Loading summary
Transcript82 lines
- [00:01]
A
Avoiding your unfinished home projects because you're not sure where to start. Thumbtack knows homes so you don't have to don't know the difference between matte paint, finish and satin or what that clunking sound from your dryer is. With Thumbtack, you don't have to be a home pro, you just have to hire one. You can hire top rated pros, see price estimates and read reviews all on the app download.
- [00:24]
B
Today you're about to make a trade. Which u do you listen to? Is it get optioning those options.
- [00:33]
A
Or.
- [00:34]
B
Let'S do a little research? Learn more@finra.org TradeSmart.
- [00:43]
A
Welcome to Solutions with Henry Blodgett. Today there've been a lot of apocalyptic predictions out of Silicon Valley. In particular that our economy and job market is about to be destroyed by the adoption of AI. Given this, and given a rough job market for folks who are just graduating from college, we wanted to talk to an expert. So we reached out to David Deming, who's a Harvard researcher and economist who has studied the impact of technology on jobs over history and very specifically the impact of AI on the current job market. So here's our conversation with David Deming. Hope you enjoy it. David, it's great to have you.
- [01:29]
B
Thanks so much for joining me.
- [01:30]
A
It is a privilege to talk to you. You've done such amazing work on this topic. So let's jump right into it. You listen to folks in Silicon Valley, particularly folks who either running AI companies or are familiar with it. They are saying we are about to have a jobs apocalypse where everything in the economy will be automated. Certainly the estimates range. But I have listened to several very smart people say, I don't know what to tell my children, maybe they should become plumbers. Then you listen to which by the way, it's a great job. Not, you know, anyway, then you listen to other economists saying, no, no, no, it's totally exaggerated. I think a colleague of yours at mit, Darren Acemoglu, says no, it's going to be a very small percentage, only 5% of jobs and so forth. So where are you in this? Are we on the verge of an apocalypse?
- [02:21]
B
Good question. I mean, it's an important question to answer. I first of all, I think we should all admit that there's a wide range of uncertainty around this, as there often is in times of history where things change. And so first thing I'll say is that we've had a lot of big technological shifts over the last 150 years. And then even going back before that before we had good data. But if you look at things like what happened when the economy electrified or when we invented the personal computer and that spread throughout society, or the Internet or steam power going back even farther, there's a long history in the US and in other countries of adoption of major technological advances that really fundamentally change the economy. And in each case, we have not lost jobs. In fact, we've added them over the course of human history. And so that doesn't mean that will always be true. This time could be different, and I'm open to the possibility that it will be. But it's a difficult question to answer, Henry, because there's no data on it. I haven't seen any sign of a jobs apocalypse. Actually. The US economy is roaring along, and so that may change, and I'm open to it. And I do think, as maybe we'll talk about later, there are some capabilities that AI has that from a first principles point of view, might replace some categories of some things that people do on the job. So I do think it's going to be tremendously disruptive. But the time horizon over which that happens and whether it actually leads to a jobs apocalypse I think is very uncertain. And I would just kind of lean towards no for the reason that the best prediction of the future is a trend line from the recent past. And, you know, so if I was, if I was a betting man, that's where I'd be betting. But I don't know for sure.
- [04:04]
A
So in addition to your academic work, one of the things you do that is a huge gift to all of us out here who have to fight our way through academic papers, is you write a substack called Forked Lightning where you've talked about a lot of the context leading up to this. And one of the articles you wrote, which was particularly vivid to me was the change in farming. And so maybe you could talk a little bit more about that. I think right now we're exposed to lots of viable talk about how the decline of manufacturing has really hurt the middle class in the United States and contributed to our current political environment. But as you point out, before that it was farming, it wasn't manufacturing, and just massive changes in the economy.
- [04:46]
B
Yeah, exactly, Henry. So if you think about like the, I would say the biggest change, economic change in human history was the movement out of farming, out of subsistence farming in particular, meaning farming, you know, growing just enough to feed you and your immediate family into farming as an occupation, into, oh, actually we're producing so much food that we don't need everyone to be a farmer. And that, in some sense, was the beginning of the modern economy. The idea that the way I can earn a living is by specializing in something that if I only produce, that wouldn't feed my family. But I produce something that has economic value, and I sell my services to the market, and then I get some money back with which I can buy food and many other things. So that's a relatively recent change in the scope of human history, and we have a lot of evidence on how that happened and how long it took. And if you just look at farming in the U.S. in 1890, about 40% or 2 in every 5 jobs in the U.S. economy were in agriculture. And most of that, as I said, was subsistence agriculture, meaning you're only growing enough to feed you and your family. You're not actually selling it on the market. And then today it's less than 2%. So the economy has gone from almost half farmers to closer to 1 in 100 or 2 in 100 and over that if you told people we're going to develop a technology that makes farming way more efficient, and you don't only need 1 50th as many farmers as you had, people would say, well, that means we're not going to have any jobs. What are we going to do all day? And yet we have farmers found lots of other things to do, and we've become much more prosperous as a society as a result. And so it's actually worth thinking about the mechanics of how that happened. It didn't happen overnight. Okay, so farming as a share of all jobs dropped from 40 to 2%, but it did so over a century, basically in a very gradual way. There were some periods of time where it happened more quickly. Those periods of time were many years after the invention of mechanized power for a variety of reasons we can get into. But I actually find that to be a helpful example because it was a pretty big change, and yet it happened.
- [06:46]
A
You know, gradually, a colossal change, and for very good reasons. Lots of people immediately want to focus on the displaced farmers, and we will get to them in just a second. But just another macro question that you point out very clearly is often in these changes, there are very hyperbolic forecasts up front about the devastation to come. So talk about that, too.
- [07:13]
B
Yeah, sure. So if you look. And when I used to teach a class about economic inequality at the Kennedy School in the economics department at Harvard a few years ago now, I taught a whole class about automation and the future of work, and I would put up A bunch of different newspaper clippings and screenshots of people being anxious about automation and technology taking our jobs, going back a hundred years. So in the Lyndon Johnson administration, there was a Blue Ribbon Commission report written about automation anxiety and the fear that technology was going to take all of our jobs. It didn't do that. In the 2000s, Carl Benedikt Fry and Martin Osborne wrote a paper where they speculated that based on current trends in computerization, something like 40% of all US employment would be automated within a couple of decades. That also didn't happen. And there's many other examples, you know, the Luddites doing their thing. So this has really gone back a long time, and it's taps into a very real anxiety, which is that technology does replace labor. It does for some categories of jobs, end them, and force people to find other work. And that's tremendously disruptive for those people, for their families, for their communities. And so that's a real challenge. But that's not the same thing as saying all the jobs go away, certain jobs go away, and then other jobs are created. And it's just very hard to forecast what those jobs are going to be. If I told you that your. If I told your great grandparents that you were going to be a podcaster or a financial journalist or an occupational therapist or a software developer, they would have no idea what those things are. And yet they're very important jobs in today's economy.
- [08:44]
A
So you mentioned the Luddites. And I have to say, recently I read an article by John Cassidy in the New Yorker that talked about the actual Luddites. And I realized that I had misunderstood the term my entire life, which is, I thought it was people who, for whatever reason, irrationally weren't interested in technology or gadgets. And it turns out, actually very specific group of people whose livelihoods were being destroyed by the industrialization in England and the creation of huge factories that are producing cloth and weaving. And this situation was so desperate for them that they actually marched on factories and became violent. And there were deaths and violence around this. And one of the things that was most striking to me about it was that it was not some philosophical aversion to technological advancement. It was the fact that technological advancement was actually destroying their livelihoods. And there was no social net. People were actually starving. So talk about that. And that'll. That'll get us started down this road of, okay, from a big picture, it looks great. There are new jobs. But if you are one of the people whose jobs is taken, it can Be very rough.
- [09:56]
B
Yeah, that's exactly right. And I would. That's why I think it's important to be clear about what we mean. So it's. If you're asking me is there going to be a jobs apocalypse and no one's going to be working and everyone's going to be on ubi, I would say that history tells us that won't happen again. Doesn't mean it won't happen in this case, but it hasn't happened in the past. But technological changes have been tremendously disruptive to human society and how well we've adapted to that in terms of policy, like how we help the losers in these new regimes and then what we do to encourage innovation so that new jobs are created. There are lots of times when that's happened. Well, and there's some times when it hasn't been so good for people. And a lot of examples of societal unrest, as you mentioned, with the Luddites and the looms associated with governments and policymakers not taking those impacts, those negative impacts of technological disruption very seriously. And so the challenge for us is not, it's not about being Pollyannish and saying, don't worry, we'll create more jobs. That doesn't just happen. You have to, you know, put your shoulder to the wheel and make it happen. And that's really the challenge we face. Not like no one's going to have work, it's that there's the AI, like other technologies is going to shuffle around the winners and losers in the economy tremendously. And we need to be prepared for that.
- [11:10]
A
One more macro context question and then we'll get to AI, which is you and Larry Summers have done. You did a great paper on, hey, let's look at actual rate of job change across the decades. And one of the very surprising findings, findings to me was despite the fact that we all talk all the time about the world is changing so fast, how can we possibly keep up? One of the findings was that actually there's been less job change. And so maybe you can give us a picture of really where we are in that whole transition.
- [11:41]
B
Yeah, happy to do that. So this is a paper, as you mentioned, it's joint work with Chris Ong and Larry Summers that is called Technological disruption in the labor market. And what we did in that paper is try to say, you know, can we in a principled way measure the rate of job change over a long period of time in the US and ask the question, like, are things changing faster than ever? And that required some data work and some assumptions about how to measure change. And so what we did was we said, okay, let's divide all work into major categories of occupation. So it's things like agriculture, blue collar work, you know, manufacturing, construction, things like that, clerical work, sales, management. So, like, pretty broad categories. And again, that's necessary because when you go back to 1880, there's no such thing as a software developer or a business analyst. So you have to broaden the categories. And then we said, okay, let's do something very simple and ask the question. If we divide all those jobs into categories and we ask what share of each job belongs to, so how big are the are the categories? So let's say farming is 40% of jobs, as I said, and now it's 2%. So you divide everything into categories, and it all adds up to 100% in every year. So the way to think about this is if you just randomly picked a worker in the economy in a year, what's the probability they'd be in one of these categories? Right? And then you ask the question, how much of those categories shifted over periods of time? So over decades. So if the economy is 40% farmers in 1880 and 38% in 1900, it's changed by 2 percentage points, right? So then we take that change, we take the. And we count winners, losses, and gains symmetrically. So a gain and a loss are the same thing. We add that up and say the 40 to 38 is a 2 percentage point change. And you add all those changes up across all the categories, and you get a total measure of what we call churn, which is basically how different is the occupational structure of the economy from one decade to the next. And if you think about it, what it means is the job distribution is totally stable. You get a churn rate of zero. And the more movement across categories there is, the higher the churn. And so the virtue of a measure like that is it's simple enough that you can go back 150 years, put everything on the same scale because it always adds up to 100, and then compare periods of time systematically. So when we did that, what we found was the most disruptive period in US history was the middle of the 20th century. This was a time when farmers were finally willing to give up their livestock and move to using tractors and other mechanized forms of farm labor. And so farms got a lot bigger and employed fewer people, but produced more crops. And then a lot of those people ended up moving into cities and working in offices or working in manufacturing. It was also a Period where the railroad, which was the main means of passenger travel for a number of decades in the US was being replaced by the personal automobile. So we were building a highway system and so people stopped traveling on the railroads. Like my grandfather worked on the railroads and he, you know, future generations did not because there were fewer railroad employees. So even within blue collar work there was a ton of change. So it's just a period of tremendous disruption. And when you compare it to the modern period, you just see that it's not today is not on that same scale. The changes in the economy are not as big. It's not because things aren't changing quickly today. It's. It's because things were changing even more quickly in the past. And so I think the lesson I learned from this Henry, is that the US economy is in a constant state of disruption. And so the bar for what it would take for AI to truly beat all that is very high.
- [14:53]
A
And you can show us these charts as one of the decades that shows that trend where things have actually seemed to be a little bit more stable in the last couple of decades.
- [15:01]
B
Yeah, actually the 2010s is the most stable period in a hundred years, which.
- [15:06]
A
Is fascinating because it certainly doesn't feel like it.
- [15:09]
B
No.
- [15:09]
A
So basically ChatGPT came out and shocked the world at the end of 2022. And that's when we first started to see these hyperbolic possibly projections about job apocalypse and so forth. Where are we in terms of adoption? I lived through the Internet. People thought that was adopted pretty quickly. I know you've studied AI adoption so far. Where are we on that?
- [15:32]
B
Yeah, so it's a great question. And it is true that the history of, of major technologies is they get adopted very quickly. And that's one leading indicator that's going to be a big deal. So what my co authors and I did in a recent study, this is bake bland and indemning. It's a paper called the Rapid Adoption of Generative AI. So giving away the conclusion in the title there what we found, we conducted the first nationally representative survey of generative AI usage in the US And I don't want to bore your listeners with details, but for those of you who know the Current Population Survey, that's the major source of labor market information in the U.S. it's a government survey that's administered monthly. It's the reason we have things like the unemployment rate and the jobs report and all that stuff. We created a synthetic version of the cps. We replicated the survey design structure. We Give the survey in the same week, we draw on the same sample of people, and then we ask them all the questions the CPS asks so that we can basically match our data to the CPS and make it look demographically exactly like the Current Population Survey. And then we add a bunch of questions about AI. So what we're doing is it's like a thought experiment. If the Current Population Survey, if the Department of Labor added questions about AI, what would they find? That's what we were trying to do. So what we found is that in late 2024, about 40% of people said that they use generative AI. That's a mix of use at work and outside of work at work. Roughly 1 in 4, 24% of people said they use generative AI at work at least once in the last week. That number has gone up a little bit in recent times, but it's still a pretty big number, especially given that at the time we gave the survey, ChatGPT was only 2 years old. And so we have a bunch of other statistics on this. But the big thing that we do to answer your question, Henry, is the cps, in years past, actually asked questions about the adoption of two other major technologies, the personal computer and the Internet. So we asked the questions the same way so that we can basically say, well, if the CPs were to also ask questions about AI, what would the rate of adoption look like of these three technologies? And in order to compare, to calculate that, we have to take a stand, like, decide what is the beginning, quote, unquote, beginning of a technology. So we date the beginning of the personal computer to the release of the IBM PC, which was the first PC to sell at least a million units. That was generally thought to be like the mass market adoption of computers. That was in 1981. And then in 1995, the Internet. As I'm sure you know, we used to be nsfnet, it was a government service that connected supercomputers. It was decommissioned in 1995 and opened up to commercial activity. That's also the year that Netscape IPO'd. So we'd say that's the beginning of the Internet. So then we could say, okay, you know, we're two and a half, three years into AI. Now. If we have 40% adoption three years in, how does that compare to the PC? Well, the PC actually had adoption rates of about 25% three years in, and the intranet, it was about 35%. Okay, now, if you look at. So on that scale, adoption of generative AI looks to be faster than adoption of PCs or the Internet. Now, critically, there's a gap between usage at work and at home. So usage at work, just work usage. Generative AI is on a similar scale to computers at work, but it's much faster outside of work and Internet, we don't have the data, they didn't separate the questions, so we don't know. But basically the way I would think about this is not to interpret these numbers super literally, but just to say it's on that scale in the sense that, you know, if I told you back in, if I showed you the Future back in 1984 of personal computers, you would say they're everywhere. They're in every aspect of our life, they bleed into everything. You wouldn't even think to ask questions about PC adoption because everyone just thinks of them as a fact of life. And what history tells us is that AI is eventually going to be like that within a few decades that it's going to be on that scale of just everywhere, all the time.
- [19:16]
A
And so what are we seeing thus far? We're not very far into it, but do we have any evidence? I know Silicon Valley again talks a lot about how a huge percentage of code creation is already automated and they're going to be far fewer programmers and so forth. What are we seeing across the economy?
- [19:35]
B
Yeah, it's interesting. So actually software developer jobs is in some ways the only thing where we're seeing anything major. There has been a pretty big decline in job postings of software developers in the past couple of years and a lot of people saying they're hiring fewer coders and things like that. I think it's a little bit mixed up, Henry, in a cyclical trend in this. So you saw a huge increase in hiring and employment in stem, particularly computer related occupations over the last decade. And a lot of that was big tech companies staffing up, making huge investments in data centers. Just like an AI investment boom. Investment in information processing and equipment, equipment, things like that related to computers has been way up. It's basically at.com levels right now as a share of GDP and it's been up for a long time over the past decade. And so I think part of what you're seeing is companies may have overshot the mark. So I'm not sure how much of this is actually AI replacing software developer jobs versus companies scaling that back. Anyway, I think we'll have to wait and see for sure. In terms of other impacts on the economy, there's really not much to point to. There's A lot of speculation, but there's not a lot of hard data on how AI has affected the economy. And I'm not surprised by that. It's only been two years, so I think it's a little too early to say what the impacts will be, so we'll have to wait and see. But certainly a lot of talk, a lot which often precedes action but doesn't always.
- [20:59]
A
All right, well, so let's look at some of that talk and then we can get into it. So just a few examples. So McKinsey, in a report that I have to describe as hyperventilating with excitement about all the productivity that's going to be unlocked and how much money everybody's going to make except for the folks who are actually replaced, has said 60 to 70% of the time we spend at work can be automated. They've said half of jobs can be automated by 2045 and they will see productivity growth acceleration by 0.1% or 0.6% going to Silicon Valley. Vinod Khosla, not known as a hyperbolicist, has said he thinks that 80% of jobs will be replaced. Elon Musk, who is occasionally known as a hyperbolicist, has said that all jobs will be replaced. And as I said, I listened to a long discussion with a very smart, longtime Silicon Valley venture capitalist who was basically saying, look, I don't know what to tell my kids, I really don't. But they can't go into law, they can't go into Hollywood because ChatGPT is already pretty good at writing scripts and that's gonna all get replaced. And so I don't know what to tell them to do. So I, you know, do we know anything?
- [22:13]
B
Like what?
- [22:14]
A
Let's say we've talked about programmers. What other kinds of jobs do you think are very exposed in the next few years?
- [22:23]
B
I don't know how anyone arrives at a quantitative forecast like x percent of jobs. Like, I don't, I don't understand any principal way to do that, but I do think it's valuable to think again from first principles about what is the technology capable of and what are the kinds of jobs in which it might be more disruptive or more helpful. And I do think white collar, information oriented work, sometimes entry level work, looks to me like something that AI is pretty good at doing. What do I mean by that? So if you want deep research to write a paper for you about some topic on which you know Nothing in the pre2022 world, you would hire, let's say, an entry level college graduate, to be a research assistant or to be some type of, you know, worker on the, you know, the entry rung of the corporate ladder whose job it is to help give you context to make a good decision, you know, so it's like I need to have a meeting where I'm going to decide on some. Make some business plan or make some strategic decision about how to run my company. And I want a bunch of people to collect and synthesize the information I need to make that decision. So, like, that's a major category of work that seems like AI is going to be pretty good and cheap at. I'm actually not sure. Having used deep research a lot myself, I'm not sure it's strictly speaking better than a really good person, but it's definitely cheaper and faster. Yeah, that's the thing that blows you up. Yeah, that's what I mean. Yeah. So it's like per unit of time spent, it's giving you much better work than a human. And so I think it wouldn't be surprising at all if companies started to be entrepreneurial and hire. It doesn't necessarily follow, by the way, that they'll hire fewer college grads. They just might hire people to do different things. And so that's where the thing I very much sympathize with having children of my own is this. I don't know what I should tell my kids to do. I think that's totally right because it's very clear just from looking at this that we're about to enter a period of a major shift in what we're asking people to do, given that they now have this intelligence on tap that is cheap and easy to access. And so I think a prediction I feel very confident making is that things are going to change. I just don't know how you kind of go from that to 70% of jobs are going to be automated or 80% or so. I just don't, like, I. You can give a number if asked, but I don't know how you're arriving at that. And therefore, I don't really have any reason to trust it, you know, and so, and a lot of it is like, if. You have to understand, like if. And I don't want to be too cynical, but a company like not McKinsey, but a company like that that is essentially trying to sell services to companies about how they should use AI has an interest in talking about what a big deal it's going to be, which doesn't mean they're wrong. You just have to understand that it's in their interest for these forecasts to be true and therefore they're more likely to make them.
- [25:01]
A
And then the other thing McKinsey and everybody else is talking about is productivity. And is there a reason that normal people who don't work in the investment business or the consulting business should care about the speed of productivity growth? How does it matter to the rest of us?
- [25:20]
B
Yeah, well, that's a great question. It matters a lot actually because, you know, it's a cliche, but I think it's true. Is that like the real risk is that people in your profession who learn how to use the AI better than you are going to end up being seen as more productive than you. So I think if you're in any job that depends on the kinds of things AI can do, you need to learn how to use it, you know, And I actually still think for most people if you are using AI, well, you can probably out compete just simple AI, you know. And so where productivity growth really matters for workers is it gets embedded into the expectation that firms have for what they expect you to do. So in a pre AI world, your boss might say, I need you to write me a research paper about topic X. I assume that's going to take you about a week, so have it on my desk by Friday. Okay. And now eventually they're going to say, have it on my desk in three hours because they know you're going to ask deep research for a draft and then you're going to edit it. And so that's productivity growth. That's what it means. It's like I can get one paper a day instead of one paper a week and that helps me make decisions 20% more effectively or whatever. And so that's how it ends up feeding into productivity growth. But the only way you get productivity growth is people can do more with less. And that means the expectations go up. So it's very, it's very fair for the average worker to care about the impacts of AI on productivity.
- [26:44]
A
And, and what are you seeing in terms of, in the actual professional world, the attitudes toward doing that, which is, you know, how. Get me the paper in two hours and I'll give you an anecdote. I was talking to partner at one of the big corporate law firms recently who saying like, yeah, you know, we're finally starting to think about maybe our class size doesn't grow every year of new people because we're in a position where a AI can, can take a 300 page corporate M&A doc and in two hours do as good a job on it as a third year associate out of Harvard Law School. So, you know, what is that? How does that go through? And, and what about the ethics of that? I mean, I, you know, you go, you're in school, people are mortified by the idea that students or professors would use AI and the working world's very different.
- [27:35]
B
Yeah, it's a great question. I mean, my sense of this is that for lots of reasons, not even primarily related to AI, there's a ton of uncertainty in the business environment right now related to tariffs, related to global political issues and so on. And so I think companies are extremely hesitant to make major moves in either direction. So I think part of what you're seeing, there's been some evidence that the job market for entry level college graduates is getting worse. And I don't think that's because firms are just saying, well, we're going to replace you with AI. I think it's firms saying, we might be able to replace you with AI, but we're not 100% sure yet. And also we're not sure what the tariff rate is on our product's going to be. So we're just going to sit tight and figure it out. And maybe we're going to use AI, but we're also going to check everything we do with a person, which doesn't actually save time because if you have AI do it, but then you have somebody read it over again to make sure it's accurate, you're not actually seeing productivity gains. You're more in the experimentation and testing phase. And so I think that's where we are with most companies in the economy. I think very few of them are just using AI with confidence in a way that would unlock productivity gains. But what you often see historically is that recessions, periods of economic weakness, are periods where people go from experimenting to actually adopting because they're forced to. If you feel the pinch on your bottom line, that's when you'll go from, oh, maybe we shouldn't hire an extra person to we actually can't. So we're going to have to find a way to use AI to replace this. And so my sense is that the next time the economy enters a period of economic weakness is when we'll see faster, meaningful adoption of AI. But I don't know that for sure. We'll see.
- [29:06]
A
And I listened to a fascinating discussion between you and Derek Thompson of the Atlantic recently, where Derek was making the point like, you know, what is the right analogy for what we've seen of AI thus far. And he basically said, is it a spreadsheet or is it a horse? And the point is that back in the pre Excel days, everybody on Wall street used to stay up all night actually adding up columns of numbers with a calculator and then writing.
- [29:32]
B
Sound like you're speaking from experience?
- [29:33]
A
Yes. No, I got there after the spreadsheet. But then one had to use it. And so suddenly the associates who knew how to use Excel had a huge advantage over the folks that could just pound numbers into a calculator. So that's the spreadsheet and then the other is a horse. We used to have horse drawn carriages and they were replaced by cars and the number of horses plummeted. So based on what you've seen, what do the better analogy is.
- [29:58]
B
So, you know, I think it's going to be some mix of the 2. I think AI will replace some functions, maybe not whole cloth, but close to it. I don't think it's particularly close to being able to do that now because of the reliability issues and the, you know, there is agentic AI, but it's in a very early stage and it's not that reliable. I've tried to use it for different things and I think, you know, if you look at again the history of horses or farming or other things, like people were loathe to give up their livestock until, you know, tractors and other farm equipment was general enough that it could do everything. You know, if you think about like, what's the value of a horse? Well, you can, you can attach it to a carriage and, you know, bring you into town, you can use it to plow a field. Like there's a lot of different things you can do with it. And so the technology is going to need to be both very reliable and very general to fully replace existing processes. So I think that will happen, but it will take some time. I think the more immediate thing you'll see is that AI is going to greatly lower the cost of certain job tasks and the time it takes to do them. And the people and the firms that are going to be the winners there are the ones who figure out what to do with the extra time that can't be replaced by AI because AI is something that everyone has. And so if you want to, a lot of what business is is beating your competitors, right? And so if everyone's got the same AI, it's not actually a competitive advantage. The competitive advantage goes to the people who figure out how to use it better. And that's going to have to do with integrating it into your process and figuring out, given the talents of my employees, what should I have them do instead? And one final thing I'll say about that, Henry, is that AI is intelligence on tap. But actually, intelligence isn't the only thing or even the primary thing we want in many jobs. So if I'm hiring somebody to be a coach or a mentor or a guide, I don't care how smart. I mean, maybe intelligence is an input into it. They could be better at their job, but that's not actually what I really care about. And so even if I could hire a, you know, AI mentor, I mean, some people are using it for that, but, like, I might not want to because I want to have a human connection. And. And, you know, there are a lot of service jobs where the person is the point. And so I don't see those going anywhere. And that's a pretty high share of jobs for which. That's true.
- [32:04]
A
You did some great work on. I think it was a study of, okay, let's. Let's see how humans differ in leadership skills and what matters. And you've talked about the importance of social skills, which lots of AIs, maybe they have it in language, but in other ways they don't. So tell us about that. What's the hope for us humans in this new world?
- [32:24]
B
I think there's a positive vision, Henry, where work becomes more human and more satisfying because we have a technology to do the things that are kind of unnecessary but rote and not very fulfilling. And you did see that with physical labor, you know, like, as we've just discussed, most of the way you could earn enough money to feed your family a hundred years ago or even longer, was working so hard that it sends you to an early grave, giving your body, you know, in terms of like, whatever, digging holes or mining coal or building cathedrals or all the things. You know, many monuments in human history were built on the backs of either slave labor or people who were working for a small amount of money because they had to do it to survive and feed their families. And fewer people. People still do that kind of work, but it's much less common. And so, in a sense, we've shifted more towards office work. But a lot of office work, it's not physically as draining, but it's intellectually draining. A job like telephone operator, where you're just running around a room connecting different phone lines, or a job where you're just typing up somebody else's notes, or you're writing marketing copy all day, like that's not as fulfilling as many other jobs. And so there is a world where instead of doing that, you're like going out on the road, talking to people, trying to sell them on different ideas, building relationships with clients. Those are still things that are very far away from being automated and they're in some sense more human and more meaningful and fulfilling. So it wouldn't surprise me if we move even more toward an economy that is built on high skilled personal services where like you can basically have the AI version that's commodified, you know, so I can get like an AI chatbot. But the really, the clients who are willing to pay more for something exclusive and personalized get to talk to a person. So the person is the luxury, the person is the expensive input in the process. And the skill that really matters is do you, can you quickly establish a relationship with someone and build trust with them so that you can get them to spend a lot of money on something? And we already see this happening in the economy, but my sense is AI could accelerate that.
- [34:33]
A
And so maybe there we do, as we get into solutions here, which is what we're aiming for on this show, maybe that is some guidance to our children or students, which is, hey, you know, learn how to form relationships quickly. Learn how to effectively. Yes. Form relationships and help people.
- [34:52]
B
Yes. And that's, you know, so I do think, you know, this is a lot, in some ways what the economy values, you know, can change and then it's, but it's still always downstream of educational curricula. So how we, you know, our school system was built in an era where the demands of the economy were quite different. You know, this is a thing that lots of people say, but it is true. And we've already seen like, if you look at, you know, college classrooms, many high school classrooms, even like preschool classrooms, you see a lot more interaction between students and with teachers and group work and project based work. So I actually think schools are already adapting to this trend. If you compare it to what a college classroom looked like in 1950, I think there's already a lot more social interaction, teamwork in the classroom. But I believe that should very much continue. I think that's what the economy wants, but I also think it's a better way to be. The skill of learning how to work with someone else is also the skill of understanding their perspective and being able to walk a mile in their shoes and understand across differences. And so that's the hopeful vision to me is a world where we build social skills in schools for economic reasons is also a world where we get along, better understand each other, and can live harmoniously. I mean, I'm not that sounds, you know, sunshine and rainbows, which I'm not opposed to, but I think it's possible. I think that's one possible outcome of this, is that we all spend more time understanding each other and working together and building relationships, which sounds great. Abercrombie is an official fashion partner of the NFL and I'm Ceedee Lamb, wide receiver for the Dallas Cowboys. You know, I'm here for Abercrombie's Cowboys gear. That's not a question, but I need a whole wardrobe to go with it. No shade to the guys, but I'm used to having the best tunnel fits. This season, Abercrombie has me covered. Shop NFL by Abercrombie in the app, online and in store when did making plans get this complicated? It's time to streamline with WhatsApp, the secure messaging app that brings the whole group together. Use polls to settle dinner plans, send event invites and pinned messages so no one forgets mom 60th and never miss a meme or milestone. All protected with end to end encryption. It's time for WhatsApp message privately with everyone.
- [37:14]
C
Learn more@WhatsApp.com this episode is brought to you by State Farm. Checking off the boxes on your to do list is a great feeling. And when it comes to checking off coverage, a State Farm agent can help you choose an option that's right for you. Whether you prefer talking in person on the phone or using the award winning app, it's nice knowing you have help finding coverage that best fits your needs. Like a good neighbor. State Farm is there.
- [37:44]
A
All right, just one more study and then we can move on to some other fascinating things that I want to get to. You mentioned a Procter and Gamble study where basically they tested individuals working on a problem, teams working on a problem, and then teams working with AI and individuals working with AI. And that was another area where it seemed to me if we're going to say that maybe it's not going to be an apocalypse, maybe we actually can all get better using AI. That was a good example of it.
- [38:13]
B
Yes, that's a fascinating study. It's called the Cybernetic Teammate Study by a bunch of authors from Harvard Business School and other places where they did an experiment with Procter and Gamble where they gave people AI teammates essentially to work with and asked, you know, under what conditions are AIs helping people work better. And the answer was that, basically, which I found very interesting, was that AI does help you, but it helps you, especially when it complements your expertise. So if you are, for example, a technical expert, but you lack, you know, knowledge of the product market, AI can help fill in that gap. Same is true in reverse. If actually what you know is the product market itself, but you don't really know how to, you know, the technicalities of it, it can fill that in. And so what the AI did for those folks is it fit in to the production process in a way that, that filled in your gap. So what I take from that is that a really good way to use AI is to think about it like a teammate that can complement your expertise by doing things that you don't do as well. So I think one of the best ways to use AI is to not to have it replace you and what you're expert in, but to have it shore up your weak points, you know. So, like the way I use AI, like if I ask AI to write, if I ask Deep Research to write a research paper about something I know a lot about, you know, higher education or soft skills or something, it's not as good as me. It's not bad. It's like undergraduate term paper good, but it doesn't know everything I know. But if I ask it to write a paper about, I don't know, the history of telephone operators prior to me writing that column, it would be much better than me at that because I don't really know anything about that. Like, I had to learn about it to write. And so in that way, that's really the value of it is a teammate that knows a little bit or knows a lot about everything, but doesn't, isn't truly expert in anything.
- [40:00]
A
So one more question and then we'll move directly to some solutions. Another fear, even among people who think, yeah, okay, you know, we'll have jobs, but there aren't going to be good jobs, is that AI is actually going to exacerbate the already extreme inequality that we have. And people are talking about, yes, there will be a few more billionaires and then Effectively we'll have 8 billion serfs who are just feeding into the AI. Are you seeing anything that gives any early indications of that? What do you think happens to inequality here?
- [40:37]
B
Yeah, so I think that's something that's worth worrying about, but I don't think it's really unique to AI. I think anytime there's a period of disruption, either driven by technological change or driven by Political change or really anything, there's always a risk that people with power and influence will capture the benefits for themselves. So I think that's something as a society we should be worried about. I absolutely agree with the premise of it. I just don't think it's particularly unique to AI. I don't think there's anything about the technology that intrinsically increases inequality. Actually the opposite of some ways. If you think about who is it most useful for, you could argue it's actually most useful to people who live in developing countries who want to participate in the global economy but don't speak English and so, or need help with, you know, kind of basic. Again, like, there's a lot of evidence that AI helps you get to above average very quickly. That's the kind of thing that would benefit people who have these gaps even more. And so when I look at the technology from first principles, I don't see something that has to increase inequality. I think, if anything, it's the opposite. But it's all in how it's used, who controls it, you know, how it's priced, how it's made available, safety concerns about it, you know, whether it's used to build a weapon or whether, like, those are all things that are not fundamentally about the technology but are about how it is laid over with the Institute, with our societal institutions. And so this is my way of saying we should be worried about it, but it's not something that the fix is not technological, it's political and social.
- [42:08]
A
So. But also, you sound more optimistic there than I think a lot of people are.
- [42:14]
B
Well, I mean, I don't know, maybe just realistic. Like, I don't, like, I think if your line of argument is like, oh, well, AI is just going to increase the inequality that's already going on, I'm like, well, yeah, but maybe that's going to happen without AI too. So it's just hard to attribute it to AI. I think we have big problems in society with concentration of power and wealth, but those problems existed before AI and they're going to exist after AI. So we should solve those problems and not think about, I guess, what is the implication of this, that we should just stop the technology in its tracks to prevent inequality from growing. I just don't. First of all, you can't do that, you know, feasibly even we could do it. I don't actually think that would address the problem. The problem is inequality. The problem is not AI. The problem is inequality.
- [42:52]
A
Yes, and there was a brief moment, I don't think it had to do with inequality, but I think it was the fear that we were going to have a Terminator situation and all be suddenly destroyed by AI that we should. A couple of years ago, hey, let's sign a letter saying, pause. And I think the discussion about that lasted for maybe half an afternoon before we were back to the race.
- [43:10]
B
That's true. Although the one thing I will say is that, you know, I think we should be taking discussions of AI safety very seriously, because even if it's not a particularly probable scenario that AI, you know, superintelligence, takes over the world and kills us all, you know, what is the probability that's high enough that you'd want to buy some insurance against that? Like, 1%, 2%? That's a really bad outcome. That's like game over for all of us. And so even if we think it's a small chance, we ought to try to stop it from happening. And so I'm very, very supportive of efforts to make the technology safer precisely because it's so powerful.
- [43:41]
A
Yes. And I think a lot of people are hypothetically supportive of those. Yes. That discussion also seemed to last maybe a day or two, but is now deep in the rearview mirror as we race, and suddenly it's a race with China, and if we don't win that one, we're toast and so forth. So it seems like we're going to be dealing with that risk when we come to it rather than ahead. All right, let's talk about solutions. So going back to the Luddites or other folks who have lost their jobs because of automation or anything like that, in the United States, our attitude seems to be, hey, whatever, your job is on you. If you have to move, learn new skills, that's on you. Based on what you've seen around the world and in the United States, what works, what. What should we be doing if we say, yes, there is going to be an acceleration in displacement here, or at least job change? What should we do that to make that less painful for people affected by it?
- [44:45]
B
Yeah. So it's a difficult question and in some ways very hard to answer because the future is so uncertain, so we don't even really know what we're trying to prevent. I do think, you know, maybe this is. This is definitely an answer that's born out of my own experience and expertise, which maybe it's not right. But I do think maybe the biggest source of inequality among people is what they're capable of and how they're trained and what skills they have. And so that's true before AI, it'll be true after AI. And so I think the solution for this ultimately has to come through the education system and to think about what are some different ways of learning on the job. Learning training that's general enough to be valuable, not just in the company you're in, but in other companies. How do we build an education and training pipeline that allows people to acquire the necessary skills for an AI future? And that's probably going to necessitate a lot of change in what we teach and what we value. And so that's not an overnight solution. But you know, it's kind of the saying, the best time to plant a tree was 20 years ago, the next best time is today, you know, and so I think we ought to prepare for like the one thing that's certain to me is that I feel, well, I feel much less certain I should say, that the economy in 20 years is going to look a lot like today's economy than I would say from today versus 20 years ago. Like, I think there's a very real chance that things will look a lot different. And so therefore what we need is a flexible education and training system that prepares people for an uncertain future. And that implies, by the way, that we shouldn't be like, if you think about what was all the rage, coding boot camps. Everyone needs to learn how to code. So let's invest in these narrow vocational training regimes. And it turns out that's like, actually now people can just vibe code. I can just crack open ChatGPT and tell it to create me a website. And so a lot of that training ended up being worthless. In retrospect, nobody knew that at the time. But that's exactly why you want education training to be broad, not narrow, because you don't know what's coming. And so I feel that the more uncertain the AI future is, the more we need, you know, a general toolkit. So I would like to see the education system build these kinds of soft skills that I've been working on. That's why I say it's self interested, not in a financial sense, but just like I think this stuff's really important. I've devoted my whole life to thinking about it, more or less my whole adult life. And so I think that's where we ought to go. But there's a lot of different ways to do it and it's difficult.
- [47:25]
A
So learn human soft skills. One solution for individuals. From what you said earlier, it sounds like we should all actually learn to use AI as well and maybe that can happen in school. So that actually brings up an important point, which is watching from the outside. I mentioned this earlier. There are rampant reports of people using AI to cheat in schools, and it's terrible and you shouldn't do it. And maybe in the workforce it's okay. I remember when one of my early bosses told me, in research, it's like, Henry, it is totally fine to plagiarize yourself all the time. And in fact, if you want to say something and get people to listen, you've got to say it again and again and again. So just plagiarize yourself all the time. You do that in school. I guess it's okay if it's yourself, but plagiarism is terrible. You shouldn't use AI. It's considered cheating. Is there a way to actually integrate it and in fact have the education system say, no, you know what? If you can write a decent research paper in six minutes, which is what ChatGPT did for me recently about this topic, do it and then actually add value to it or present it in class or what have you.
- [48:35]
B
Yeah. So first thing to say for those listeners who don't know, usage of generative AI is completely ubiquitous on college campuses. When I went, I live on the Harvard campus, and so I would go talk to students in the dining hall, and I started off by asking them, do you use it? And people are kind of sheepish to tell a professor whether they use it or not. I say, and what share of your friends use it? 80%, 90%, 100%. Basically, everybody's using it all the time, not just for school, but for other things. And so that's what's happening. Okay. Period. That's happening now. And I suspect I have some kids in high school. I think it's happening in high school, too. And so we just have to adapt to that. And I think what it really pushes on is the purpose of education itself. So what are we trying to accomplish in schools? And I think what we see now is that the way we assign and grade work and the way we think about learning objectives in classes, it's almost perfectly designed to make cheating the best and easiest use. You know, because we're asking people to write papers at home and turn them in, and they're getting feedback from professors. That is all on a common rubric, meaning it's not about your personal journey. It's just like, there's a right answer or there's a way to do this, and I'm just grading you against that. And so it's That's a system where it's very Easy to use ChatGPT to substitute for your own work. And it's actually not that easy to use it to make you better. And so we just need to change. I mean, we don't need to like, I don't think it's anybody's fault, you know, that we haven't solved this problem because the technology is only two years old. But like, we need to adapt to this future and make it so that we need to reverse that calculation to make it Easy to use ChatGPT or other, you know, generative AI products to facilitate learning and hard to use it to substitute for learning. So what does that mean practically? It means sort of what you say, Henry, which is like either blue book exams where you come and you write it in person, that's the low tech solution. Or you assess people based on their ability to use the knowledge they've gained to do something, make a presentation, persuade somebody else that your argument is correct, argue both sides of a difficult issue like sell something, make a product, something. That's more using learning to produce something of value, where you have to make some judgment about what that is. And that's my preferred way of being AI proof, which is to say, well, that's something where the AI will help me generate a lot of ideas if I know what I want to do. But if I just tell the AI, write me an A plus term paper or give me a good, it's not going to do. That's not enough scaffolding. So you have to decide how you want to use it. And that's the learning is that you have this tool and you're figuring out how to use it to get something you want, that you want to make the case is valuable to other people. And that's where I think there's a ton of opportunity to redesign the learning environment to make that easy and to make that work better.
- [51:27]
A
And do you see, do you see signs that the education system is open to that?
- [51:35]
B
I see signs that some people I know who are entrepreneurial are open to it. I think students are definitely open to it. I think education systems are slow to change. And so part of why I say I start this part of the conversation by saying everyone's using it is just so my colleagues understand that whatever system you've designed that you think is getting rid of the problem of AI using classroom, it's just not working. It's just not. I don't care what anybody says, your students are using it much more than you think I know my students are using it. It's not, you know, and so I just think we have to come together on that. And I don't think that's happened in many places yet. But it will. It has to.
- [52:17]
A
And by the way, it is an extraordinarily useful tool for getting yourself up to speed on topics. It it's like talking to somebody who knows way more than you do. And even if it makes mistakes and so forth, the fact that you can interrogate it, it's incredibly useful. So it is.
- [52:35]
B
And that's why I think we'll get there, because it is so useful for learning. We have to find ways to, you know, bend it to our purposes, which is how to get students to learn better and deeper. Running a business comes with a lot of what ifs, but luckily there's a simple answer to them. Shopify it's the commerce platform behind millions of businesses, including Thrive Cosmetics and Momofuku, and it'll help you with everything you need. From website design and marketing to boosting sales and expanding operations, Shopify can get the job done and make your dream a reality. Turn those what ifs into Sign up for your $1per month trial@shopify.com SpecialOffer this.
- [53:16]
A
Episode is brought to you by Indeed. When your computer breaks, you don't wait for it to magically start working again. You fix the problem. So why wait to hire the people your company desperately needs? Use Indeed's sponsored jobs to hire top talent fast and even better, you only pay for results. There's no need to wait. Speed up your hiring with a $75 sponsored job credit@ Indeed.com podcast terms and conditions apply.
- [53:45]
C
Eczema isn't always obvious, but it's real. And so is the relief from EBGLIS. After an initial dosing phase, about 4 in 10 people taking EBGLIS achieved itch relief and clear or almost clear skin at 16 weeks. And most of those people maintain skin that's still more clear at one year with monthly dosing.
- [54:03]
D
Ebglis Lebricizumab LBKZ, a 250mg 2ml injection, is a prescription medicine used to treat adults and children 12 years old of age and older who weigh at least 88 pounds or 40 kilograms with moderate to severe eczema, also called atopic dermatitis, that is not well controlled. With prescription therapies used on the skin or topicals, or who cannot use topical therapies, ebglis can be used with or without Topical corticosteroids. Don't use if you're allergic to ebglis. Allergic reactions can occur that can be severe. Eye problems can occur. Tell your doctor if you have new or worsening eye problems. You should not receive a live vaccine when treated with ebglis. Before starting Epglis, tell your doctor if you have a parasitic infection searching for real relief.
- [54:38]
C
Ask your doctor about epis and visit epis lily.com or call 1-800-lilyrx or 1-800-545-5979.
- [54:47]
A
So lots of, as we look at solutions, lots of emphasis on education and changing education to again, focus on what humans can do to each other. And let's automate the stuff that can be automated. Anything from not necessarily AI, but just your study of the labor market change over time and inequality, or are there any policies around unemployment retraining that the government can do? UBI Universal Basic Income that you mentioned earlier, which has been again, a very favored Silicon Valley solution to this because suddenly 70% of people are going to be out of work. We got to have a universal basic income. Are any of those particularly good and effective?
- [55:28]
B
Well, you know, I think in terms of, you know, more narrow policies or things, you know, I don't know how much of it really has to do with AI. I think there are some things in the US about our workforce training system that could be improved. We don't take it seriously. The US spends about 20% of the average of other OECD countries on active labor market programs, meaning job training, apprenticeships, work subsidies, things like that. So we have a very less a fair system of transitioning people from education into jobs. And that's great if you get a bachelor's degree from a place like Harvard, works really well for you because employers want to hire you. It's not so great for people who have skills, but those skills are hard for them to signal to the market because either they're kind of specialized. So like I learned how to be a nurse because I went to this two year community college program, you know, in Boston. But like the, you know, city and like the program in New York is totally different. And so if I move there for personal reasons, like nobody knows what I can do. So there's a lot of little things like that where we're not very good at. We haven't really designed a system that helps people signal their specialized skills to the job market. And so I think there's a lot we can do on the national level to basically certify different work Oriented credentials in a way that helps people move around and move up a career ladder, create a real career ladder for jobs in fields like information technology, advanced manufacturing, allied health, where you move from being a medical assistant to a radiology tech to a nurse or whatever. So there's a lot of infrastructure I think we can build around jobs, particularly jobs that don't require four year degrees. That would really help people and would really reduce inequality and we create more family sustaining jobs. I think that actually has nothing to do with AI at all, or very little to do with AI, but it's very important and I think it's something people are calling for and it's very bipartisan. So I think it's important to mention here because I've been talking about that and the work I've done with my colleagues at the Project and Workforce at Harvard. We've been talking about this for a long time. It has some AI inflections, but actually, even if the technology never existed, it would still be an important thing to do in my view.
- [57:27]
A
And when you say we, you're referring to national government, local government, companies. How much do companies should companies do here?
- [57:36]
B
So what I mean is some combination of national firms. So companies that hire in multiple labor markets and the federal government should do this. Why? Because actually there's a lot of really interesting work happening in states and cities around workforce pipelines. So if you look in the Boston area, for example, there's a really. There are a lot of really good programs and things like cardiac sonography or licensed vocational nursing, basically health careers or IT careers. It's just that those things are very bespoke to a local labor market and a set of employers that have particular needs and they don't always translate to other places. But a big thing that makes people upwardly mobile is having lots of labor market opportunities. So if you go to a community college and you train, you get an associate's degree in some very narrow thing that was designed with one company in mind. That's great as long as you're at that company. But if the company goes out of business or they change their technology, you have nowhere else to go. And so you have very little market power. You have very little chance to move around and move up. Whereas if you have a bachelor's degree for all its warts, it's a very general skill set that everyone understands that you can take. You can get a bachelor's degree in North Carolina and move to Wyoming and get a job. And everyone understands what that means and that enables people to move up the job ladder. And so what we need is a system like that for people who have something less than a four year degree, but that requires federal action precisely because it has to live at the national level, not at the local level.
- [59:01]
A
So one of the things we like to do on this show is talk about people's personal experience and particularly life lessons, and navigating these tough and competitive worlds. And I tried to learn a little bit more about your background because from the outside, boy, have you been successful. And, boy, do you do a lot of cool stuff. And I don't know where you find the hours in the week to do what you do, but I salute you. But to the extent that you're willing to talk about it, how did you get to where you are? And did you ever go through periods, like the periods that I went through where there were many years where I didn't really know what I wanted to do, and I was sort of doing a lot of work and not getting very far and so forth? Your career looks just up and to the right, and maybe there's some lessons you can give us there.
- [59:47]
B
Yeah, I mean, it's interesting it looks that way because that's not the way that I experienced it. Henry. I mean, I would say that I had a very normal childhood. My family was truly middle class. My dad was a Methodist minister and my mom was an editor at a religious book publishing company. And I lived in small towns outside Nashville, Tennessee. And then eventually in Nashville for the first 15 years of my life, moved to Ohio for my mom's job when I was 15. She became a publisher at Pilgrim Press in Cleveland, and I went to high school in Shaker Heights. And then I went to state school, Ohio State. Go Bucks. And eventually, and then I got my master's degree at Berkeley because I wanted to live in California. I thought I wanted to be a lawyer, and I took the lsat, and then I decided to work at a law firm in D.C. and that didn't really work out for me. I didn't love it that much. And so I decided to go to policy school. I was going to work in the federal government. And then when I got to Berkeley, I worked with some professors and they were like, hey, have you ever thought about research? This seems like something you might be good at. And I was like, oh, no, not really, but maybe I'll try it. And then they wrote me good letters and I got into grad school. And so it was like I could tell a story that packages that all in some narrative, but the Truth is, I had no idea what I wanted to do with my life until I was like, 26. And even then, I wasn't 100% sure. When I went into my PhD program at Harvard, I wasn't sure. But the thing is, I had a lot of different jobs. I worked in landscaping, I. I worked as a waiter, I worked in construction. All for different periods of time. I worked. I just did a lot of things because I didn't know what I wanted to do. Then when I found research, I was like, wow, I love this. I'm just a nerd, Henry. That's the bottom line is I'm just a nerd and I love it. It doesn't feel like work. And I've been doing. I've been interested in the same broad types of questions for 20 years, and I've just been plugging away at them. And I guess what I would tell your listeners is that something I didn't know at the time, which is they're just huge returns to being an expert on something, to knowing more than almost anyone else about a topic that other people care about. And you can't do that overnight. But if you just pick your corner, you don't need to be the best at everything. If you pick your corner of the world that matters to you and you just learn everything about it, eventually people come calling and they want to know about it. And then all the good things in my life anyway have flown from that. That like I basically, in 2005, when I started the PhD program at Harvard, I decided I was going to work. I was going to be an economist, and I cared about social issues and education, and I just read every paper and talked to every person and went to every seminar for 20 years, more or less. And eventually I learned everything there was to know. And then I started writing things that other people wanted to hear about. And so my story is really just a very simple one. So I'm just kind of a normal guy who geeked out about something for long enough that other people wanted to hear about it. That's really all there is to it.
- [62:32]
A
I think a lot of folks are going to be very relieved to hear that amid this incredible career that you've had, there were, there was, at least, especially in the early 20s, that period of being a waiter in construction, both jobs that I have done. And actually, ironically, listening to you, I found myself to the same conclusion, which is I loved research, too. I happened to do it on Wall street as opposed to academia. But I think that observing that the lessons that I would draw, too, is you did a lot of things, and you probably figured out what you were good at and what you liked. And to me, having talked to a lot of people on this, I think back on my own experience those early years. The most important thing to figure out is what you are naturally good at relative to other people and what you naturally love to do relative to other.
- [63:21]
B
People and what the world is willing to compensate you for. It's those three things together.
- [63:25]
A
But if you figure those two things out, you can actually usually find your way into the compensation.
- [63:32]
B
That's 100% right. That's 100% right. And I think I always tell, like, I tell students here that even though that period in your life when you're not really sure what you want to do, it can feel like it's lasting forever and you have this urge to get closure. And I tell people, if you think there's some moment when you're going to know, okay, I did the right thing. Like, I figured it out, and now I've got all the answers. This never comes. Like, nobody knows what they're doing. Everyone's just muddling through, okay, nobody knows anything. And it took me a long time to realize that, but once I did, I found it very empowering because it's like I'm just making stuff up. And in the course of this interview, I probably said some dumb stuff, maybe I said some smart stuff, but you just give yourself a break. And if you're driven by the intrinsic love of what you're doing and just trying to understand and explore, good things generally happen.
- [64:14]
A
All right, so if your students come to you and say, yeah, thank you, fine, and I'm impressed that you found your way through it, but you weren't competing against this ubiquitous super intelligence that can do everything better than you. What do I do? In the name of AI, what is your advice to your students?
- [64:31]
B
I mean, I think. I don't think I would change anything about what I just said except understand that AI is a tool that can help you do it much faster. Like, I think I could have gotten to the, you know, frontier of knowledge in my field much faster if ChatGPT existed, you know, when I was in graduate school now, so could have other people. So it doesn't follow the. Therefore I could have, you know, become an expert. But you have to understand that, like, that's just a tool, and it doesn't really. Like, these tools are not going to tell you what to do with your life or what to specialize in. I still think There are, if anything, even higher returns or higher value to really geeking out on something and then also having good, you know, people skills and like caring about others and doing the right thing, those things are all really valuable laid on top of a true expertise in something.
- [65:16]
A
And what about advice to folks who are relatively junior in the workforce at a law firm or another office job and had been looking at this career that seemed very safe and self assured. You do well and you progress and suddenly again there's this new technology that maybe is threatening that future. What's your advice to them?
- [65:37]
B
Here's some advice for young people that you don't have to take seriously, but I think you should. Okay, so one of the things I discovered when I became, I arrived in positions of authority, I became a professor instead of a student and then I got tenure and then I started like all those things is that the things that I was doing like answering emails, promptly, doing my homework on someone before I met with them, being interested in them and not just counting on them to be interested in me. I thought of those things as like the basic blocking and tackling of everyday life. And I think partly that was because of my upbringing, like you know, growing up in a church oriented household, like I had some sense of doing the right thing and like it's being respectful of people's time and all that stuff. I don't know where it came from, but that's I think what it was. I thought everyone did those things. And then when I got on the other end of it, I realized that actually a lot of people don't do those things. And so if you just show up, are totally engaged, do what you say you're going to do, you're reliable, you follow through, you're the kind of person people want to talk to. You know, it's not rocket science, it's just being reliable and interested and open minded like that gets you 90% of the way there. It really does. You will stand out just by fulfilling your obligations and doing the things people expect of you. And I would just say to people, don't get that wrong, because it's free money, it's money lying on the sidewalk. Be the kind of person that people want to talk to, that comes prepared, that respects people's time. I know that sounds like an old guy giving young people advice, but I'm just telling you, a lot of people your age won't do it. So if you do it, you have a leg up. So it can be totally self interested. It's not to flatter my ego. It's really not. It's just to get a leg up for yourself. And it's easy. It's much easier than being an expert or whatever. So that's my advice.
- [67:16]
A
What I'm reminded of, as you say, that is going back to what you're saying about social skills. It's the interpersonal relationships. And you reminded me of it. And just because it's so old that people probably amused. One of the best how to Live Life books out there is still Dale Carnegie's how to Win Friends Influence People. It's a terrible title. It sounds totally manipulative and so forth. But the main message of it's exactly what you just said, which is, you know what people like. They like it when you're interested in them. Not when you sit there and give a speech about yourself, but when you're actually interested in them.
- [67:52]
B
I tell my kids, too. Like, they say, you know, oh, they're teenagers, so they're like, oh, social interaction. It's so awkward. What do I say? And I said, listen, whenever you're in a conversation with someone and you don't know what to say, ask them about themselves, because that's everyone's favorite subject. And yes, that sounds cynical. You're just trying. But it's really just when I do that, I find that interesting things emerge and I find common ground with people, and the conversation doesn't become about them, it becomes about both of us. But the impetus is me asking about them. And so that's one concrete example. I read that book every five years or so just to remind myself that's how important I think it is.
- [68:24]
A
It's tremendous. Professor Deming, this has been so great. Thank you so much for sharing your time and expertise with us. And I'm actually more optimistic that I'm with us at the beginning of this conversation.
- [68:35]
B
One conversation at a time. We're changing the world. Thank you so much, Henry. It was great to be here with you.
- [68:39]
A
Thanks.
- [68:39]
B
All right, take care.
- [68:42]
A
Solutions is produced by Meghan Cunnane. Jim Mackle is our video editor. Our theme music is by Trackademics. Nishat Kurwa is Vox Media's executive producer of podcasts. Thanks for listening to Solutions from the Vox Media Podcast Network. I'm your host, Henry Blodgett. We'll see you soon.