David Autor on the Labor Market
When economic news, especially that revolving around working, gets reported, it tends to get reported in aggregate – the total number of jobs affected or created, the average wage paid, the impact on a defined geographic area. This is an approach labor economist David Autor knows well. But he also knows that the aggregate often masks the effect on the individual.
In this Social Science Bites podcast, Autor, the Daniel (1972) and Gail Rubinfeld Professor, Margaret MacVicar Faculty Fellow, Google Technology and Society Visiting Fellow at the Massachusetts Institute of Technology, examines two momentous changes to global economics and how they play out for individuals. He explains to interviewer David Edmonds how the rise of China’s manufacturing dominance and the widespread adoption of artificial intelligence likely are and will affect individual people accustomed to do specific tasks for pay.
What he finds is not as straightforward as the headlines alluded to above. Take China and its remarkable ascent and how that impacted the United States.
“[The rise] benefited a lot of people. It lowered prices. It allowed American companies to kind of produce a lot of products more cheaply. You know, it’s hard to imagine Apple’s growth without China, for example, to do all that assembly, which would have been extremely expensive to do in the United States. At the same time, it displaced a lot of people, more than a million, and in a very geographically and temporarily concentrated way, extremely scarring the labor market. Now those people also got lower prices, but that’s not even remote compensation for what they lost. And now there are new jobs — even in those places where those trade shock occurs — but it’s not really the same people doing them. It’s not the people who lost manufacturing work.”
Concerns about these shocks have been widespread in the 2020s, but the tough if erratic talk about tariffs coming from the U.S. president centers on the idea of restoring something (while ignoring question of that thing ever existed or if it makes sense to go back). Autor argues that the administration actually is asking the right question – but they are arriving at the wrong answers, He notes that the U.S. currently has a half a million unfilled manufacturing jobs open already, a sizeable figure relative to the nation’s 13 million manufacturing workers. But that number itself is roughly a tenth of China’s 120 million.
“We cannot compete with them across every front. .. What we should be very deeply worried about is losing the frontier sectors that we currently maintain. Those are threatened. So aircraft, telecommunications, robotics, power generation, fusion, quantum computing, batteries and storage, electric vehicles, shipping. These are sectors that we still have (except for shipping, actually) but China is making incredibly fast progress, and instead of trying to get commodity furniture back, we need to think about the current war we’re in, not the last war.”
At MIT, Autor is co-director of the Blueprint Labs, whose “scholars apply their unique expertise to pressing problems in education, health care, and the workforce,” and co-director of the MIT Shaping the Future of Work Initiative. Off campus, he is a research associate and co-director of the Labor Studies Program at the National Bureau of Economic Research.
To download an MP3 of this podcast, right-click this link and save. The transcript of the conversation appears below.
David Edmonds: Based at MIT, David Autor is one of the world’s leading labor economists. He’s well known in particular for his research into how the rise of China has affected employment in the US, as well as how the labor market adapts to new technology. David Autor, welcome to Social Science Bites.
David Autor: Thank you. It’s a delight to be here. I’m looking forward to this conversation.
David Edmonds: We’re talking today about the labor market. You’re a labor economist, naive question, but what does a labor economist do?
David Autor: Great question, which I’ve never been asked, by the way. You know, the labor market I think of as the most important social institution, right? Most people spend a great deal, even the majority of their waking hours, in work. Work doesn’t just provide income, but it provides identity, status, friends, purpose, and in some sense, I also think it’s kind of the ballast of the ship of state of democracies, because in countries where most people are working, everyone is viewed as a contributor as well as a claimant. So labor economists study things that affect employment, opportunity, skills, earnings mobility, and not just on average, not as you know, what is GDP? What is the stock market? But like the distributional consequences, how are different people faring?
You can have a labor market that seems to be doing well on average, and yet no one’s at the average. Some people are doing extremely well. Some people are doing extremely poorly. So we care a lot about the distribution. A lot of my work focuses on workers without a four-year college degree — who are the majority of American workers, by the way. Forty percent of US workers have a four-year college degree. Sixty percent do not. So, we ought to be especially attuned to how that group is doing.
David Edmonds: I want to get onto that distinction later between the highly educated and the less highly educated. But is it true to say that a labor economist is primarily interested in change, in shocks to the labor market, or how the labor market adapts to change.
David Autor: Certainly, that’s the most interesting part. But even without shocks, we’d say like, “Well, what determines the distribution of income? Why do some people make more than others? Why do some people advance over the lifetime and some do not? Why are some employed and some unemployed?” But certainly, the work that I do is very focused on two sources of change. One is technological and all of the incredible technological developments over the last five decades that have been beneficial, but also disruptive and in many cases unevenly beneficial and harmful to some. Then the other source of labor market change that I’ve studied a lot over the last more than a decade is the effect of China’s monumental rise as a kind of economic and manufacturing superpower – how that’s affected workers and communities the United States.
David Edmonds: We’re going to talk about both those areas. What are your key sources of data? Where do you get your stats from?
David Autor: Well, you know, a lot of my work uses representative data sources that are produced by the US government. So, the Census of population, the American Community Survey, the Dictionary of Occupational Titles and the O*NET, which are data on occupations and what people do. And then also data on trade flows by goods across countries. I use that a lot.
And then I study specific technologies and try to understand how they may be changing the tasks that people do. I think much more in terms of tasks than jobs. So, a job is a constellation of different tasks. You do many, many things you obviously, you write your research, use talk on podcasts, you spend time fixing your equipment to make that work right. Those are all part of your job. And so technology primarily displaces or creates new tasks, rather than wholesale eliminating jobs. But that’s not always good. If your things that you do that are valuable are all of a sudden automated, well, your skills, that expertise that you painstakingly develop, is suddenly less valuable.
David Edmonds: So, that’s interesting, because somebody who does a part of what I do might now self-identify as a podcaster, but the skill set is just identical to what a radio broadcaster did 25-30 years ago.
David Autor: Yeah. I mean, occupations are the words we use to describe things, types of jobs. But first of all, they’re broad. And second of all, within an occupation, there’s incredible diversity of what people do and the range of things they do. And then, moreover, there are new things that are being created that didn’t exist, right? A podcaster is a new job …
David Edmonds: … a new job with old skills.
David Autor: Yeah. So sometimes it’s a remix, a remix of existing skills. Many times, what I think of as new work is work that requires expertise that didn’t previously exist or wasn’t previously valuable. That expertise could be as someone who does prompt engineering, right, for example, or, you know, there was a time when we didn’t need pilots. We didn’t need people who were air traffic controllers. We just didn’t have those technologies. A lot of new work is a very important part of the way the labor market evolves and demands expertise in new things that we do.
But I want to be clear, not all new work comes from new technology. It also comes from changing tastes, changing income levels, changing demographics, right? So we have, as our populations are aging, we have all kinds of new specialized activities that just care for the elderly. And the converse was true in the 1960s and 70s, when we had such young population.
David Edmonds: And do you pick up these new jobs? How, from Census material, how do you work out what these new jobs are?
David Autor: Yeah, actually. So in recent work with my co-authors, Anna Salomons and Caroline Chin and Bryan Seegmiller, we actually did use data from the Census Bureau from 1915 forward. So people write in their occupational titles on their Census forms, and then the Census Bureau has to categorize those. They create these basically lists from people of our era might call phone books, of all the occupations that people might write in, and then they classify those into codes. But if they notice something that people are writing in that they haven’t seen before, and they see it enough time to then go to a supervisor and say, “Well, can we call this a new title?” And then those get added to the list. So we actually use the flow of new titles across more than eight decades to measure the types of new work appearing, what broad categories they’re in, what education levels they are associated with. And so yeah, that’s been an important part of my work over the last five years.
David Edmonds: And the increase in the number of titles, has that been linear, or has there been an exponential increase in the number of jobs?
David Autor: I would say the time that we see the most growth of new titles is between 1940 and 1950 over the Second World War, when the society changed the most. It’s an irony that I really appreciate, our labor market and especially the structure of occupations, has been more stable over the last few decades than in other eras. Now, the data, I should say, are not perfectly suited to making these kind of cardinal numerical comparisons, because the sort of level of effort with which the Census does this across decades does not appear to be consistent. But we do think, I think we’re pretty confident, that between 1940 and 1950 there’s an explosion of change.
And I should also say work disappears, right? The types of tasks that people used to do just are no longer done. We don’t have elevator operators anymore. We don’t have toll collectors. And so part of the dynamic is that old tasks are eliminated, and new tasks are created. There’s no economic law that says those things have to be in balance. And moreover, the people who are, you know, losing one set of tasks are probably not the people who are gaining the new ones.
And that’s true of so many changes. Same with, you know, if we talk about the effect of trade on labor market and China’s rise, right? It benefited a lot of people. It lowered prices. It allowed American companies to kind of produce a lot of products more cheaply. You know, it’s hard to imagine Apple’s growth without China, for example, to do all that assembly, which would have been extremely expensive to do in the United States. At the same time, it displaced a lot of people, more than a million, and in a very geographically and temporarily concentrated way, extremely scarring the labor market. Now those people also got lower prices, but that’s not even remote compensation for what they lost. And now there are new jobs — even in those places where those trade shock occurs — but it’s not really the same people doing them. It’s not the people who lost manufacturing work, right?
David Edmonds: So this occurs in the early part of the century when China joins the World Trade Organization and there’s an explosion of Chinese exports and trade with the US. And as you say, it was very localized, the effect. Now you might expect, if you were free market economists, that the economy would adapt very quickly, and that these people would move very quickly into new work, is that what happens?
David Autor: You might expect that, and I think a lot of economists did expect that. I think labor economists did not really expect that. And that is definitely not what happened. And the reason is labor market adjustment, the way people change careers, is really cohort by cohort, generational, rather than within a lifetime. Most people don’t go from, you know, this year I’m a manufacturing worker, next year, I’m a doctor. The following year, I’m a lawyer. Year after that, I’m a truck driver. People make their skills investments and career investments in young adulthood, and if they change careers later in life, many people are changing because the thing they were doing is suddenly not viable, and they often end up doing something lower paid that uses a more generic skill set. You know, if we stop doing what we’re doing, you or I could hopefully get a job doing cashier, checkout, waiting tables, doing lots of tasks that require common sense, that require language, that require physical dexterity, but aren’t very specialized. We would earn less money doing that work.
And that’s the problem is that most adults, when they’re displaced from a career job, end up in a lower-paid job. The change occurs when the next generation comes in and says, “Well, I guess I’m not doing manufacturing. I’m going to invest in something else.” Now you could say, “Well, isn’t this all inevitable because things change?” And there’s where the rate of change matters so much, because there’s a natural ebb and flow. Let’s say 5 percent of people retire out of an occupation every year. Well, that means in 20 years, you could replace everybody or go to zero without firing anyone. But if you did that all in the course of seven years, well, 35 percent of people would go and the other 65 percent you’d have to fire. So the rate of change is incredibly important, and that’s why shocks that are so concentrated in time are so much more scarring than ones that occur over a long period of time.
David Edmonds: And the Chinese shock, you would say was a very dramatic shock?
David Autor:. Extremely dramatic. US manufacturing employment fell by 22 percent between 1999 and 2007. We’ve never seen anything like it.
David Edmonds: Wow. So which particular industries and which cities were most affected?
David Autor: Sure. So it was labor-intensive industries; it was not high-tech manufacturing. So things like textiles and shoes and clothing, building materials, commodity, furniture, assembly of toys and dolls. And most of this manufacturing was in the South Atlantic, the Deep South. There’s some of it in the Northeast and so on. But these were not the frontier sectors of American manufacturing. They were big in the 19th century, they had often migrated over time towards lower wage areas. There were many, many women employed in the types of manufacturing that were displaced because they were over-represented substantially in textiles and garments. And it’s important to emphasize, manufacturing. – unlike, you know, kind of like drug stores and supermarkets, right? — is very concentrated in places. Everywhere you go you’re going to find, you know, whatever, a Wal-Mart, a CVS, whatever you need, manufacturing where it’s done, it’s done in a very condensed way, because there’s a lot of benefit to doing it in groups, because you have suppliers and buyers and lots of knowledgeable workers. And it’s not even just manufacturing, it’s specific products. So Martinsville, Virginia called itself the sweatshirt capital of the world, Hickory, North Carolina called itself the furniture capital of the world. You know, trade affects prices, right? When all of a sudden, the thing you’re making is now available for 40 or 50 or 60 percent less equally good, or almost equally good, even it doesn’t have to be equally good at that price difference, right? It’s not like you just say, “Well, we’ll cut our price.” Your work is no longer viable. Your business is going to have to close, and that’s what happened.
David Edmonds: And we’re talking now over two decades on, you mentioned Hickory, North Carolina, if I was to go there today, would they still be feeling the effects of that shock, or would people have now moved on to work in new industries?
David Autor: Well, it’s interesting, because both are true. So first of all, you would certainly see abandoned mills and so on. You will certainly find people who have hung on in those sectors, or people who’ve lost that work and are very angry about it. You’ll also find all kinds of new innovation, but it’s a different set of people. You’ll find many Hispanic Americans. You’ll find many immigrants. You’ll find more women working and fewer men working.
My colleagues, Gordon Hanson and David Dorn and Margaret Jones and Brad Setzler and I have a recent paper about we call it “Places versus People: The Ins and Outs of Labor Market Adjustment to Globalization.” And if you go to the places where these shocks occurred, they have rebuilt. They’re different. The wages are lower. They’re doing many more services, not manufacturing. The fraction of people who are present, who are working is a little bit lower, but essentially, they’ve reconstructed themselves in terms of doing healthcare, not the exportable kind, but care of the elderly and so on, some education, retail, warehousing, trucking, entertainment and recreation. But the people who are doing that work are not the people who are in manufacturing, and the people are manufacturing, you might think, “Oh, they’ve moved elsewhere for greener pastures,” but the opposite occurred. They’re more likely to stay. And so the places have substantially recovered, which is great. The people who were there at the point of impact have not.
David Edmonds: That’s fascinating. I have to be careful how I phrase the next question, because your president seems to change his mind on a daily basis, but the implication of his tariff policy is that America should be going back to producing what it produced 20-30, years ago. Is that the right strategy? Should Hickory, North Carolina be going back to making furniture?
David Autor: No, I’m afraid not. I understand and relate to the issues that the Trump administration is trying to address, right? So Treasury Secretary Scott Bessant said the American dream is not just about cheaper TVs, and I agree with that. There are more important things than simply having the lowest price of everything, especially if you’re not working, what do low prices really do for you? But it does not make sense any longer to try to produce labor-intensive, mostly commodity products in the United States, and we’re not good at that. Our costs are too high, and people don’t even want to do that work. You know, there’s half a million unfilled manufacturing jobs in the United States right now.
But more importantly, we have 13 million manufacturing workers. China has 120 million. We cannot compete with them across every front. This would be like Ukraine trying to surround Russia and attack it from every border. That’s not going to happen. What we should be very deeply worried about is losing the frontier sectors that we currently maintain. Those are threatened. So aircraft, telecommunications, robotics, power generation, fusion, quantum computing, batteries and storage, electric vehicles, shipping. These are sectors that we still have (except for shipping, actually) but China is making incredibly fast progress, and instead of trying to get commodity furniture back, we need to think about the current war we’re in, not the last war. And so that is my concern. They have the right question and the wrong answer.
David Edmonds: I wanted to focus on two shocks. There was the China shock, and there’s the AI shock, which is coming, or perhaps has already begun. You said something earlier about you don’t regard jobs as jobs, as it were. You regard them as a set of tasks. And I wonder whether that’s helpful when we think about AI.
David Autor: Well, I hope it’s helpful. And here’s why. Many people think that just being exposed to a technology is like as a risk. If you’re exposed, you know your job is going to atrophy, your wages are going to fall. But that’s way too simplistic, right? Think of the jobs of air traffic controller and crossing guard. These are basically the same job, right? An air traffic controller, crossing guard. They’re there to prevent collisions between things, right? Airplanes, people, children, cars, et cetera. One is highly exposed to technology, air traffic controllers. The other is not, crossing guards. Air traffic controllers make five times what crossing guards make where the world is in desperate need of more of them. They’re incredibly exposed to technology. And people say, “Oh my god, exposure to AI, that’s the new weasel word for you’re going to lose your job.” That’s not right. For some people, that exposure is really going to be a negative, because it’s going to automate away their expertise. Right? If you’re a language translator, if you’re an illustrator, medical transcriptionist, some types of computer coding, that’s really a threat. And why? Because you have a set of tasks. It’s taking the most expert ones and all of a sudden making them into commodities — too cheap to meter. On the other hand, if you are a doctor, an attorney, it actually says, “Look, you can automate some of these sort of supporting tasks that you do and then really specialize.”
But I don’t want to suggest it’s only like the super high-end people who benefit, people who run construction firms. They spend a whole lot of time doing paperwork that they hate, writing notice letters, writing, you know, personnel reports and so on. AI just says, like, we can just get that done. That’s not what we’re good at, and we don’t have to do anymore. And my hope is that if we use AI well, we will enable more people to do jobs that require judgment and knowledge without as much formal education. More people could do coding. More people could do skilled repair. More construction workers or contractors could do kitchen design. More workers with a bachelor’s degree in health services could do medical-technical jobs. So I feel like if we use AI well, it’s actually complementary to the knowledge that many people have. I mean, that requires design. It’s not a given.
So just to summarize my point, why it’s useful to think of tasks is you shouldn’t think that technology is either good or bad. You need to think about it in relation to the value of your expertise. And does it commodify it and make it such that anyone can do it now — good for consumers, bad for workers, right? Or does it make it more valuable? Because it eliminates a bunch of the stuff you aren’t really specialized to do, that’s not what you’re good at.
David Edmonds: That stat you mentioned earlier, about 40 percent of Americans having a four-year higher education degree. And I know from the ‘70s onwards, the gap between the highly educated and less highly educated increased, although it’s been narrowing in the last decade or so. What do you expect to happen to that gap in the AI revolution? Do you expect the gap to grow or to narrow?
David Autor: Good question. So it’s hard to forecast these things. Inequality has been rising strongly in the United States and most of Europe since, really, the early 1980s, although it plateaued over the last 15 years and even compressed during the pandemic. And in some high-income countries like the UK, the kind of college, high school premium or the college, non-college premium has fallen over the last 10 years. So there are elements of what I would think about. One is AI is certainly going to devalue some types of white collar work. On the other hand, we already have desperate need for people to do skilled hands on, work in electrical, work in plumbing, in construction, skilled repair, and also to care for people’s physical infirmities, because we’re getting older. So I think there’s going to be some rebalancing towards people who do skilled trades. Now that’s highly skilled work let me be clear. It’s just not as highly educated, because so many of the skills are acquired outside of the classroom.
Now, will AI eliminate the billionaire class? Absolutely not. There’s going to be people who are making extraordinary riches. But from the American US perspective, I don’t think we hate billionaires nearly as much as many other countries do, right? I think many people agree what we should be most concerned about is the opportunities of people who are not in the elite. You know, the 50 percent of Americans, 60 percent Americans don’t have a college degree, 50 percent of Americans are, by definition, at or below the median. Do they have economic security? Do they have jobs that are not demeaning? Do their children go to good schools? Do they live in safety? Can their children pursue the careers they want? Are they not constrained because their parents aren’t in in the right walk of life? So I think opportunity is a word that almost everyone agrees is something we should be focused on. It’s something I care about a great deal, and so I worry that AI could erode opportunity. I’m also cautiously optimistic that we can use it well to increase opportunity.
David Edmonds: AI is obviously going to be able to deal with some of the tasks that you’ve already mentioned. What. About the generation of new tasks. Is that possible to predict?
David Autor: It’s super hard to predict. We already see like, some examples of this, like, so people use AI a lot, for example, for ideation. You banter back and forth with AI to think about a new business plan, or even, what do you say on a date or for a job interview. People use it also for taking a messy data source and saying, like, help me see what’s going on here. People use a lot for coding, of course, but they also use it for tasks that involve even reviewing AI’s output and deciding, is it right, is it ethical? Think about, for example, the advent of the web created all these people who are website designers and developers. That’s not a highly technical job, actually at this point, but it’s a very skilled job. It takes a lot of design sense and capability, and so I think people will use AI to create new services and products that will be very complementary to specific skill sets. But it’s harder for me to predict more specifically than that.
David Edmonds: I suspect it’s the market that will sort those things out, and the government is not very good at predicting what AI is going to generate in terms of new jobs. If that’s right, what is the role for government?
David Autor: Yeah. I mean, I think the role for government in the AI era is a few things. One is regulation. Of course, that’s the most obvious, right? We should be concerned about risk and safety. We should be concerned about intrusive monitoring, non-consensual pornography, misinformation and deep fakes. That’s one thing.
The second thing is the government should be investing we’re in a real technology race with very, very fleet-footed competitors, and we need to invest in research. And the US is cutting right now at an incredible rate, the scientific basis of this country, in health and in science, from cutting the National Institute for Health, the National Science Foundation. These things have helped the US lead the world in these front areas since the 1940s. I mean, if you’re talking about making America great again, this is something America is great at. We’re really putting it at risk.
The other thing is, the government can steer the direction of these technologies in socially useful ways, right? So two areas where AI could be incredibly useful to us are in healthcare and education. Both these areas are incredibly expensive, and they have slow productivity growth, right? Because why they use tons and tons of highly skilled workers who aren’t getting faster at what they do, right? So we could use AI in those areas to make healthcare more accessible, more affordable, create better classes of jobs like nurse practitioners and, you know, medical technical workers who can do more things with this technology. And this is true in education as well. And look even in the US, which allegedly has a private healthcare system, 40 percent of all health expenditure United States is Medicare. That’s government money. A lot of education is government money.
So the government can say, “Look, we’re going to set up challenges for how do we redesign education? How do we redesign healthcare, to use people well, to make these services better?” And most of the service making it better is not pure automation. It’s creating better collaborative tools to make people more effective in doing this incredibly rich, incredibly hands-on, incredibly nuanced work. And so the forecast of healthcare will be automated, right? You know, the AI radiologist is a great example. There was a great newspaper article by Steve Lohr in The New York Times last week, and he said, you know, your AI, radiologists will not be ready to see you soon. And the point was, radiologists love AI. They’re like, wow, that’s a small part of what we do is reading those scans. Thanks for making that so much faster. Now we can, like, you know, spend more time consulting. We can, we can actually look at a broader set of diseases and diagnoses. We are more accurate.
And that should tell us that we’re thinking too simplistically about automation. Everything’s automation. Most things are not automation. Most transformative technologies. They’re not transformative because they make the same old things better, cheaper, faster. They’re transformative because they enable us to do things we couldn’t previously do. They give us new capabilities. We didn’t have software engineers 100 years ago, we didn’t have pediatric oncologists, we didn’t have kitchen designers. So much of the work that we do involves things we didn’t do 100 years ago, and it’s because we have created a much richer set of possibilities. And what makes them so important is not the code and hardware that goes into them, per se, but the imagination that it takes to think of those things. The computer that landed the first set of astronauts on the moon, right, the Apollo Guidance module has much less processing power than you would find in your phone, much less processing power than you would find in my washing machine. What made it so impactful was the imagination to say, “Wow, with this digital technology and this calculating capacity, we can do a task we’ve never been able to do, which is guide a spacecraft onto the moon.” And I think those moon shots are where so much of the potential lies.
David Edmonds: So I have a side hustle in podcasting. You, a professor. Which of us is in greater danger in the face of the AI onslaught, and which of us is in a better position to take advantage of it?
David Autor: It’s a hard question. I think we both have to adapt. I think podcasters, they’re some of the most personable, the most interesting, engaging people. They are hugely influential, right? I think Ezra Klein in the United States has a cult following. We didn’t have a way of doing that for one person, right? You required a huge organization for that, and that is a testimony to people’s creativity and the kind of charisma and smarts and insight that people have. And I think that’s always going to be in demand.
Are professors threatened? I mean, at some level, yeah, of course. I mean, first of all, if you cut all our research funding, that’s not going to make our lives any better. Also, you could say, “Look, maybe we won’t need them as much in classrooms,” although I don’t think that’s true. I will say that for me, AI has opened vistas of research opportunities, and not just to study AI, but to use the tools of AI to do things I couldn’t do. I mean, when talking about your moonshot, I couldn’t work with text as data 20 years ago, I could not take books and turn them into data sets that I could analyze. I’m so excited. I feel incredibly fortunate. I’m 60 years old. I was there at the first computer revolution, right? I was learning to program mainframe computers when I was in middle school because there was a college nearby, and I was so fascinated by this, I didn’t get to think I’d get to see something this transformative again, four decades later, five decades later, and it is really different.
And of course, there’s a lot of uncertainty and there’s risk, for sure, I don’t think most of the risk is labor market risk, as much as political risk, weapons proliferation risk, misinformation risk, but you’ve got to think that some of our most important problems in terms of climate adaptation, energy generation, malnutrition, inefficient farming, making education, healthcare more readily available, these are much more within reach than they were even 10 years ago, because we have better tools.
I’m not an AI doomer. I’m not an AI utopian either. I think one needs to recognize that very good and very bad things are going to happen because of AI and because of humanity, and so one should be highly optimistic and also quite concerned simultaneously.
David Edmonds: David Autor, thank you very much indeed.
David Autor: Yeah, this was a pleasure. Thank you so, so much for having me in a great conversation. It’s an honor to be on the podcast.
