements of boot making did kind of 0:02:01.760 --> 0:02:06.120 go extinct and new jobs appeared in their place. 0:02:06.800 --> 0:02:10.320 Absolutely, all right. I know you've got questions about all yeah. 0:02:10.320 --> 0:02:12.800 So on that note, you know, we had a couple 0:02:12.840 --> 0:02:16.760 of different people ask what happened to pay? And I 0:02:16.800 --> 0:02:19.480 know you might not have the specifics in this circumstance, 0:02:19.560 --> 0:02:23.760 but what do we know about technological revolutions in general 0:02:24.000 --> 0:02:26.320 and their impact on wages? 0:02:27.000 --> 0:02:30.280 Right? This particular study didn't have any pay data, so 0:02:30.360 --> 0:02:33.480 I can't say anything about Victorian England in the eighteen fifties. 0:02:33.919 --> 0:02:36.239 And I know I'm going to surprise you, but I'm 0:02:36.280 --> 0:02:39.200 not actually an expert on Victorian England in the eighteen fifties, 0:02:39.280 --> 0:02:43.760 even though I dress in linen. Okay, So what typically happens, 0:02:44.320 --> 0:02:47.720 it's not always the case that wages rise immediately. The 0:02:47.760 --> 0:02:50.600 period after in the Industrial Revolution was when Charles Dickens 0:02:50.639 --> 0:02:54.160 started writing, and Dickensie and Tail is full of young 0:02:54.240 --> 0:02:59.040 children laboring away in factories in rags, barely able to 0:03:00.160 --> 0:03:04.080 make ends meet. So it took the immediate transition is 0:03:04.120 --> 0:03:08.720 not necessarily good news. What happens though, over time is 0:03:09.040 --> 0:03:12.679 if the pie is bigger, and I don't want to 0:03:12.680 --> 0:03:15.440 say there's magic, which means the pie is always genuinely 0:03:15.480 --> 0:03:17.400 distributed in good ways. But if the pie is big 0:03:17.480 --> 0:03:20.320 enough and humans are an important part of producing that pie, 0:03:20.600 --> 0:03:23.320 some of it comes back to roost. So the story 0:03:23.360 --> 0:03:27.040 on wages is there's two stories. One is what's happening 0:03:27.080 --> 0:03:32.640 on average on average, eventually bigger pie bigger servings to 0:03:32.680 --> 0:03:36.600 workers in the short run, that may not be true. 0:03:37.000 --> 0:03:39.000 And then on average was doing a lot of work 0:03:39.840 --> 0:03:43.400 if you were a cord swainer of that word, I 0:03:43.440 --> 0:03:45.520 don't even know what it means, but I love I 0:03:45.520 --> 0:03:48.200 think I might be a cord swaener in my next life. 0:03:48.200 --> 0:03:51.840 That's one of the old artisan or bootmaking jobs. The 0:03:51.880 --> 0:03:55.160 demand for cordswainn has felt dramatically and so it's very 0:03:55.480 --> 0:03:59.680 likely that cord swainers saw their pay go down. If 0:03:59.720 --> 0:04:02.000 you want to to give a modern day interpretation of this, 0:04:02.600 --> 0:04:06.080 the world of translation is changing right now due to AI. 0:04:06.480 --> 0:04:10.680 It's very likely the current translators are seeing lower wages now. 0:04:11.120 --> 0:04:13.760 The thing we learned from Victorian bootmaking is cord swain 0:04:13.800 --> 0:04:16.279 has went away, but the price of boots fell, so 0:04:16.320 --> 0:04:19.400 therefore the number of boots we make went up, and 0:04:19.440 --> 0:04:22.560 that created more demand for folks to work in factories, 0:04:22.920 --> 0:04:26.039 you know, foreman's and foremen and machinists and riveters and 0:04:26.080 --> 0:04:28.159 so on, which would be the equivalent of today. What 0:04:28.200 --> 0:04:30.279 we're seeing is the rise of the prompt engineer. And 0:04:30.360 --> 0:04:32.799 I want to be clear, prompt engineers almost too glib 0:04:32.839 --> 0:04:34.640 for what's happening in a current moment. The reason I 0:04:34.720 --> 0:04:36.320 use that is that's a job title. You and I 0:04:36.360 --> 0:04:39.000 both know. One of the points that's really important to 0:04:39.240 --> 0:04:42.640 understand is there's going to be occupations we've never heard of, 0:04:43.080 --> 0:04:46.520 and if we follow it all the path of the 0:04:46.560 --> 0:04:49.600 boot makers, I'm not sure we will. Those other occupations 0:04:49.640 --> 0:04:51.720 are going to grow. Part of the reason they're going 0:04:51.760 --> 0:04:54.200 to grow is they're highly productive. That means is going 0:04:54.200 --> 0:04:56.400 to be high wages. Those high wages are also going 0:04:56.440 --> 0:04:58.800 to draw new and younger workers into it. 0:04:59.000 --> 0:04:59.560 Well, it was Gary. 0:04:59.600 --> 0:05:02.599 I mean, have any other evidence from like countries that 0:05:02.640 --> 0:05:06.159 have industrialized more recently. I'm thinking, like China, do we 0:05:06.200 --> 0:05:09.080 know what's happened? What happened with wages? My sense is 0:05:09.080 --> 0:05:12.800 that we saw the benefits of increased wages more quickly 0:05:13.200 --> 0:05:15.120 than maybe Victorian England did. 0:05:16.560 --> 0:05:19.960 Yes, I think that's right. So you know, so China 0:05:20.200 --> 0:05:25.520 famously has moved from largely a rural agricultural economy to 0:05:25.640 --> 0:05:30.400 an urban manufacturing economy. It is still the case there 0:05:30.440 --> 0:05:33.920 is immense poverty in the rural areas. But what's happened 0:05:33.960 --> 0:05:36.120 is there's been opportunities in the cities that have pulled 0:05:36.160 --> 0:05:38.640 workers out, which is why the average Chinese person's living 0:05:38.640 --> 0:05:43.719 standards have risen dramatically. And a very human reaction is 0:05:44.320 --> 0:05:47.400 won't these technological changes just make the rich rich and 0:05:47.480 --> 0:05:51.200 leave everyone else behind. In order to tell that story, 0:05:51.200 --> 0:05:54.800 given how much bigger the pie gets, that story is 0:05:54.839 --> 0:05:59.680 implicitly suggesting that the returns to the share of the 0:05:59.680 --> 0:06:03.320 pie that capital gets must be extraordinary or must shift 0:06:03.360 --> 0:06:08.880 the more almost implausibly. Right, So, the average income in China, 0:06:09.000 --> 0:06:10.400 I don't have a number in front of me. I'm 0:06:10.400 --> 0:06:12.080 going to guess that over a period of decades it 0:06:12.160 --> 0:06:13.240 rose literally tenfold. 0:06:13.640 --> 0:06:14.039 I think so. 0:06:14.960 --> 0:06:17.920 And you know, if the average wage of the work 0:06:17.960 --> 0:06:22.480 arose sixfold or eightfold, could have risen tenfold, that's all plausible. 0:06:22.960 --> 0:06:25.919 Saying that it all went to capital is just literally implausible. 0:06:27.120 --> 0:06:31.239 I don't want to be mistaken for saying trickle down economics, 0:06:31.920 --> 0:06:35.480 but this here is where we push the technological frontier. 0:06:35.520 --> 0:06:38.599 We learned to produce more, and you can't produce more 0:06:38.640 --> 0:06:41.960 without people engaged in the production process. That was true 0:06:41.960 --> 0:06:44.159 of bootleather, it was true of manufacturing, and it is 0:06:44.200 --> 0:06:47.360 still true of AI, at least at the current moment. 0:06:47.440 --> 0:06:50.560 Someone needs to ask, you know, AI puts us in 0:06:50.600 --> 0:06:53.680 a world where the price of questions has fallen to zero. 0:06:53.760 --> 0:06:56.080 But sorry, the price of answers has fallen to zero. 0:06:57.000 --> 0:06:59.479 But someone's got to figure out the interesting questions. That's 0:06:59.480 --> 0:07:00.000 the human. 0:07:00.480 --> 0:07:04.080 Okay, I want to dig into the demand side a 0:07:04.120 --> 0:07:08.279 little bit more, right, So, cheaper boots meant more boots 0:07:08.320 --> 0:07:11.160 and kind of we had a funny comment of someone asking, 0:07:11.240 --> 0:07:16.600 like how many AI images can I actually consume? And 0:07:16.640 --> 0:07:19.800 so I'm just curious kind of that idea of like 0:07:19.920 --> 0:07:23.640 demand saturation, Like is that a thing? Like how might 0:07:23.680 --> 0:07:25.040 that play under this story? 0:07:26.360 --> 0:07:29.680 Yeah, Megan, let's get personal about this. How many pairs 0:07:29.720 --> 0:07:32.000 of leather boots do you wind me? 0:07:32.400 --> 0:07:34.200 I tend to do one at a time, and I 0:07:34.240 --> 0:07:36.440 replace it every one to two years. 0:07:37.080 --> 0:07:39.320 Wow, you're really good at throwing out your old boots. 0:07:40.000 --> 0:07:42.120 Well, they usually fall apart, and that's what I know. 0:07:42.240 --> 0:07:43.880 I have to get rid of them. 0:07:44.400 --> 0:07:47.320 Okay, So I'll tell you about me, which is several 0:07:47.400 --> 0:07:50.160 years ago. I just signed my textbook contract and I 0:07:50.200 --> 0:07:52.480 wanted to treat myself, and I bought a beautiful pair 0:07:52.520 --> 0:07:55.120 of brown leather boots to go up just below the knee. 0:07:55.800 --> 0:07:58.680 You can decide whether that's masculine or not. I don't care. 0:08:00.080 --> 0:08:03.040 Then we're in Spain and I bought a pair that 0:08:03.160 --> 0:08:07.160 sort of looked part Spain, part Texas, and it's partly 0:08:07.200 --> 0:08:10.200 because I loved the Bob Dylan song boots of Spanish leather, 0:08:10.400 --> 0:08:13.520 And every time I put on those boots and I 0:08:13.560 --> 0:08:15.520 had a chance to wear them recently. What was the 0:08:15.560 --> 0:08:19.280 big country music star who came to Michigan recently. 0:08:19.360 --> 0:08:21.480 Zach Bryan. Last year, I wore. 0:08:21.360 --> 0:08:25.960 Those boots to a Zach Bryan concert. I wanted to 0:08:26.000 --> 0:08:28.480 wear them with cut off shorts, but Betsy wouldn't let me. 0:08:29.800 --> 0:08:31.560 So we got that pair of boots. And then, because 0:08:31.560 --> 0:08:33.760 I'm an Australian, I've got to have a pair of Blundstones. 0:08:34.520 --> 0:08:37.079 And then the first pair of Blundstones I brought over 0:08:37.120 --> 0:08:39.160 from Australia thirty years ago. Honestly, I still have them. 0:08:39.960 --> 0:08:42.040 I then bought another pair of Bludstones. Someone else gave 0:08:42.080 --> 0:08:43.800 me another pair of pluts. I got a lot of boots, 0:08:43.840 --> 0:08:45.360 and I live in a cold weather area. I need 0:08:45.360 --> 0:08:48.520 a lot of pairs of bits. That is so there 0:08:48.520 --> 0:08:49.800 are lots of ways you could look at that. You 0:08:49.840 --> 0:08:52.960 could say that's the lunacy of the current moment. How 0:08:52.960 --> 0:08:55.480 many pairs of boots do you need? Or what a 0:08:55.600 --> 0:08:57.920 joy it was to have a special outfit for my 0:08:58.000 --> 0:09:00.560 Zach Bryant concert and honestly did give me a smile, 0:09:00.600 --> 0:09:02.400 and lots of people add a good laugh at me 0:09:02.679 --> 0:09:06.040 and with me, and those smiles of what economics is 0:09:06.080 --> 0:09:09.160 all about. So I don't feel like I'm done with 0:09:09.280 --> 0:09:12.320 leather boots yet. So next big revolution, I'm more for it. 0:09:12.679 --> 0:09:16.360 Let's come back to AI. AI is doing cognitive work. 0:09:17.080 --> 0:09:20.040 I understand people might object to the word cognitive, but 0:09:20.320 --> 0:09:25.640 cognitive adjacent. How many AI images can you consume? Well, 0:09:25.679 --> 0:09:28.760 I will tell you I remember ancient history. I think 0:09:28.760 --> 0:09:31.960 it was four years ago. Whenever I would want to 0:09:32.080 --> 0:09:33.840 write a column for the New York Times and we 0:09:33.960 --> 0:09:36.360 want to have art with a column, and they would 0:09:36.360 --> 0:09:38.480 actually have to call someone from the art department, and 0:09:38.520 --> 0:09:41.360 there is something visually pleasing about space on the page 0:09:41.360 --> 0:09:44.800 and so on. We're now writing on substack, and substack 0:09:44.840 --> 0:09:47.840 has a vast image library. We do have access to AI. 0:09:48.120 --> 0:09:53.319 Anyone can now produce somewhat something semi professional. The point 0:09:53.400 --> 0:09:56.080 that I think a comment is making that's really important 0:09:56.200 --> 0:10:01.800 is the value of the last image I created isn't 0:10:01.800 --> 0:10:05.840 that high? But the first doesn't I created? Were they 0:10:05.880 --> 0:10:08.320 were images that went into my textbook, for instance, or 0:10:08.360 --> 0:10:10.680 images I put into a substack and so yeah, the 0:10:10.760 --> 0:10:13.160 last one's not that valuable, but all the ones in 0:10:13.200 --> 0:10:16.000 between were quite valuable. You know, if there's too much 0:10:16.040 --> 0:10:19.600 AI images in your life, cry uncle, demand be artisanally drawn. 0:10:20.559 --> 0:10:24.080 But also remember that there are still people all around 0:10:24.080 --> 0:10:28.560 the world doing without, and when we create more, we 0:10:28.679 --> 0:10:30.079 help those who do without. 0:10:30.440 --> 0:10:33.400 Yeah, I think one thing that I've been thinking about 0:10:33.640 --> 0:10:37.560 is kind of where we're seeing AI show up first. 0:10:38.240 --> 0:10:41.880 And I think in the US. 0:10:40.600 --> 0:10:44.079 It's in a lot of media to the point where 0:10:44.120 --> 0:10:47.920 it's for some people it's not actually like adding to 0:10:47.960 --> 0:10:48.439 their lives. 0:10:48.440 --> 0:10:49.319 In some ways, it can be. 0:10:49.280 --> 0:10:54.080 Detracting, right AI slot, Whereas I've heard that like China 0:10:54.120 --> 0:10:57.520 has been focused a little bit more on robotics side 0:10:57.559 --> 0:11:01.080 of AI and things that are actually like helping people 0:11:01.120 --> 0:11:04.839 with domestic chores, like things that are a little maybe 0:11:05.480 --> 0:11:08.320 have seemed to have a little more value, And so 0:11:08.800 --> 0:11:10.800 I don't know, I think I think a lot of 0:11:10.800 --> 0:11:13.959 our conversation has been really focused on what where we're 0:11:13.960 --> 0:11:17.240 seeing AI now, which is maybe arguably not the best 0:11:17.320 --> 0:11:21.760 use case for it, and it's making it hard to 0:11:21.800 --> 0:11:26.080 think about kind of the wider applications. 0:11:26.600 --> 0:11:28.280 Let me try something brave. I want to connect this 0:11:28.320 --> 0:11:32.000 to Victorian England and boot making. So people complain that 0:11:32.040 --> 0:11:34.400 when the cost of producing an image gets low, people 0:11:34.600 --> 0:11:37.240 produce low quality images, and we call that aislop or 0:11:37.280 --> 0:11:40.520 low quality text. Well, it's also the case you've been 0:11:40.559 --> 0:11:42.280 to a shoe store. There is such a thing as 0:11:42.679 --> 0:11:48.000 boot slop. Right back in the day, in the eighteen fifties, 0:11:48.040 --> 0:11:51.080 boots were artisanal though are handmade. They were workers boots, 0:11:51.080 --> 0:11:54.000 They were made to last. They were a major investment. 0:11:55.160 --> 0:11:57.960 Not quite as important as a car, but closer to 0:11:58.000 --> 0:12:01.199 that than how we think about it today. All technologies 0:12:01.240 --> 0:12:04.199 create slot. I have seen some horrendous pairs of boots 0:12:04.240 --> 0:12:08.000 in the world. You're broader question of like, where's the 0:12:08.040 --> 0:12:10.280 best use case for this? This is a moment of 0:12:10.280 --> 0:12:14.080 immense transition. The only way to figure out the future 0:12:14.160 --> 0:12:16.559 is to try a bunch of things. And if people 0:12:16.559 --> 0:12:19.880 don't like AI generated images, and the verdict seems to 0:12:19.880 --> 0:12:22.839 be that they don't, we're going to try using fewer 0:12:22.880 --> 0:12:25.240 of them. You know, you try some things, they work out, 0:12:25.320 --> 0:12:27.840 You try others they don't. I think there's a much 0:12:27.880 --> 0:12:29.840 deeper question here, and this is where a lot of 0:12:29.840 --> 0:12:33.440 the discussions in Silicon Valley go. There's one occupation where 0:12:33.440 --> 0:12:36.000 it's come to absolutely dominate the workflow, which is it 0:12:36.040 --> 0:12:39.079 turns out that AI can do the work of a 0:12:39.160 --> 0:12:43.600 highly trained coder. A bunch of coders right now are 0:12:43.640 --> 0:12:46.080 going to scream at me in the comments and say, no, 0:12:46.160 --> 0:12:48.040 they can't do what I do. But I will literally 0:12:48.080 --> 0:12:50.880 tell you Bill Gates told me that it can code 0:12:52.160 --> 0:12:54.199 as well or better than him, and that's been his 0:12:54.240 --> 0:12:57.880 life's work. How is that for a little just name 0:12:57.960 --> 0:13:03.360 dropping in there? The deeper question is how much are 0:13:03.360 --> 0:13:06.280 we all like coders. There's lots of reasons to think 0:13:06.320 --> 0:13:10.800 coding is special, Like it's literally codified, it's a very 0:13:10.840 --> 0:13:16.120 simple language. At its heart, coding is things like if 0:13:16.120 --> 0:13:18.560 then statements, it's loops and so on, And so the 0:13:18.640 --> 0:13:21.840 ability for a computer to learn a constrained set of 0:13:21.880 --> 0:13:28.040 tasks that to some people involve constrained imagination, we shouldn't 0:13:28.080 --> 0:13:30.080 be surprised, and then the rest of us should be fine, because, 0:13:30.120 --> 0:13:32.719 of course AI could never do economics. Of course, it 0:13:32.760 --> 0:13:34.640 could never paint a landscape. Of course it could never 0:13:34.679 --> 0:13:38.120 sing an opera. But let's come back to singing an opera. 0:13:39.120 --> 0:13:41.040 At a fairly deep level, it's got a lot in 0:13:41.040 --> 0:13:44.120 common with coding. It's a fairly constrained language, a set 0:13:44.160 --> 0:13:46.680 of notes on the page. We can look to a 0:13:46.679 --> 0:13:48.840 long history to figure out what works well and what 0:13:48.960 --> 0:13:52.600 works poorly. On and on it goes. You might think 0:13:52.640 --> 0:13:55.760 that's a useful analogy. You might not. The point is 0:13:55.840 --> 0:13:58.520 neither of us knows. And so what we are going 0:13:58.600 --> 0:14:01.280 to see is, you know, the question is, look, codeers, 0:14:01.320 --> 0:14:05.800 their lives are changed forever, same as cord sweiners to 0:14:05.920 --> 0:14:09.880 you right now, megan to me, to our viewers, they 0:14:10.600 --> 0:14:13.600 like chord swiners and coders or are they more like 0:14:14.520 --> 0:14:21.000 jobs that continue? Because that wildly human aspect is so important. 0:14:21.200 --> 0:14:23.640 And when I stayed it that way, lots of people 0:14:23.680 --> 0:14:26.040 have a particular view, But I really think it's an 0:14:26.080 --> 0:14:26.640 open question. 0:14:26.920 --> 0:14:27.240 Okay. 0:14:27.320 --> 0:14:29.880 That leads us well into the next question. So the 0:14:29.960 --> 0:14:33.760 video also stressed the importance of peace, and I think 0:14:33.800 --> 0:14:37.080 you said Hillary Vpon stressed that the moderate piece of 0:14:37.160 --> 0:14:41.280 change in Victorian boot making really helped shield incumbent workers. 0:14:41.320 --> 0:14:43.440 They were able to retire in their roles. It was 0:14:43.520 --> 0:14:47.560 just new people didn't enter. So what do you take 0:14:47.600 --> 0:14:51.400 away from that, like, should policy makers be trying to 0:14:51.520 --> 0:14:55.480 slow the peace of AI? That seems difficult to do 0:14:55.560 --> 0:14:58.920 in a global economy, I mean, should they be trying 0:14:58.960 --> 0:15:01.880 to direct AI to certain goals? 0:15:02.000 --> 0:15:03.520 Like, what do you take away from that? 0:15:03.960 --> 0:15:06.160 Yeah, let me throw out a few of Hillary's facts 0:15:06.200 --> 0:15:09.080 first because I think they're really interesting. So what you 0:15:09.120 --> 0:15:13.760 saw is, over you know, a period of several decades, 0:15:14.240 --> 0:15:17.080 literally two thirds of the jobs in what you call 0:15:17.120 --> 0:15:20.240 old occupations, the called swayiners and the like, disappeared. The 0:15:20.320 --> 0:15:23.440 number of people working in bootmaking didn't change. Two thirds 0:15:23.440 --> 0:15:25.640 of the occupy, the folks who are in the old occupations, 0:15:25.680 --> 0:15:28.640 their jobs disappeared, a new set of jobs turned up. 0:15:28.680 --> 0:15:31.480 They ended up taking two thirds of the new jobs. 0:15:31.560 --> 0:15:35.560 So it really is the total level employments the same, 0:15:35.600 --> 0:15:39.400 but the tasks being done all literally are just fundamentally different, 0:15:39.920 --> 0:15:43.320 wildly different. And what was interesting about this case is 0:15:44.000 --> 0:15:46.720 a majority of the folks who had the old artisanal 0:15:46.760 --> 0:15:52.480 bootmaking roles ended up as old artisanal bootmakers. What happened 0:15:52.560 --> 0:15:55.880 was they slowly retired and died and their sons didn't 0:15:56.000 --> 0:15:58.160 enter the same trade. And I mean sons literally, because 0:15:58.160 --> 0:16:00.680 that's how this worked. There's an interesting side story, by 0:16:00.720 --> 0:16:03.160 the way, which is the women who worked as bootmakers, 0:16:03.280 --> 0:16:06.800 they did leave bootmaking altogether. They were the ones whose 0:16:07.400 --> 0:16:10.440 professional ambitions had to give way to the economic reality. 0:16:10.440 --> 0:16:15.120 The bootmaker couldn't employ mum and dad, and so then 0:16:15.120 --> 0:16:17.000 what you had was the sons of the bootmakers either 0:16:17.040 --> 0:16:19.960 went into new boot making occupations or went into something 0:16:20.000 --> 0:16:23.560 different altogether. And so there was a lot of numbers 0:16:23.560 --> 0:16:25.640 to say what you said, which is the old bootmakers 0:16:25.720 --> 0:16:28.160 retired in their jobs, but no one else was hired 0:16:28.200 --> 0:16:32.600 as a kordswaanner after that. So where does that bring 0:16:32.720 --> 0:16:36.280 us in the current moment. So the headline is Dario Amode, 0:16:36.280 --> 0:16:40.720 the CEO of Anthropic, says that within five years, was it, 0:16:40.840 --> 0:16:44.720 half of all white collar work will be done by AI. 0:16:45.960 --> 0:16:48.600 So first thing is he's a technologist, not an economist. 0:16:49.520 --> 0:16:52.120 Economists have studied history. When you look at history, things 0:16:52.160 --> 0:16:55.320 never happened quite that fast. And so what he should 0:16:55.320 --> 0:16:57.800 have said is he believes the code will be capable 0:16:57.840 --> 0:17:00.920 of doing half of the work within five years. And 0:17:00.960 --> 0:17:02.280 if he says it that way, by the way, I 0:17:02.280 --> 0:17:07.480 actually think that's true, or it's in the neighborhood of true. 0:17:08.040 --> 0:17:12.359 But organizations are run by people. Organizations run on people. 0:17:12.480 --> 0:17:16.200 Organizations adapt and adopt slowly. There's no way it happens 0:17:16.280 --> 0:17:20.040 at quickly, but to the extent that happens quickly, it 0:17:20.119 --> 0:17:22.399 means that every member of our audience is wearing a 0:17:22.400 --> 0:17:26.040 white collar right now should be terrified that their job's gone. 0:17:26.080 --> 0:17:29.720 In five years. That would be an economic disruption larger 0:17:29.760 --> 0:17:33.440 than the global financial crisis, larger than COVID, larger and 0:17:33.480 --> 0:17:36.920 eything would be an economic upheaval that would cause political 0:17:36.920 --> 0:17:39.800 and social change of really an extraordinary measure. So then 0:17:40.000 --> 0:17:43.720 a very common conversation here I'm just reporting, in fact, 0:17:43.760 --> 0:17:47.480 a very common conversation among there's a group of economists 0:17:47.520 --> 0:17:50.400 and technologists and policy wonks who they meet each other 0:17:50.400 --> 0:17:52.439 at the same conference every three weeks and talk to 0:17:52.480 --> 0:17:58.000 each other. A very common conversation is how do we 0:17:58.040 --> 0:18:01.160 slow this down? And actually some of the folks from 0:18:01.200 --> 0:18:03.560 the AI labs are part of that conversation and they're 0:18:03.600 --> 0:18:08.920 saying that out loud. There's a range of policy solutions. 0:18:09.000 --> 0:18:11.359 Is what's called a token tax, which is basically tax 0:18:11.400 --> 0:18:16.480 people for the use of AI. You can require the 0:18:16.720 --> 0:18:19.840 businesses hold on to their current workers even as they 0:18:19.840 --> 0:18:23.000 adopt AI. You can imagine all sorts of regulatory responses 0:18:23.000 --> 0:18:25.639 that are designed to slow things down. You know if 0:18:25.640 --> 0:18:29.320 we had a strong union movement so on. So, first 0:18:29.359 --> 0:18:32.199 of all, want to just acknowledge something, which is this 0:18:32.400 --> 0:18:36.640 echoes the Ludites who were worried machines would take their job. 0:18:36.680 --> 0:18:36.840 Now. 0:18:36.840 --> 0:18:40.399 The funny thing is when economists talk about Luddites, they 0:18:40.400 --> 0:18:43.880 almost always mean it in a majorative way, someone who 0:18:43.920 --> 0:18:45.880 thinks that the future is coming to take their job 0:18:45.920 --> 0:18:49.960 and the future will make their lives worse, but actually 0:18:50.080 --> 0:18:54.679 have a little empathy. The Ludites were right, technology was 0:18:54.760 --> 0:18:58.000 coming for their jobs. Change was coming, and if change 0:18:58.040 --> 0:19:03.480 came quickly, change would be painful. That's true. And the 0:19:03.480 --> 0:19:06.040 economists intuition that if we adopt the best technologies all 0:19:06.040 --> 0:19:07.399 the time, the pie will be as big as it 0:19:07.400 --> 0:19:10.040 can possibly be, and therefore we have the capacity to 0:19:10.080 --> 0:19:14.000 feed more people. That's also true. Of course, having the 0:19:14.000 --> 0:19:16.119 capacity to feed more people doesn't mean we actually do it. 0:19:16.320 --> 0:19:18.200 That's also also true. 0:19:19.359 --> 0:19:22.600 Are other video on recent cuts to snap. 0:19:23.720 --> 0:19:27.400 Yeah. Right, we are currently the richest America has ever been, 0:19:27.840 --> 0:19:32.520 but our safety net is not showing that and people 0:19:32.880 --> 0:19:35.679 are not enjoying the spoils. And so that's where we 0:19:35.720 --> 0:19:38.719 get all these other conversations about universal basics can come 0:19:38.720 --> 0:19:39.840 and so on. But I want to stick to this. 0:19:40.320 --> 0:19:43.399 What you're hearing a lot of is a sense that 0:19:43.440 --> 0:19:45.680 if we could slow things down, the transition costs would 0:19:45.720 --> 0:19:49.840 be a lot lower. Let me point something out why 0:19:49.880 --> 0:19:53.879 this sits quite uncomfortably on the economics. Saying we have 0:19:53.920 --> 0:19:59.119 an extraordinary new technology that can create greater abundance and 0:19:59.160 --> 0:20:05.640 we should stop using it feels totally upside down. It 0:20:05.720 --> 0:20:10.160 requires a language of talking about greater abundance versus transition costs, 0:20:10.600 --> 0:20:13.240 and transition costs is just a language economists have never 0:20:13.240 --> 0:20:15.680 been very good at. If you think about the debate 0:20:15.720 --> 0:20:19.120 about international trade, our view as economists was always will 0:20:19.119 --> 0:20:21.639 we get the pie to be bigger? And what we 0:20:21.680 --> 0:20:25.119 didn't recognize is towns like Flint, Michigan will disappear, and 0:20:25.160 --> 0:20:28.360 that transition cost is very real. And in our economic 0:20:28.400 --> 0:20:30.760 models we assume that the guy who's an auto worker 0:20:30.760 --> 0:20:33.320 becomes a home healthcare aid tomorrow, and the reality is 0:20:33.400 --> 0:20:36.679 he doesn't see our past work on pink color jobs, 0:20:36.760 --> 0:20:41.640 masculinity transitions, and so on. So I think my profession 0:20:41.720 --> 0:20:47.200 hasn't got yet a refined enough language to have this 0:20:47.240 --> 0:20:50.040 conversation coherently. And the problem is if we don't have 0:20:50.080 --> 0:20:55.120 it coherently. It's happening incoherently instead. So I do think 0:20:55.200 --> 0:20:59.119 Hillary of a Pond, who warned me having coffee with It, 0:20:59.720 --> 0:21:02.600 that the pace of change is central to the economic 0:21:02.640 --> 0:21:04.760 and social disruption. I think she's one hundred percent right. 0:21:05.080 --> 0:21:07.840 What's a lot harder is what to do with that observation. 0:21:08.280 --> 0:21:13.360 Okay, So another element of change that you discussed was geography. 0:21:13.920 --> 0:21:18.000 The new technology dramatically changed the geography of where people 0:21:18.000 --> 0:21:21.280 were working. The factories tended to be based in kind 0:21:21.320 --> 0:21:26.399 of two major parts of England and human geography. We 0:21:26.480 --> 0:21:31.440 call that agglomeration. Uh, look at you right bringing it in. 0:21:31.560 --> 0:21:35.440 But with so much of today's work being digital remote, 0:21:35.680 --> 0:21:39.760 I'm curious if you think geography is going to be part. 0:21:39.640 --> 0:21:44.080 Of the AI story. Yeah, what what's your take on that? 0:21:44.800 --> 0:21:47.800 I thought the geography story was fascinating. So basically every 0:21:47.800 --> 0:21:51.239 town had bootmakers. The old school bootmakers went away, we 0:21:51.280 --> 0:21:54.439 moved to factories, and factories agglomerated, excellent use of the 0:21:54.480 --> 0:21:58.040 term in a small number of cities. I am not 0:21:58.119 --> 0:22:02.560