David Autor on AI’s impact on jobs, expertise, and labor markets
01:03:20 Share

Everybody knows AI could be used for all these great things. Everybody also knows it could be used for really terrible things. And your belief about what's going to happen is not really a belief about AI. It's a belief about what humanity will do with this opportunity. Will we squander it or will we make the most of it? The future is not a forecasting exercise. It's a design exercise, right? We're building it. It's a matter of making good collective choices.

Hi, I'm Reid Hoffman. And I'm Aria Finger. We want to know how together we can use technology like AI to help us shape the best possible future. We asked technologists, ambitious builders, and deep thinkers to help us sketch out the brightest version of the future. And we learn what it'll take to get there. This is possible.

In the early 2000s, David Otter and his co-authors documented the China shock, or the rapid influx of imports from China that reshaped U.S. manufacturing, hollowed out middle-skilled jobs, and contributed to deep regional disparities in income and employment.

Today, we may face a new AI shock where advances in machine learning promise to automate a broad swath of tasks. From data entry to diagnostic imaging, potentially upending whole sectors. The pressing question is, will this shock exacerbate inequality or can we steer it towards widespread opportunity?

To help us understand the parallels and the pivotal differences between those two seismic shifts, we've invited David as one of the foremost experts on technology and labor markets.

David Otter is the Ford Professor of Economics at MIT and co-director of the Work of the Future Task Force. His landmark research on the China shock has become foundational for policymakers grappling with globalization's labor impacts. And now he's turning his analytical lens towards AI's coming wave. We sat down and talked with David about AI, jobs, expertise, automation versus collaboration, and labor markets.

Here's our conversation with David Otter. All right, David, I am so excited because oh, so many years ago, I was an economics major in college and my life diverged and my path not taken is being an economist. And so you are a huge celebrity for me. So this is great. And we have another thing in common because

When I was in high school, I used $700 to buy a 1983 stick shift Toyota Camry. Oh, okay. And I have it on good authority that you perhaps, in between college graduation and sort of what was next, you drove across the country. And I'm going to, I don't know anything about cars, so I'm reading this right off the sheet. In an eight Ford Speed stick shift 1980 Dodge Colt RS, they cost you $250. Wow.

That's right. I spent one third of what you spent on your car. I know. Well, you were frugal. So I want to ask, what did you learn about this cross-country trip?

Oh, actually, it kind of has set up the rest of my life, believe it or not. Because I was, you know, I finished college, I'd studied psychology, and I'd done computer science separately just as a concentration. I'd worked as a programmer, and I wasn't very satisfied with either career. I liked the question of psychology, but I didn't really love the methods. And in computer science, I loved the methods, but I didn't really like what I was working on as much.

And so when I finished college, I really didn't know what I was going to do with my life. And I went on a seven week cross country trip in this little roller skate of a car with my girlfriend. And we were at the big top Chautauqua in Minnesota and listening to NPR. And they talked about this computer learning center opening up at a black Methodist church in San Francisco called Computers and You.

funded by kind of Silicon Valley folks who were trying to bridge what they called the divide at that time. And I thought, oh, that sounds cool. I'm going to show up and volunteer there. So I showed up and volunteered there. And then I ended up as the director of education for three years. And that kind of got me interested in technology and work and inequality. So that was the main thing I learned on that trip.

I love it. That's awesome. That's very cool. Let's cut to some of the very current and very important things. Part of the reason why we've been looking forward to this podcast for, you know, some time now. So let's start with your China shock work, which revealed how import competition hollowed out U.S. manufacturing regions and over decades. Now, you know, we may be entering what some might call an AI shock. So what,

What are parallels and differences between a wave of low-cost goods, flooding markets, and a wave of cognitive automation reshaping tasks, jobs, industries? So let me put this in context by talking a little bit about the China trade shock. So the China trade shock refers really to two different things simultaneously. One is China's incredible economy.

growth, you know, as a function of its own changes in policies. And that starts in the eighties, continues to 1990s under Deng Xiaoping. And, you know, China just has this incredible productivity growth. They adopt, you know, Western techniques and technologies. They allow foreign direct investment. They allow hundreds of millions of people to move from unproductive rural agriculture into export processing zones and

And so that is a tribute to their own talent and turnaround of a country that had been really in quite a bit of trouble for quite a while. And then in 2001, they become a member of the World Trade Organization.

They also get permanent normal trade relationships with the United States in 2000. And that induces a really a surge in exports to the United States because of falling tariffs, but it changed investment environment in all kinds of ways. China actually became more competitive as a result of having to reform and open because of the WTO. And this had many benefits, lowered prices. It was great for China. It was great for many, many countries. It

It had a very severe impact on U.S. manufacturing and labor intensive U.S. manufacturing, particularly in the south and southeast. So furniture manufacturing, textiles, clothing, doll assembly, a lot of labor intensive, not the high end manufacturing, not mostly cars or airplanes, electronics manufacturing.

And this, in a very short order, caused a loss of more than a million manufacturing jobs. Now, a million is not that large in a labor market of 150 million people. And so if this were distributed evenly across U.S. counties, it'd be a few thousand people per county. You wouldn't necessarily notice it.

But that's not the way manufacturing works. It's very regionally concentrated and not just, you know, there's manufacturing here, but where manufacturing occurs, it's specialized, right? There was a town that was called, you know, they called itself the sweatshirt capital of the world and another town that called itself the furniture capital of the world, you know, Hickory, North Carolina. Right.

So they concentrated on doing one thing and did it well. And all of a sudden, their work was just non-viable. And many big factories closed in very short order. The period from 2001 to 2007 was when this occurred. 22% of all U.S. manufacturing employment was lost between 1999 and 2007. And then cumulatively about a third once we go into the Great Recession.

And, uh, and so this was just a, uh, incredibly concentrated kind of, uh, loss of jobs and loss of the viability of entire towns and, uh, the industries they were built around. And so it was experienced, uh, you know, as extremely scarring, uh, by the people in those places. And, you know, 20 years later, those places have kind of rebuilt. They're really quite different, but the workers who were initially in those locations actually have not moved on. They've not kind of moved up or out, uh,

And many of them are still in relatively low-paid manufacturing or other low-paid work. So it's been a very, very difficult, challenging, scarring transition. Okay, so that's the China trade shock. So when we talk about the AI shock, it's useful to draw that analogy, not because I think the analogy is exactly correct, but because it's instructive to think about the differences, right?

So, you know, one thing that they have in common, of course, is it could happen quickly. It's not so much a policy choice, although maybe China's trade WTO accession was, but it will affect a lot of people potentially really rapidly. I think there are three very important differences. One is, again, that regional concentration. We don't expect the impacts of AI to have anywhere near that type of local impact.

For example, we've lost more clerical jobs over the last 30 years probably than we have lost manufacturing jobs. But no one talks about the clerical shock. Why not? Well, one reason is there was never a clerical capital of the United States. That wasn't the way it worked because there were clerical workers in every industry. Similarly, AI, it will affect jobs and roles and occupations, but it will not affect resources.

regions in the same with the same degree of of concentrated impact. So one is it's not going to have the same regional component to trade shock really impacted the viability of industries themselves. Right. You know, all of a sudden it just wasn't competitive to have a U.S. you know, commodity furniture manufacturing industry that was making furniture for Walmart and Target that was now all moving overseas. Right.

And AI will much more, again, affect specific occupations, lines of work, the way work is organized within occupations. But it won't wipe out entire industries. Or if it does, I can't think of many that would be in that category. So it will not be as kind of holistic. And then the third thing is the time trade shock was experienced by U.S. firms as a pure negative competitive shock. All of a sudden, they couldn't charge the prices they were charging. Someone else was charging much less.

And so from a firm perspective, this was all bad. AI will be experienced by many firms as productivity increasing.

So it may still lead to displacement of workers. I don't, in fact, it will, I don't want to suggest it will not, but it will have a very different, you know, texture. It will not be perceived as all bad news. It will be perceived as rapid change. But some of it will be firms saying, well, we've got additional efficiencies. We can offer more services. We can offer lower prices. They may still shed workers. I don't want to say they won't. It will not be in any sense, a repeat of the giant trade shock.

One of the things you talk about, which I think is interesting, is everyone's sort of obsessed with jobs. Like, will AI create jobs? Will it displace jobs? How will it be? And you're saying, listen, we actually have a shortage of workers right now. We have plenty of jobs. The real question is actually about wages and income inequality. And what does technological change do to that? And so historically, technological change has sort of the, it's been skill-based, raising returns to education. And so the question on AI, like,

Could AI usher in an era of task-based change instead where like the premium flows to say emotional intelligence rather than former credentials? Like how could AI reshape this? And is there a positive scenario, negative scenario? Because we don't want to have the inequality that we had in the past.

So first, let me agree with what you just said. We're not running out of jobs. We're much more running out of workers. And this is true in most industrialized countries where we have low population growth, low birth rates, and now in the United States, heavily, heavily restricted immigration.

And that creates real challenges, not just because it's hard to find the workers, but it also means you're going to have a large retired population that's expecting, has earned the right to a decent standard of living in retirement. And you need workers to support that, to provide that financing. So that is not my concern. And we've had, you know, we have very tight labor markets. We have low unemployment rate and we have, you know, for quite a while now, for more than a decade, the greater concern is the value of skills. People are paid to a substantial extent, you know,

for expertise, their know-how in specific activities, right? In the, in, you know, rich industrialized countries, there just isn't much of a value to just pure physical labor anymore. And by expertise, you know, I mean know-how and let me, and I don't want to equate that with schooling because there's lots of expertise that's gained, not through schooling, some schooling, hopefully some is gained through schooling as well.

Expertise means like how to bake a loaf of bread or code an app or diagnose a patient or remodel a kitchen. These are all valuable forms of expertise. But for expertise to have market value, it needs to have two things. One is it needs to produce a service that people value, right? So it's got to be data science, not card tricks. The second thing is it needs to be scarce.

Because if everyone is expert, no one is expert. It just won't pay very much. And this is the threat that automation sometimes poses, is it can devalue a skill set very quickly, not because no one needs the skills anymore, but because the machine can do it better, cheaper, faster, right? So we've seen that when phone-based routing changed the value of taxi services in London, where people used to work from memory. It used to be that...

Touch typing was a very valuable skill, not so much anymore. And a lot of mechanical skills used in manufacturing have been, have lost value because either that work is done overseas or because it's automated.

And this is the concern that one could have. And actually, let me even say this more broadly. Over the last 40 years, computerization has led to a lot of hollowing out of both production work and office work, right? There was a lot of skilled work.

following codified rules and procedures that required expertise and knowledge and practice. And because those rules and procedures were well understood, it was feasible to turn them into computer code and have machines execute them.

As it has happened, it's not that we've run out of jobs in the interim and nothing like it, but a lot of the people who would have been doing office work and manufacturing work four years ago now find themselves doing services, food service, cleaning, security, entertainment, recreation, hospitality, home healthies. And that's socially valuable work, no judgment there, but it's not expert work. Most people can be productive at that with very little training or certification.

And that means it won't pay well. So, you know, I like to give the example of, you know, you know, think of it like a crossing guard versus an air traffic controller. You know, at some fundamental level, those are the same job, right? The job is to prevent, you know, collisions between, you know, people and vehicles or vehicles and other vehicles. And yet crossing guards make less than a quarter of what air traffic controllers do. And, you know, they're protecting our children's lives when they go to school in the morning. So they're doing valuable work.

But they're not going to be paid a lot, unfortunately, because there's almost no training or certification required to become a crossing guard. So if we suddenly ran short on crossing guards, we could get the air traffic controllers to go do that in the morning. But if we suddenly ran short on air traffic controllers, as we are now doing, we could not get crossing guards to do that job. Google Gemini here. The U.S. has faced a continual shortage of air traffic controllers over the last decade, a trend marked by persistent understaffing and declining numbers in many facilities.

This ongoing problem stems from a confluence of factors, including past hiring constraints and disruptions like government shutdowns and the COVID-19 pandemic, coupled with a lengthy and difficult training process that sees high failure rates.

Additionally, attrition due to mandatory retirement ages and the demanding nature of the job, along with inefficiencies in workforce allocation, have exacerbated the shortage, leading to concerns about controller fatigue and flight delays. So expertise is really critical. And so the threat that rapid automation poses, to the degree it poses a threat, is not running out of work, but making the valuable skills that people have highly abundant so they're no longer valuable.

I mean, there's a whole set of different questions here that I think are really interesting. But one of the ones I actually want to start with is something I've been thinking a lot about this, which is you say, well, okay, the question exactly as you say, that it's really important that we, you know, kind of like for the quality and economics of work is that it's valuable to the market and that it's, you know, relatively rare to, you

you know, kind of uncommon at least in order to kind of command a premium in kind of economics and wage. Have you seen any work or speculation or theorization that you think is interesting around how AI can help people acquire those skills? Like being adaptive to it, like learning new skills, being in kind of new areas because AI

You know, most of the analysis that I've been running across has been, well, look, you know, coding is an important thing and AI is going to be doing a bunch of coding and you take a zero sum amount to the amount of coding there is. And you kind of predict something, you know, challenging. All of which strikes me as has a bunch of assumptions in it that are.

Certainly not necessary and maybe steerable. So I'm curious if you've seen anything that's, you know, things that we should be paying attention to, things that we should be trying to emphasize, things that we've been learning.

So first, let me agree with you. You know, there's a lot of thinking in the world that goes of the following, you know, let's look at people who are exposed and kind of if you're exposed, you're hosed, right? Like, that's it. Your work is going to shrivel and shrink and die away. But of course, we know that's not true from, you know, like, think of the air traffic controllers versus the crossing guards. Who would you rather be? Right? So.

In many, many cases, technology actually makes us more effective and valuable. We would be worthless without the technological tools that we have. I couldn't do my work. Air traffic controllers couldn't do their work. And if you're a doctor, if you don't have a stethoscope, you're much worse off. So in many ways, our technologies are complex.

complementary to our expertise and they are amplifiers or force multipliers for our expertise because they shorten the distance between intention and result. They give us superhuman powers to see things, to do things that we could not do with our bare hands or our naked eyes and so on, or even do with our cognitively. We couldn't compute them fast enough. I do a lot of statistics. My computer inverts matrices in milliseconds. It would take me months if I even remembered how to do it. So we should be thinking about

How do you get to be on the right side of this equation? Right. Does technology, is your technology going to make your work more expert or is it going to basically displace the valuable expertise that you have? And for different people in different roles, those will have different answers. And so let me go back to the question you asked, Reed, which is, you know, what about people acquiring expertise? This is super central, right? So, you know,

The argument I've been making is work that pays well is decision-making work, right? Where you actually have to, where the stakes are high, it's a one-off choice, right? How to land this plane, how to care for this patient, how to remodel this kitchen, right? Even, you know, how to season this, you know, meal at a restaurant. And there aren't simple rules. If there were simple rules, it'd already be automated, right? So it actually requires a lot of discretion.

And the problem is a lot of that work is done by highly educated people, right? So, you know, the people with BAs and MDs and MBAs and so on, they kind of monopolize the commanding heights of the modern economy, right? Whether in medicine or in law or in design or in, you know, education, technology and so on. And a lot of the people who utilize

used to do valuable work in offices and clerical offices and factories, right? They're kind of been pushed into generic work that's not expert. And the good scenario would be one where we were able to use AI to support people to do more valuable decision-making work, both to use their expertise more effectively and to acquire it more efficiently. And I actually think that's one of the great challenges of our era is

is to figure out how to create tools, AIs, that support people using their expertise better and learning faster. And I think that's very hard because the opposite can occur, right? If you rely too much on technology, you kind of won't bother. And you'll just say, well, it'll do the job for me. And so I like to distinguish between what I call automation tools versus collaboration tools.

Right. So automation tools are tools, you know, like, you know, like my, the automatic transmission on your car or the elevator that goes between floors or the toll taker that when you drive through a highway toll takes, takes money. Right. These are examples of successful automation. And those all used to have be jobs, right? You used to shift your own car, right? I know my Dodge cult, the eight speed car, uh, you know, that had a stick shift. Uh, there used to be elevator operators and there were lots and lots of toll takers. Right.

And this is successful automation. All of the specialized knowledge that was required is now fully encoded in machinery. It's done. We're happy with that. There's no problem. Most tools are not that form. Most tools require you to bring some expertise to the table to use them, right? So stethoscope, great thing. No use to me. I wouldn't know what I was hearing, right? Chainsaw, good for lumberjack. Not so good for my children. And most tools...

They're good because they allow you to take some knowledge and capability you have and do it faster or further or better. And so if we use AI well, we'll be using it a lot as a collaboration tool.

to enable us to make better decisions, to do harder tasks, to solve problems. And I don't just mean in research or whatever. I mean, you're an electrician. You go out to a site, you encounter an unfamiliar problem, you use your AI, and it pulls up the relevant information for you and helps guide you to do the work. You shouldn't do that if you're not an electrician. Don't go opening up fuse boxes. But if you have the basic skills and then you have a tool that can support you, you could probably go further with that. And so...

That's the way that would be successful use for collaboration in many, many cases. And again, I'm not morally opposed to automation. If you can automate something fully successfully, great. In most cases, we can't, right? There's actually an illusion, a kind of hubris that, oh, we have technology, it's superhuman, therefore expertise is dead, it can do whatever we want. You know, Jeffrey Hinton famously predicted about a decade ago that we would need no more radiologists.

And because it's so obvious that AI will just do this better. Now, AI is now used by radiologists. They love AI, but they're not fewer radiologists. They just do more of what they did. They're more useful because they have better tools. And a lot of their work is not just looking at scans, right? It's all of the communication with the patient and the other caregivers and so on. So it's a mistake to think

That everything, because we have good technologies that we can automate everything. There's a problem with thinking you can automate when you can't. You're going to design badly. So automation, and it's not just automation, can either increase the expertise of your work by eliminating the supporting tasks and allowing you to focus on what you're really good at.

Right. So I don't spend time inverting matrices. I can just work on what the statistics mean. Or it can de-skill your work by automating the expert parts and just leaving you with a sort of last mile. Right. So both are possible. Now, it also often creates new work that requires new forms of expertise. But there's it's usually different people who are doing that work. So in terms of using the tool, we should be thinking about, well, where will expertise be needed? Where will be displaced? Right.

And how do we enable people to do expert work with better tools, right? People who might be shut out, like if a world in which more people who don't have a four-year college degree can do software development, can do some legal work, can do medical technical work, can do kitchen design, right? That's a better world in my opinion. And then finally, how do we create tools that enable people to get better at that stuff faster?

You ended with the word faster, which I think is really apropos because the thing I think, one of the things that scares people the most about AI is time. They say it'd be one thing if this was happening over a hundred years, but it's happening so quickly. So I have a two-part question. One, I thought you really sort of

illuminated why speed can be tough for me when you talked about the difference between an occupation disappearing over 20 years versus seven years. So I would love to sort of hear what that difference looks like in terms of unemployment, et cetera. And the second thing is sort of from hearing you talk about expertise, it would make me think like, okay, the superstars are going to get even better. Like the, you know, the people who are educated are going to be even more enhanced by AI. But you've talked about how AI could potentially

potentially be good for the middle class? Like, how can we think about that? Is it because we can retrain workers more quickly? Like, how can this tool actually be good for the middle class? So let's first talk about the speed. So, you know, labor markets have a natural rate of adjustment. If you think a career is 30 years, let's say, that means kind of 3% of people will retire out of anything every year, right? So, you know, if you wanted to eliminate

a third of people without laying anyone off, you just wait a decade and they'd all be gone. And most labor market change for adults, it actually doesn't happen mid-career, right? It's at the choice of the entry point. And so it often occurs across cohorts, right? So in the areas that saw these big China shocks, right, the adults who were manufacturing

have not primarily moved on to something else. Some have moved into lower paid services. It's their kids who never enter manufacturing, right? So it really does matter how fast this goes on. You know, we talk about autonomous vehicles all the time, right? If autonomous vehicles are

come Labor Day this year, replaced all long distance drivers, right? That would be a very serious problem because there are more than, you know, 2 million, I believe more than 3 million, you know, people who just do their living in driving vehicles. So that would be catastrophic, not because autonomous vehicles wouldn't be a good thing, but because that would be so much job loss all at once. If it happened over 25 years, right, that's kind of a manageable problem.

Right. People wouldn't enter the occupation. People would retire out of it. And it will happen actually much more slowly because even if this problem were solved tomorrow morning, right, it takes decades to replace all that capital. Right. You're not just going to throw away all your trucks. Right. You're going to replace them slowly over time. And so the concern with AI, I think that a lot of people have is it's just going to boom, you know, and it's absolutely the case that machines can acquire skills much more rapidly than people can. Right. Once you have a machine that does something, wow, then you have a lot of machines that do the same thing.

And there are, we will see this. I mean, we should not be naive to think that won't occur. If you're a language translator, you're under threat. If you're an illustrator, you're under threat. I think a lot of people who do just sort of workman software coding, there will be fewer

I actually, it's not completely certain, but you know, we do see a big decline in employment in computer coding right now in software development, not our site. I shouldn't say in the software engineering, but in, in the people who write programs. Right. And so it's quite possible. Now I think we alluded to this earlier, um,

It also depends on how much, what demand looks like, right? So like if we, you know, got like really, really good and cheap and fast at colonoscopies, people still wouldn't be lining up with their proctologist's office to get more of them, right? But it is the case that if we get better, cheaper, faster coding, right, there's a lot of demand for software, right? Like you can't buy an appliance that doesn't have a microprocessor and it doesn't have embedded software running in it. You know, that, I mean, I'm sure my toaster oven has that.

that. I know my coffee pot does because it regulates the temperature and tells me what it is. So it may be that we'll just get a lot more software coding, but it may also change what we use it for. So when people start developing websites back in their mid-90s, that was all about skill and HTML, writing markup language. Those websites, if you go back and look at them, they're so incredibly horrible looking. It's hilarious how primitive they are.

Now, there's lots of people who build websites for a living, but it's not really a technical skill. It's design, right? It's how do you present information? It actually has a different skill set that's involved. So that's kind of cool, actually. So the rate of change is a concern. It absolutely is. Okay, so now let me come to the last part of your question. You say, well, how could this be good for anyone? Well, this is not a given. This is kind of a good scenario, but

As I argued a few minutes ago, a lot of valuable work in advanced economies is monopolized by elites, right? Professors, for example, or lawyers or medical doctors or financiers. And we don't face that much competition.

Right. You know, there's huge barriers to entry. There aren't that many of us. You can't print them that fast. And and so, you know, we have this, you know, lock and, you know, that's fine for us. Like, yeah, things are expensive. Health care is expensive. Education is expensive. Legal service expensive. So but that's OK, because I get paid a lot, too, because I do that. Right. But for most people, they don't do that. It's just expensive. Right.

So if we can enable more people to compete in those domains, if we can enable more people to enter, you know, software development, more people to do medical services, more people to be lawyers. Now, it doesn't I don't mean that everybody can do anything. And I don't mean that they you know, we don't need doctors anymore because people have A.I.,

But you could imagine more supporting roles for people who don't have as much elite education to do that type of valuable work. In other words, just instead of, you know, the last wave of technology pushed so many people out of the middle and down towards low paid services. It would be great if we could use this technology to enable more people to move up into new opportunities. That's the good scenario.

Now, have you seen anything that kind of is how...

other than just getting people into this, how they should start learning these tools in order to start kind of beginning to learn their own paths on what the new contours of how jobs will be transforming, how markets will be transforming, how industries will be transforming. And is there separately anything that you've been looking at as kind of a macro thing to say, you know, here is like a, you know, kind of call it a 30,000 foot or a 40,000 foot map

by which people should start thinking about how these transformations are going to happen. Learning to use AI well actually is an important skill in itself. And it is, you know, the first, like the first instinct that you have to develop. I mean, I remember, you know, I'm old enough. Like I remember a world before Google, uh,

and I remember developing the instinct to Google things, right? You'd be at a dinner party or whatever, and you'd say, oh, I thought President Taft said this, and I think someone said that. And then someone would say, well, why don't we Google it? And everyone would be like, oh, Google it. And then we learned, and then we sort of developed this kind of,

a subroutine or an instinct or whatever, this habit to turn to it. And knowing actually when to turn to AI for things is also like not obvious. It's something you get better at. Like, for example, I was talking with some of my research assistants and we were like looking at this table and say, could this be made into a figure? And, and we're trying to say, well, would it work as a figure?

And we would say, well, what if you did this way? And then I said, oh, I know. And I just took a picture of it. I stuck it into an AI and say, make a bar chart of this. No, now rearrange it like this. And did it like that. Of course, five minutes, we all did on the screen. And we said, yep, this won't work. But that was a very time efficient way to do it. And so I always have a chat window open. And I use it to sort of bounce ideas or look things up and so on. But another important thing to realize about AIs, and this goes to this distinction between collaboration versus automation, AI isn't really useful for things that you don't understand.

because it's not that reliable and it will misunderstand for you and lead you astray. So if you're using AI to do something you really don't get, you're kind of out over your skis and that's not a good place to be. So that's why I say it's a good collaboration tool because it's complimentary to, you know, if it's something you know about, then you can adjudicate, oh, this makes sense. This doesn't make sense. You can ask the right question and you can kind of, you know, filter and interpret that knowledge.

If you're trying to get to do something for you that you don't, you know, write me a paper about, you know, the currency of the Roman Empire or something like, you know, it might very well make up some Roman currency you've never heard of and you wouldn't know. So that also is part of learn to use AI is learning the instinct of when you know enough to know if it's doing something useful.

And when you can use it, you know, so you've got to have, you know, it's like I could tell an AI, you know, tell me how to do a surgical procedure on someone. I shouldn't do that.

Right. If I'm a surgeon, right. And this is an extreme search scenario and there's no other surgeon around. I can like look at the AI, give me instructions and I could probably do it. So you want to use it in the domains where it can collaborate with you and augment you, but it can't automate away substitute for, you know, just fundamental lack of knowledge in some area. That's a dangerous place to be.

So sort of thinking about the possible future we want to shoot for, like you're talking about AI being complementary. And that's one of the visions that obviously Reid spoke about in Super Agency. It is, OK, how can we have more people sort of moving up and being able to do before they could be a nurse, now they can be a doctor. Before they were a radiologist assistant, now they can be a radiologist because they have these additional skills. And it's actually, we work a lot with this organization Opportunity at Work, which is trying to increase the

returns to skill-based work as opposed to just degrees. If you have the skill, you should be able to get the job. So that's sort of one vision that in the future, everyone who perhaps is in a low wage job right now, or not everyone, but some of those people can move up into middle wage jobs. How fantastic would that be to add, you know, $20,000 to each job?

Another sort of positive vision, actually, for AI that some people talk about is that less people will be doing jobs. Like we will have so much money, our government will have it or the big AI companies will have it or someone will have it, that these jobs are not necessary and we're able to give out a UBI or we're in this new world of sort of freedom and only working 10 hours a week, which sounds a little fantastical. So I would love you to comment on sort of those two visions, both of which are sort of positive as possible.

positive visions from sort of the new AI future. Yeah. So the first vision, of course, I'm all behind. And it doesn't need to be, by the way, that everybody is doing middle or high, you know, skill work, right? So if we just extracted half of the people who were doing, you know, leisure and hospitality and janitorial services from that work, the half who remained would get a big pay increase, right?

Right. Because firms would have to compete for the more. The problem is there's too many doing it. And that's why wages are so low. And there's a well-known, you know, kind of economic parable, right? Like why do the wages of barbers rise over time? They're not getting any faster cutting anyone's hair.

And the answer is, well, they have to be compensated to be barbers as opposed to being something else. So if there's, in the long run, if productivity rises and there aren't lots of people, if you will need to convince someone to do food service or cleaning, whatever, and they may want to, they may not. But if you want to, in a competitive market, if there aren't that many people available, you'll have to pay them more. So it doesn't have to be everybody. Just more opportunity benefits a lot of people.

including not every, and they don't have to all take that opportunity benefit. On the notion of, you know, we'll have all this income and therefore have all this leisure. The concern I have with that is not that we won't have wealth, but they will have a lot of trouble distributing it equitably.

You know, we are already an incredibly wealthy society, right? We're arguably the wealthiest society humanity has ever seen, and we don't really have any real scarcity here. And yet we have a lot of people who are quite poor and don't have access to health care and don't have a safe housing, don't have safe neighborhoods, don't have good schools. Right.

that's not because those resources don't exist. It's that we don't have a system where people that we really want to distribute that much to people who don't somehow earn it on their own through the labor market. And I don't know that more wealth is going to solve that problem.

And I really worry about a world in which so much income is concentrated. The notion that we are sort of reliant upon the generosity of strangers through the tax and transfer system to sort of, you know, you know, make us, you know, take care of all of us. I just don't know if that's if that's a reliable thing to do. I don't you know, it generally doesn't work that well. You know, the U.S. U.S. is not getting more generous as a society, even as it's getting wealthier. We seem to be getting less generous generally.

Uh, and, um, so that, so, you know, I, I like to compare two scenarios, what I call the, um,

the, uh, the Wally and the Mad Max scenario. So you've all seen the movie Wally and, you know, it's a future where basically people, you know, sit around on, you know, kind of hovercraft armchairs, uh, watching, you know, holographic TV, uh, drinking big gulps and they all weigh 300 pounds. Right. And this is supposed to be some sort of future dystopia because there's no work to do and everyone's bored. But I view that as the good scenario, uh,

Because the more likely scenario to me looks much more like Mad Max Fury Road, where everybody's competing over a few remaining resources that aren't controlled by some warlord somewhere. So you can have a world that's very wealthy and yet most people don't have anything. And so that's why I like to think about work because I actually think the labor market has so much going for it. Two things. One is it's intrinsically a lot more

equitable or equal than the capital market, because everybody in a society that doesn't have slavery and doesn't have labor coercion, everybody owns no more than one worker, right? They just own themselves. And so we all start off, you know, at a kind of a, at a, a relatively even starting point. And so, you know, 60% of the income in the United States is labor income that goes first to workers. And I, and so it just creates a much more shared, uh, resources.

Work also has a lot of virtues. I think, you know, it gives people identity, it gives them structure, it gives them meaning. I mean, not all jobs are good. So that's a luxury for me to say, you know, work is all great. I don't want to say that for a minute. But the other thing is, in a democratic society, if most people are working, then it's easy for people to say, well, these are people, all contributors to our society. Of course they have a vote, right? You know, they, of course they were the, were the co-owners of society, right?

Whereas if we're in a world where we say, well, like all the money, you know, comes out of a fountain in, you know, in San Francisco, you know, next to the, you know, open AI headquarters or something, then it's much harder to say that everybody deserves their share. I mean, I might, I might agree to that, but I don't know that everyone else will. So that's why I'm, I'm not excited about a world in which the resources are mostly coming from machinery and capital and everybody expects to be supported. Now, let me say it.

Important to emphasize, you know, we do much, much, much less work than we used to, right? So at the beginning of the 20th century, you know, 120 years ago, 125 years ago, people worked on average in the United States about 3,000 hours a year.

Now we work on average about 1900 hours of the year. Right. We now, you know, we've invented the weekend, you know, we have vacation and so on. Additionally, you know, people used to enter the workforce as soon as they were physically able. Right. You know, 10 years old and they would work until they died.

Right. And now, you know, our side people enter the labor force, you know, 16, 18, 20, 25, their PhD students, 40, 45. And then they retire when they have, you know, 20 years of health remaining. So we work a much smaller percentage of our healthy lives than we used to. So we actually have much more leisure. So we've handled that well. It's not that we've just become this kind of overworked society. That's kind of a myth.

So I'm in favor of that. I'm in favor of us all working somewhat less, or at least those who want to work less. I'd like to work more. But that's very different from there being no need for people to work and that they just hope that the society will care for them. That I'm not as confident in. Well, one of the things that I think this discussion highlights in a really interesting way is there's kind of two issues that I think get combined in the inequality discussion, right?

one of which I'm extremely sympathetic to and I think is very important to navigate, and one of which is complicated. The complicated one is like the, well, no, I want there not to be that much of a gap between me and you. I just think that a gap is an issue. You have two cars, I have one, and I think that's a real big deal, whatever that kind of thing is. And I tend to be

a little bit more, you know, how do we really figure out what the right mechanism of the gap is? But a lot of people, that's their real driving issue in their rhetoric, but I think it's more complicated. Now, the one that's not complicated that I think you're highlighting that is extremely important that we solve

is a notion of increasing quality of life across everybody and increasing, and quality of life isn't just like, oh, look, I can afford the cheeseburger. That's great. But it's also kind of like, do I have meaningfulness and control and respect and personhood and place within community and society?

And that kind of breaks down into two components. One component is, you know, kind of what are you learning to do that gives you kind of a unique position in your community and your workforce, you know, and your, you know, your kind of your tribal group. And, you know, that's part of where I, you know, super agency, trying to embrace the future, learn, become AI curious, you know, step into it, don't get forced into it, but like jump into it first and foremost.

But the other one's also reducing, this is the abundance thesis, reducing the cost of a whole bunch of things. So like, for example, part of how you increase quality of life is you say, well, one of the things that AI can create is a 24-hour, seven days a week medical assistant where very, very few people, even amongst, except when you get to super wealthy, actually have that in the US. Has anyone gotten good results

kind of sense of, of both this, not just the, the, the thing we're talking about, like the jobs and the transformation of the industry, but also on the kind of quality of services and quality of life that,

Based on, you know, because with AI, I can see not just a medical assistant that can essentially one for free for everyone on a smartphone, but a tutor, a legal assistant, a set of things, none of which I think will take away actually jobs. I think there'll be a whole bunch of coordination, you know, kind of amplification there. I mean, they'll take away some jobs, but they'll also create a bunch, I think, in this environment.

but in that kind of quality of life that can possibly come from AI based on helping people in all these ways? I mean, for sure. I mean, look, we, even people who are, you know, much, much less affluent in the United States, you know, in terms of the sort of physical quality of life is much higher than it was, you know, even 40 years ago, you know, people, most people didn't have air conditioning, right? They didn't have private transportation, right?

They certainly didn't have all the things that come from a mobile telephony, you know, and mobile phones aren't just about entertainment, right? They're about communication. They're about access to services and information. So there are many, many ways that there are a lot of things that our material standards of living have improved because we've gotten better technology for doing it. You know, so people live in larger houses. They have more indoor plumbing. They have more electricity. They have bigger TVs. And again, you know, more importantly, they have, you know,

cars, air conditioning. And I do think, yes, AI will make a lot more service can be used to make a lot more services less expensive.

And, you know, we use the web this way all the time, right? You know, the amount of time we spend, you know, waiting on phone queues or trying to buy airline tickets or dealing with banks and so on. It's, you know, it's actually a lot more convenient than it used to be. I remember buying plane tickets on the phone or, you know, waiting in line to see a bank teller.

So that's I think that's very important. But, you know, I don't think low prices are sufficient because, you know, what are the things that are really defining for, you know, kind of a good quality of life in addition to the work you do? It's do your kids have opportunity? Right. Are your neighbors neighborhoods safe?

Are these good schools? And also, equally important, do people get a fair shake when they start? My problem with inequality, I'm with you. I don't care if I have two cars and some people have 20 cars. I'm still okay with that. But it is the case that inequality can lead to dynasticism.

where the next generation is like one set of kids starts so far ahead of the others that, sure, even if their family got wealthy on the merits, then there's no longer anything close to fair chance in the next round. And so I think that's something we... And those things are harder to deliver just through low prices, right? Just good schools...

good opportunity, safety. A lot of that depends on basically having a level of affluence in your family and in your neighborhood that supports those things. Maybe we'll get better at that. But so that's the key thing. So what matters to people is, of course, economic security, like from day to day, can I pay my bills? Am I going to eat? Do I have a roof over my head? But then do I have a job that provides me stability? And then do I have a way to ensure that my family

is going to, you know, is going to do at least as well as I have done and get good opportunity commensurate with their own hard work. I do think what is important, actually, about a lot of this is uncertainty. So if you didn't have the uncertainty that your rent would go up next year, if you didn't have the uncertainty that someone would get cancer and so you would go into medical debt and you declare medical bankruptcy, like those aren't really sort of

price issues. Those are more uncertainty issues. And so one of the things that I think a lot of people talk about is declines in unions and collective bargaining. They leave workers more vulnerable to pass shocks. Obviously, there's also more uncertainty about your wages. Are you going to be fired? There are certainly some negatives about unions. Sometimes they lock in workers that we don't want, et cetera, but they do sometimes provide that certainty. So my question for you is, as we want sort of gains to be broadly applied, are there

institutions, policies, collective bargaining? Like, what are the things that we can put in place to make sure that these gains are broadly applied? Certainly collective or, you know, kind of worker representatives are, you know, there might have been a time when we had too many, but now we definitely have too few. And the decline of collective bargaining at this point, I think, has left workers very vulnerable. So I would like to see more of that. I would like to see

more investment in our schools, in our children, those things would make a big difference. And then I do think, you know, people in the United States have much greater level of economic insecurity than, than do other people in, in less affluent economies working the same jobs. And we tend to tend to think it's a, it's a necessary factor of life. And it just isn't. If you work on a McDonald's in, you know, in Norway or Denmark, or even in France, you're going to have vacation, you're going to have healthcare, uh,

You're going to have sick leave. And this does not have to be inaccessible or only accessible to people who are middle and upper class. And that's a choice that we make, but we tend to think it's inevitable and it's not. The U.S.

we're so impressed with ourselves that we forget to make the right comparisons about, you know, and this actually, and the other way too, people don't realize how affluent we are. Like, you know, people think that America has been in decline, but you know, our productivity has risen 30% relative to Europe over the last 20 years. We're much, much better off. Our labor markets actually work really well. So, you know, everyone's talking about making America great again, but you know, actually we're pretty, we've done a lot of great things. Um,

We should recognize that. I mean, we're also, you know, the U.S. is an incredibly innovative country, right? You know, AI comes from here so much of the community area or the internet era, but we have an incredible culture of innovation that has been so important to us. So we should recognize both our strengths and our good fortune, but also recognize we can learn a lot from other models. And it doesn't, not everything that is bad is inevitable. School shootings are not inevitable.

And people not having adequate health care, not inevitable at this level of income. Those are choices that we're making. I don't think we're making the right choices. So one of the things that I think, you know, is a kind of a question that we're going to we're going to see happening that will be a little bit different within the universe is, you know, previously kind of workforce transitions and so forth have happened to what I would think of as industrial timeframes.

which is, you know, part of like when you look at our educational system, it's like, you know, train, then deploy for a lifetime versus, you know, kind of a constant recycling. And I actually think that part of what happens as we get into the new world of AI is that you'll need to have more constant adaptation. It's part of the reason why my very first book that I wrote, Startup of You, was how we're all going to have to be more entrepreneurial and how we think about our work and our careers doesn't mean start businesses. It means, you know, how to do that.

And obviously, part of my optimist is thinking about how AI can help with that. Is there any kind of good work, good lenses, good principles for people thinking about now these transitions are, call it, within, as opposed to within a 30-year timeframe, they're within a 10-year timeframe?

And so that kind of moving along and adjusting is going to be one of the things that's going to be important throughout kind of a human life, a human career. And so have any principles or work been done so far in this? I'm convinced it's going to be very important.

Yeah, I'm convinced as well. I would say the work is early. Like, you know, like, for example, I think everyone who looks at AI says, oh, my God, there's so much potential here for education. It's amazing how little education has changed in the last millennium or so.

And even, you know, if we walked into a classroom today from, let's say, you know, we all left school 30 or 40 years ago or 20 years, we wouldn't be surprised by a single thing in there. Right. Or the way things are done, it's really not different. So I think we need to figure out how to use these tools to get better at class.

education, at teaching ourselves and learning new skills. And I think, you know, one of the things we know about, especially for adults making transitions is they learn much more successfully kind of experientially than they do going back to classrooms, right? It's shocking to me as a professor, but not everybody loves being in a classroom.

And, you know, we sort of forget this lesson over and over again. Right. So remember, like MOOCs, massively online open courseware, right, supposed to just everyone was now going to be a, you know, a botnet herder or, you know, an ethnomusicologist or whatever they wanted to be. And they really weren't very successful. And, you know, people don't talk about it much anymore. Why aren't they that successful? They were like simulated classrooms. Right. And

And it's like everything you hated about a classroom and worse, right? Like who wants that, right? At least a classroom is a social environment, whereas watching a video is not. So if we want to use these tools to make education better, we need to make what do better, what makes education effective, right?

And that is kind of engagement and immersion. So I certainly think that there's a lot of skills that people can learn in simulated environments, right? So, you know, like we do this, right? If you want to fly a plane, you're going to spend a bunch of time in a flight simulator. If you're, you know, learning to do medical procedures, you're going to start off on animatronic dummies that bleed and scream, right? Why don't we do that for, you know, plumbing and electrical work? Well, it's clear, you know, we do it in a few places because it's so damn expensive. We don't do it everywhere else, right?

even though we could. But we can now. That will be possible. So I think that's an important lesson. It's also the question of how do we get, there is a lot of research now on how should we interact with AIs? What should they tell us to help us learn? And one thing, the consensus that literature is what they should not do is just sit around telling us what to do, right? That's not how we learn.

And in some sense, they need to interact with us in a way. Like say, you know, there's one theory that says like, you know, if you're coming to a choice, you know, there's A and B. The AI could say, do A. But it could say, well, you could take A and this would happen or B and I predict this would happen. The contrast between those things is actually very instructive for learning. So if you want to help people learn, it's not sufficient to tell them what to do. You need to give them information first.

That supports that choice and enables them to reason about it. So there's a lot more work going on now about what are the ways. And I'm doing some experimentation this myself. I'm, you know, I'm a Google tech and society visiting fellow. And one of the things we're trying to do is stand up experiments.

on expertise and to ask whether tools that are built to support experts do they not they just help them get better results probably they they will but do they help them acquire judgment faster right because in so many you know if you're a lawyer if you're a doctor uh if you're a researcher if you're an actuary if you're a carpenter right you develop judgment over time that's part of the expertise that makes you so valuable right it's not just like what you learn from a book you

You learn how to make the right decisions at the right time, at the right moment, you know, and, and so we, I would like to understand better. What are the tools that enable people to do that faster? Get those skills sooner because of course it takes a long time expertise, you know, it's great, but you know, it takes a long time to get expertise. It's slow, it's expensive. And even the best experts are fallible and most people are not the best experts, right?

So, you know, so that so I think that that's the question we should be asking. And that's where the research is that I see is about how do you do these interactions in a way that makes people smarter? And I think kind of the key principle for me is automation versus collaboration. That, you know, automation is when you tell people what to do and that takes out expertise and collaboration is when you.

harness expertise, that you give people the information and, you know, and collaboration is, you know, like if two people, two experts look at the same problem and reach different answers, and then they put their heads together, the answer they come up with may not be the average of A or B. It may be C. It may be saying totally that they weren't considering, right? Two people collaborating could actually come up with a different answer than either of them thought originally, which is not something that can happen when a machine tells you what to do.

So I think this principle of collaborative design, to me, seems like very essential for thinking about how we help people learn and acquire skills more efficiently. We'll now move to a rapid fire. Is there a movie, song or book that fills you with optimism for the future?

So I gave this one a lot of thought and I came up with the answer. It's not quite in any of these categories, but it would have to be the tiny desk series of concerts on NPR. I don't know if you guys watched tiny desk, but Oh my God. So, you know, this has started like 20 years ago, Bob Boylan, who was their music editor started inviting bands to come play at his desk.

And they would just sort of film those. And over time, this has become an institution. So like major acts that, you know, so like, you know, Taylor Swift and not Bob Dylan, but, you know, almost everybody else you've ever heard of shows up. But the greatest thing about Tiny Desk is they're in an intimate space. So it's a small group and it's often people you haven't heard of. And the range of music that you encounter is so incredible. Things you would not see. Like, you know, I go see Anderson .Paak's Tiny Desk concerts.

or go see the time guest concert of the Buena Vista Social Club. These like, you know, 15, 20 minute productions. I look forward to this like so much.

And the amount of music that I've encountered that I never would have heard of that gives me so much joy. So Tiny Desk. David, what is a question that you wish people would ask you more often? You know, I wish people would ask more often about, and I ask people this all the time, like, what is your counterlife? What is the thing you would be doing if you weren't doing what you're doing? And that often tells you about something else that they're really passionate about. And, you know, in America, especially, we always ask people what you do for a living. What's your work? What's your job? And that so much summarizes people's identity, but counterlife.

kind of too much. So yeah, asking people their counterlife, I think, which people would ask me. Oh, so now you're going to ask me what is my counterlife? Yeah, David, okay. Yeah, let's hear it. You know, I think there's another world I could have been a sailor. I love to sail. I still sail. In fact, I'm in a place where I sail. And I could have imagined doing around the world sailing, doing racing, crewing. So that would have been okay.

Where do you see progress or momentum outside of your industry that inspires you? People in, you know, in the West, certainly the United States, if they look back over the last 20, 34 years and they go, eh, it's been mid, it's not been a great, you know, half century, but actually this has been the best 40 or 50 years that humanity has ever experienced, right? The amount of people brought out of poverty, right?

We've never had a global middle class until now. And partly this is China itself. China has reduced its poverty level from 70% to effectively a couple percentage points, and that's more than a billion people. But it's also created prosperity in Central and South America, in Sub-Saharan Africa. And in fact, growth in Sub-Saharan Africa has been really strong and a livelihood to improve. It's still very poor, but it's much less poor and health is better.

And so it is I think we overlook how much progress there has been in so much of the world. And so, you know, we don't appreciate our good fortune, but we also don't appreciate the good fortune of others. So I think that is overlooked progress. All right. Our famous final question. Can you leave us with a final thought on what you think is possible to achieve if everything breaks humanity's way in the next 15 years? And what's our first step to get there?

So if everything broke our way, if we really did this right, you know, we would make, it's not that we put ourselves out of work, but we would give people more secure and fulfilling work. We would give them more access to education and access to better healthcare everywhere. And those things alone,

would kind of improve welfare in so many dimensions, not just in terms of material standard living, not just in comfort, but investing in our kids, creating opportunity for the next generation. So I think that would be great. And it's feasible. I mean, many of these things are feasible. If we think we're not going to do them, it's not because we couldn't do them. It's because we're somehow not

delivering on what is feasible. And that's the kind of sad thing. Everybody knows AI could be used for all these great things. Everybody also knows it could be used for really terrible things. And your belief about what's going to happen is not really a belief about AI. It's a belief about what humanity will do with this opportunity. Will we squander it or will we make the most of it?

And I think most people, you know, I don't think anyone thinks we'll absolutely make the most of it. I'm sure many people think we'll totally squander it. But I think many people are really deeply, deeply uncertain. So, you know, in terms of breaking our way, this is, you know, as my friend Josh Cohen, a philosopher, you know, likes to say, you know, the future is not a forecasting exercise. It's a design exercise, right? We're building it. And so breaking our way is not just a matter of,

it's a matter of making good collective choices and that's extremely hard to do. And so that is what's feasible, but not easy. If we were going to do that, like where would we start? I would say, look,

Healthcare and education, two activities in the United States that's 20% GDP. A lot of it's public money, actually. And this is where there's such great opportunity, where AI could be a tool that could be so helpful to us in a way that other tools have not been. And that's where we could be really investing. And investing doesn't just mean more treatment for rare diseases. It means things that make healthcare more available to everyone so people have longer lives and higher quality lives. That's where I'd like to get started.

Special thanks to Surya Yalamanchili, Saida Sepiyeva, Ian Ellis, Greg Beato, Parth Patil, and Ben Rallis.

And last but not least, a big thanks to Abby Abizorius.