[MUSIC PLAYING]
Hello and welcome to Data Nation, a podcast from MIT's Institute for Data Systems and Society. I'm Liberty Vittert and I'm here with my co-host Munther Dahleh, the head of MIT's Institute for Data Systems and Society. In this episode, we chat with Dr. David Autor. David is the Ford professor of economics at MIT.
With all the chatter about ChatGPT and the development of artificial intelligence, we contacted David to get some real talk about the current state of AI and what we might expect as the technology continues to advance. Well, thank you, David, for doing this, first of all. My pleasure. Thanks for inviting me. So this is a very hot topic. And the whole idea of AI, and the future of work, and the workforce is something that is on everyone's minds.
I'm going to start very broad and ask you about your opinion. Where do we stand today? It's like, what is the general comment about where AI is in terms of the workforce and what we should be expecting in the future? Yeah, I generally don't like to say, we're at an inflection point. Things are changing faster than they ever have, et cetera. But we are at an inflection point, and things are changing really rapidly.
We've obviously had computers all around us for the last four decades if you've been around that long. The office personal computer, the IBM PC was introduced in 1982, and that was a big deal. And people thought that it's going to change the world. It did change the world in some ways.
But the progress of computing over the last four decades has been very linear in the sense of, if you want to get a computer to do something, you first had to make explicit all the rules, and formulas, and procedures. And then you got to write that up in code. And then you can have a machine that didn't understand what it was doing and was not flexible or adaptive carry out those rules and procedures.
So it was like you had to basically, first, put everything on ice and figure it all out and then have a machine do it. So we call that codifying routine tasks. And so that's really good for office work. It's good for repetitive production work, anything that's calculating filings, sorting, storing, retrieving. But the key thing is, all knowledge had to be made explicit. It had to be formalized and codified before a machine could execute it.
And that's actually very different from the way that we do things because we do many things without understanding what we're doing or knowing the rules. We use what we call tacit knowledge. So you never went into a classroom and learned how to ride a bicycle there. And if someone asked you to teach a class on riding a bicycle, you wouldn't know where to begin. But you can ride a bicycle.
And similarly, if you're trying to tell a funny joke, or make a persuasive argument, or recognize someone who you haven't seen in four decades, you don't know how you do that, but you know how to do it. But the fact that you didn't know how to do it meant it was very hard for you to write a computer program to do it because the rules were tacit, they were not explicit.
And so the progress of computing was very predictable because we knew this was a very incremental procedure to go from playing chess, to spell checking, to checking grammar, to calculating trajectories, and all kinds of stuff. So AI is fundamentally different from prior generations of software and computing capabilities because you don't need to explicitly know the rules. You can say, look, here's the question, and here's the answer.
And now, you tell me, you figure out how we got from there to there. So here's a billion objects. They're labeled as such and such. Now, I'm not going to tell you what makes a chair a chair. You just look at the thing and tell me which of these things is a chair. And it turns out, that capability has become very general.
So it doesn't just apply to recognizing objects, but generating language, generating faces, predicting the text you're going to write as you write an email, and parsing huge bodies of knowledge and pulling out relevant things. So that is extraordinary. And many people, including our colleagues here at MIT, didn't think we would get here this fast. I remember 10 years ago, there was a group of computer scientists and economists who used to meet rather frequently.
And it was just at the beginning of the contemporary AI spring, after decades of AI winter. And some people were saying, well, look, this thing, it's not going anywhere. We'll see what these things are doing. They're just going to get the right answer on average and get every interesting case wrong. We're going to hit the flat of the curve. And other people were saying, wow, it is this new AI. I mean, I'm surprised.
I didn't think you could get this far just based on this notion of these neural nets that are just assimilating statistical associations and then doing things. But it does seem to be really working better than you'd think. And now, I think we're at a point where a lot of people are saying, wow, this is gone further, faster than we thought was possible.
Whether you're looking at ChatGPT, or you're looking at how fast computers can do translation between languages, or even if you're writing an email, the fact that it seems to be able to predict what you're going to say next, that's pretty extraordinary. So I do think this is a time of substantial uncertainty about what's feasible. And the reason I say uncertainty is because, I said a minute ago, we have this linear roadmap, like how do we get from here to there? Well, we know the route.
It's a long one. It's a slow one. And now, we don't have a roadmap. I think when I ask computer scientists this question, what can you confidently say AI will not be able to do 10 or 20 years from now? They don't have a very compelling answer. They're really not sure what it won't be able to do. So I think it's a time for a lot of blue skying and I think one should be hesitant about saying what won't happen.
I think one should also be hesitant about saying, you know what will happen, because I don't think anyone has a very, very clear picture. David, I can't help but hear the excitement about what's happening around AI, but without bringing up the fear, and that the jobs are going to be taken away, that the workforce is going to be lost, and the real fear that people have around AI taking away jobs.
But in your Ted Talk, and correct me if I paraphrase you wrong, you mentioned this really interesting phenomenon that has human workers who've almost had increased opportunities and capacity to be employed in the face of increasing automation, which is what no one thinks is happening, but, I guess, is. And you had this great example with bank tellers and ATMs to explain how automation doesn't necessarily mean less jobs, it could mean more.
So could you maybe explain how that works and why we shouldn't necessarily be as scared as I think a lot of people are? Sure. Absolutely. The example of bank tellers, by the way, goes to Jim Bessen, who's an economist at Boston University, which I did use with attribution in my Ted Talk. And is the case that decades after ATMs were introduced, people were very surprised to discover that the number of bank tellers had increased, not declined. And the answer is, well, why?
We all thought they'd be automated by now. What are they still doing here? And the answer was a couple of things. Two pieces. One is that, because ATMs lower the cost of opening a branch, banks began branching much more aggressively. And so they would hire a couple of ATMs and a couple of bank tellers. And boom, they had more people. But the other is that the bank teller job changed.
It went from being less of a dispensing machine for cash, and much more bank tellers became salespeople who were selling accounts, credit cards, loans, not always for the better. And so the job changed. And so it moved away from people doing the most routine cash-handling transactions, and much more to adding value through these customer interactions. Now, it is the case that the number of bank tellers is now in decline. So this, it's not a universal law.
But I think there are three ways to think about, well, why doesn't automation just always replace us? We have automated ourselves to a huge extent. 40% of employment was in agriculture at the turn of the 20th century. And now, it's under 2% And it's not because we're eating less, it's because we've become so much more productive in agriculture. And so very few people need to do it. But there are few things that work in a different direction.
So one, of course, is that a lot of what automation does is it raises productivity. And when we have more productivity, people have higher incomes. And when they have higher incomes, they spend more. People are pretty insatiable. People's consumption rises at about 1.01 the rate of their incomes. And so no matter how wealthy we get, we seem to never run out of wants.
And so that consumption itself creates demand, and not just demand for more of the same, but for more experiences, more services, more luxuries, more travel. And a lot of those things are labor-intensive. A second thing to recognize is that many of these things that we think of as automation are actually just tools for us to do what we do better.
So whether you're a writer, or a researcher, or you're a roofer using a pneumatic hammer, the last thing you'd want is for someone to take away the tools that you use to perform your basic work. So many times, the technologies are very complementary to what we do because they allow us to focus on what we're really good at and get rid of some of the boring, repetitive stuff.
A third, and this is the hardest to anticipate, is that we often use these technologies to create new goods and services that demand labor. And some of those things are having to do with technology itself. Obviously, you need AI programmers, you need people who handle the hardware, and so on. But often, it's something completely different.
It's a set of luxuries that are made feasible by new tools, whether that's like a virtual environment, whether that's everyone buying a personal assistant, whether that's everyone using the extra money to have new experiences. Many of the jobs in which people work now are things that just didn't exist 80 years ago. There's all kinds of medical specialties. There's all kinds of new types of engineering, but there's also new types of sommeliers, and counselors, and coaches.
And so in work with my coauthors, Caroline Chin, Brian Seegmiller, and Anna Salomons, we estimate that about 60% of the work tasks that people do in 2020 were not present in 1940. So there is a lot of new work that's created. So I see no reason to think we're going to run out of jobs. But that's only half the story. And the other half is, well, what jobs? What will people do?
And there, it's very important to ask whether the new technologies that we create, are they going to make our labor more valuable, scarcer, or are they going to compliment our expertise, our creativity, our judgment?
Or are they going to commodify us and make us the last mile of the task, where the machine does the hard part and the person is just the one picking the thing off the shelf and putting it into a box, or the person just makes sure that the truck, as it comes off ramp, it doesn't run into a post, but most of the driving is done by a vehicle, and so on? And that's a real concern. And we've seen both happen over the last 40 years.
So people who do professional, and technical, and managerial work over the last four decades have been strongly complemented by computerization because they've made access to information and computation research much cheaper and faster. If you're a doctor, or if you're a lawyer, or if you're someone who has to oversee a larger organization, having really cheap access to information and calculation processing makes your labor more valuable. It allows you to do what you do faster.
However, if you were an office clerical worker, or you were a person doing repetitive production work, many of those jobs have been eliminated. And the people who did them, if they didn't have the opportunity to move up into the professions, let's say, many of them end up doing rather generic work. So food service, cleaning, security, entertainment, recreation. There's nothing wrong with that work. It's valuable work, in fact.
However, because many people of sound mind and body can do those jobs, they're not well-paid. The labor is not specialized, it's not scarce. And so that's the concern with these technologies, not that we're going to run out of work per se, but that they will reduce the value of specialized skills that people have. And the balance between how much they complement, how much they substitute, and for whom, that is the area, I think, for legitimate worry.
David, this is actually really interesting and raising many, many questions in my head. So I'm going to go back a little bit to your inflection point. And I remember, when I was a kid, and for the first time, they allowed us to use the calculator in a math test, my grandfather was appalled. He's like, if you use a calculator, what's left of the problem. And I think that a lot of it was about the definition of what is routine, and what is creative, and what is problem-solving.
And I think you were getting to the point where, at that point, with the inflection point we're at, by saying that we're not really clear at this point, what is actually routine in what we do? Stuff that we thought was actually creative now looks like it's routine because a machine is able to do it and does it in a systematic way. And that set of routine things is growing.
And so I think some of the fears that we're having right now is, how much creativity do we have, and is there a limit to this? So it is those jobs that define us as humans, the ability to rise above any other machine and do something that no one has done before, and you thought a new idea, I think there's another concern that machines are going to replace us there. And I don't know what your thoughts are about this.
Yeah, so it is absolutely the case that the domain of things that could be done by machines is expanding, and probably, arguably, expanding very, very rapidly. Now, creativity is a high standard to hold most jobs to. I don't think most-- we all use creativity in various ways, but most of what we do is not creative. Most of what we do is accomplishing thing that needs to be done, and that's still always been valuable.
So when you say, I'm doing agriculture, I'm doing farming, I don't know if creativity is the top thing on your list, although there's creativity involved. But you'd say, look, I'm doing a lot of things that can't be done any other way. And it's important work. It needs to be done. And then I think we also have to ask ourselves, and I think this is implied by your question, what is creativity? What does that actually mean? And you'd think of, well, writing a new paragraph, well, that's creative.
You have to produce a new sentence, and a new idea, and you got to structure it. And then we're surprised to discover that that can be simulated, replicated by machines. And a lot of what we do is actually, at some level, pretty predictable.
Now, I would say that there are important distinctions between what AI is doing and what people are doing, and not just how they do it, which obviously is pretty different because we don't have the same level of horsepower for doing certain things, but we do other things much more efficiently. But what AI is doing is-- or something like ChatGPT, in particular, different AIs do different things.
It's doing a lot of prediction of what you would expect to see, what sentence should follow what, what things should go together. It isn't doing another thing that we do a lot of, which is verifying against fact, verifying against logic. So it actually is missing a very important part of cognition. You can say to ChatGPT, how many funerals did John F. Kennedy attend after he was assassinated? And it will tell you, it'll give you a list, and it won't be zero. It'll just make something up.
And that's because there's no model in ChatGPT that says the world works this way. There is such thing as time, and causality, and there is such thing as rules of mathematical operations that constrain what can go together. It's actually funny. It's actually quite ironic that the frontier of computers are machines that are bad at facts and numbers. You wouldn't think that was where we'd end up, but that's where we are right now.
And so I do think there's a lot of things that you can potentially use ChatGPT for that it doesn't do very well and need, at this point, to be disciplined. You can ask ChatGPT to write a syllabus for your class. It'll list papers that don't exist by people who would be likely to write them.
So you can ask yourself, you say, well, maybe one zone of complementarity is I can use the machine for the first version of something to write the draft of something or help me produce ideas for X, Y and Z. But then, I could add a lot of value in that setting. But we don't know how much. We don't know where that complementarity comes in versus substitution. And of course, ChatGPT is just an example of a technology that's advancing quite rapidly.
As we think of what more and more AI can do, the one thing that always comes to mind, especially in popular culture, is AI becoming human in some capacity, or having human intelligence, or human emotions. Do you foresee AI becoming more or becoming emotionally intelligent in the future? Is that possible?
And would we be able to overcome our mistrust towards AI's lack of ability to socially interact or be able to more carefully explain how it comes to decisions, rather than these black box algorithms where no one really knows what's going on and we have all the bias issues that we see? But is that possible, what we see in the movies? OK, so I'm not a computer scientist, so I don't want to overextend my expertise here. This is all hearsay about what it can and can't do.
I think we should ask whether we want it to have that capability. It's not obvious to me at all that we do. There's a real model in the kind of AI computer science community of replicating human capabilities with machines. But we already have human capabilities. If we just have machines at human capabilities, is that the best we can do? Maybe we should think about what we could use them for that we can't do. So having them simulate the emotions of my teenage kids, I don't know if I want that.
But there are a lot of things that AI could do that could be very valuable. So, for example, we could use it to make education more immersive, cheaper, more accessible. We could use it to lower the cost of medical care and bring expertise to more people. We could use it to augment people who are doing skilled work and allow them to do a broader set of tasks. And when I say skilled work, it could be construction, or diagnosis, or repair, or maintenance.
It doesn't have to be-- school work, I don't mean to evoke the image of people at MIT sitting at their desks. I mean, there's a lot of skilled hands-on work in the world in which we bring foundational skills, and then we want to be able to broaden our expertise. I'll give you a very concrete example. If you want to change some plumbing in your house or rewire something, you can go to YouTube and you can watch video that says, this is how you change the plumbing. This is how you do the wiring.
Most people shouldn't do that. They're going to drown themselves or electrocute themselves. Not a good thing. But if you have some foundational skills in dealing with plumbing or home electricity, you could then use that video to help you do more things. So it's a compliment to a basic skill set. So you can imagine using AI to enable people to do more things.
So that could be, I'm a jet engine repair person, but I've always trained on Rolls-Royce engines, and now, I need to work on a Pratt & Whitney engine or GE engine, would be an example. In medicine, we have this phenomenon of tasks are always going to the most expensive person in the room. Whatever it is, it always moves upward. It's got to be done by the doctor. And that's very difficult. That's very expensive. Those doctors, there aren't that many of them. They get paid a lot.
Why can't we devolve some of those tasks to people who have expertise, but are slightly less rarefied? And they could do a lot of them. In fact, this has happened in medicine. So a lot of tasks have been moved from only being done by the primary care physician to being also done by the nurse practitioner. And you could imagine that you could use AI to enable more people with foundational skills to do a broader set of valuable things. And that would be terrific.
And here's a point I want to emphasize. My colleague Daron Acemoglu makes this point very often. What AI does is not up to AI, at least not at the moment, it's up to us, where we want to invest in it. Where do we want to put it? So China has the world's best surveillance state, they brag, and I don't have any reason to disbelieve them, they can physically locate any of one of their 1.3 billion citizens within an hour. They also have the world's best content filtering system.
They can delete stuff as soon as you post it, even before it goes out on the web. Now, that's an impressive technological achievement. It wouldn't be feasible without AI. But that's not because that's what AI does, that's because we're trying to put its money on AI. That may not be the highest use. And so what capacities we develop depends a lot on what we prioritize, what the incentives are, and what people imagine they're supposed to be doing with this technology, is highly malleable.
I really like your point. So I think it's definitely true that we are now in control of how we build these systems, but we do a lot of that without guidance. There's TikTok collecting data about me, flipping through these videos, and so forth, and building a model of who I am so that the next thing, they can sell me something. I mean, this whole system isn't necessarily targeting social responsibility, but doing a lot of random things.
The science fiction story is that we are at a battle between us and machines, and someone is going to win at the end. And part of that, I think, is not so much that the machines become so intelligent, but that we become stupid because we are becoming followers of what the machines tell us to do. I follow the GPS. I don't even look at the sites anymore where I am. I no longer know where I am. I'm just looking at this little device. The same thing you described in medical profession.
There are situations now where the doctor never looks at you, I mean, actually, constantly looking at the screen, looking at the next suggestions of what they do and they're not looking at the patient. So this equilibrium. I mean, how do you think we're evolving, and how do we intervene to make that equilibrium something that we would like to have? Yeah, so one way to say this-- I mean, I think are several things you're saying, but let me try a couple of them.
One is, and this goes back to [INAUDIBLE] previous question, is we often don't know what the machine is doing, as she said. It's doing a thing, and it's a black box to us. That's a very hard problem to solve. I described earlier the fact that we have all this tacit knowledge, and we don't know how to codify it and write it in software. Machines now have all this tacit knowledge that they don't know how to communicate to us either.
You can't just look into someone's head and look at the neurons, and say, oh, I see what they're thinking. The information is there somehow, but we have no idea how it's represented. Similarly, with AI, it's just a billion weights on different connections. You could look at that all you want. You could stare at it for years. It wouldn't tell you what the thought process is. So it is opaque to us. And why is that a problem? Well, one is it makes it hard for it to predict.
You can imagine, every once in a while, a self-driving car does something crazy and we don't know why. The other is, it's disempowering or it makes it hard for humans and machines to interact, because where do you develop the expertise in the practice if you're so reliant on the machine? So for example, we know there have been horrific air disasters because, basically, autopilot stopped working and pilots didn't know how to fly the systems without them.
And this is why you might have the same concern about writing. So now, every time you want to write something, you'll just mumble it into ChatGPT and it'll turn into something coherent, and then you can work on it from there. In some ways, that's good. We wouldn't want someone to not be able to do engineering calculations because they weren't good at division and multiplication. Have a machine do that. But it is a concern.
Will we be able to do the higher level tasks if the foundational tasks are not things we master? So I don't really need my kids to learn times tables anymore, at least not memorize them to the extent they used to, but they do need to understand the basic operations of addition, subtraction, multiplication, division to do the rest of math. So it is a concern how these things will interact.
So I think there's an optimistic story that's often told that says, oh, well, human and machine together are better than either one of them separately. And first of all, that may be only true for a time. And second, humans are generally more expensive than machines. So it may be that the human machine is better ignoring cost, but you may say, well, after accounting for costs, I think we'll just take the machine. So it is hard to predict where this goes.
And I try to discipline myself on this, too, in the sense of I know that my foresight about what will be created and augmented is much less good than my foresight on what is going to be substituted. And I can predict lots of things that will be substituted. I wanted to bring it back to something both of you mentioned. Munther, you said it, this idea of robot against person, but it almost seems, in some ways, it's going to be person against person using AI or country against country using AI.
And David, you brought up China, who's decided to really invest in AI to be a surveillance state. A lot of the facial recognition systems they're using came from US Tech, and they're using it against the weaker people. We've seen these very terrible consequences. And so in a world where we're trying to create something new, this new AI, where in a perfect world, we'd be able to cross borders and share IP, what's the recourse to enforce this cross border IP protection?
What could be the collateral costs that we're going to have to accept, and how necessary is it going to be to protect national defense, whatever country you are? Because it does seem like it's almost going to be country against country building these robots. Yeah, I mean, I think, I agree. It's really people against people. It's not machines against people, at this point, but people against people using machines.
And the concern is that the technology is not expensive and it's not that hard to get it to do what you want. So you can make an analogy, oh, well, in the nuclear age, we had very successful nuclear containment and we have had that. For over seven decades, there have been no additional uses of thermonuclear weapons. And even countries today, after many years, are struggling to produce the technology. And so it's been an incredibly successful regime of control of a dangerous potential weapon.
And AI is not like that because hardware is cheap and getting cheaper. The tools are easy to access and getting easier. The barriers to entry are really, really, low, so it's extremely difficult to contain it in the way you might want to if you were very scared about it.
And I do think we're going to face a lot of challenges, and not just at the level of country versus country, but even at the level of fake news versus real news, at the level of misinformation, at the level of persuasion, of overcoming people's barriers to trust when they shouldn't be overcome. I do think we've produced a technology that we're not necessarily especially well-equipped to immediately handle its repercussions. Now, of course, we've said this about many technologies in the past.
People thought TV would rot everybody's brains. People thought the Walkman, the little personal stereo system, would turn everyone into the urban zombie who was totally tuned out from the environment, and so on. That hasn't happened. But I do think social media has had real consequences that we haven't handled especially well. And so I think there's reason to think it's going to take a while for us to figure out. And it is another irony.
I mentioned the irony that we have machines now that are great at language and are terrible with numbers and facts. And it is the case that we live in an era where, with all the information, more information at our fingertips than the world has ever had, we seem to be less certain of truth as a consequence of our ability to manipulate. And then at the country level, I mean, the thing obviously that the US has recognized as was the lever point here is semiconductors.
This is the thing that all of this runs on. The whole world, at this point, runs on semiconductors. And it is still the case that semiconductors are arguably the most intricate technology that humanity has ever produced. And the leading semiconductors of the day basically can only be produced by one company, which is TSMC. Morris Chang was an MIT student. Someone who's at the building I'm sitting in is the Morris & Sophie Chang Building.
And it uses another machine that's only made in the Netherlands, which has a set of suppliers, 300 deep, based all around the world. So if the leverage point that I think the US has correctly identified is control of production of semiconductors is the way to limit others AI capacities, if that's your objective. So David, just to bring it home to MIT and to other educational institutions, there are many aspects of educating the masses is going to be critical.
One aspect, obviously, is what you described, which is automation or technology displaces people from certain jobs, but then you create new jobs. Some of the people you displaced are not ready for the new jobs. And I think the countries that do very well are the ones that educate the people ahead of time or at least anticipate so that the new cadre now is able to tackle the new jobs and so forth. How can universities, MIT and others, manage this educational piece?
Yeah, so some of it is education and some of it is assistance. So if you look across industrialized countries, everybody loses jobs. And job losses is damaging in many ways, both economically and psychically. And those are not independent. But the cost of job loss varies a lot across countries. It's much higher in the United States-- I mean, for an individual being displaced than it is, for example, in Denmark or Norway.
And it's because we do so little to assist people who have been displaced either financially, or in terms of retraining, or even the psychological supports that are necessary to get back into the workforce. So part of what we should be thinking about, some of it is exactly as you said, it's preempting, it's investing ahead of time, and some of it is assisting when the inevitable occurs. So I think we need to think about both.
But then in terms of going back to the education questions, I think it starts well before university. If you said, well, I have to redesign the educational curriculum from the ground up now because so many things that we used to think are important are now being done by machinery. Some of the things that we teach are archaic, like lots of memorization of facts and tables of numbers.
Some things actually are as valuable as they've ever been, which is quantitative reasoning, communication, the ability to adjudicate around facts, to reach conclusions, to form hypotheses. So what do people need to do? They need to be able to read. They need to be numerically literate. They need to be analytically literate. And that's something that we haven't historically taught. Now, it used to be the skill was how to go to the library and find facts.
Now the problem is how to adjudicate and throw away most of the facts that are available to you, how to you actually assemble them into something useful. We've gone from a world of information scarcity to information abundance and the scarce capabilities, the ability to organize that information, to organize that information. And by organize, I don't mean sort it alphabetically. I mean, tell a story, draw an inference, produce a hypothesis and say, well, what facts fit with that.
And then, of course, to communicate with that to people about it verbally in the form of presentation, in the form of writing, even writing assisted by ChatGPT, and to lead others. So those skills will continue to be highly, highly valuable. So we've got to start with that, foundationally. And I think that's a K through 12 thing.
Colleges and universities are-- first of all, the majority of adults don't have a four-year college degree and will not have a four-year college degree in America in the next several decades. At this point, under 40% of the labor force has a four-year college degree. And that's rising very slowly. So I think there's a lot to be done at the level of vocational training. If we say, well, what's more at risk from the advances in AI?
Is it blue collar construction, trades, repair, electrical work, plumbing, or is it people who are doing management tasks? I would much more say the latter. So we should be investing in allowing people to use those other skills, a lot of skilled vocations. And then in terms of colleges, I think this is where you want to reinforce those analytical skills, as well as expert skills.
And there's often a view that if you give people a lot of expertise, you're making them, in some sense, that's so specialized, maybe it's narrow. But in general, expertise is complementary to other things. The more you know, the more you can use that across other domains. We don't want to say, oh, we over train these doctors. They're too expert. They can't do X, Y, and Z. Usually, the more you know, the more you can learn and apply.
A work by David Deming, an economist at Harvard who has looked at the value of social skills versus formal technical skills, and many people think, naturally, oh, STEM has become so important. It's all about STEM. It turns out that the fields that managerial skills, at least in the last couple of decades, have become more important, but particularly, interpersonal skills that combine with technical skills.
So it's not the person who is just sitting in the back room, crunching numbers, but often, the person who is translating between a body of expertise and other people. So a lot of what doctors do is they don't just look at charts, but they communicate with patients. Or if you're a contractor designing houses, it's not just engineering calculations, it's figuring out what the person needs. And that's true if you're an attorney, if you're a counselor, if you're a marketer.
So at that interface between the technology, and the domain expertise, and then other people and their needs, that's where there's a lot of valuable things to do. So when you think about where are people most useful or where are they most valuable, they're useful in many ways, but where are they most going to be paid the most, it's where they can bring specialized expertise to bear, not just in a technical sense, but in a translational sense, in an adaptive, interactive sense with other people.
Thank you for listening to this month's episode of Data Nation. You can get more information and listen to previous episodes at our website, idss.mit.edu or follow us on Twitter and Instagram @MITIDSS. If you liked this podcast, please don't forget to leave us a review on Spotify, Apple, or wherever you get your podcasts. Thank you for listening to Data Nation from the MIT Institute of Data Systems and Society.