Latest / What's Your Problem? / How AI Could Reduce Inequality

Transcript

How AI Could Reduce Inequality · What's Your Problem?

• 0:15
  Speaker 1: Pushkin. I'm Jacob Goldstein. This is what's your problem? And
• 0:32
  Speaker 1: my guest today is David Otter. David is a labor
• 0:36
  Speaker 1: economist at MIT, and among other things, he's done really
• 0:40
  Speaker 1: influential work on how new technology affected American workers over
• 0:44
  Speaker 1: the past several decades. In particular, his work makes a
• 0:48
  Speaker 1: compelling case that computers contributed to rising inequality in the
• 0:53
  Speaker 1: late twentieth and early twenty first centuries. So I was interested,
• 0:58
  Speaker 1: and I was pleasantly surprised when I recently read some
• 1:01
  Speaker 1: new work of his that suggested that AI might actually
• 1:05
  Speaker 1: drive inequality down. It's a nuanced story and it's a
• 1:11
  Speaker 1: really interesting case that he makes, And of course we
• 1:14
  Speaker 1: don't actually know what's going to happen, but whatever happens,
• 1:18
  Speaker 1: David Otter's framework is a really useful way for thinking
• 1:21
  Speaker 1: about the way technology affects work, affects who does what,
• 1:26
  Speaker 1: who makes how much money? So to start, I asked
• 1:29
  Speaker 1: him to talk about what happened when computers started to
• 1:33
  Speaker 1: become common in offices back in the nineteen eighties, so you.
• 1:37
  Speaker 2: Know, information technology, The computer was different from many other
• 1:42
  Speaker 2: technologies that preceded, and what made computers different traditional computers
• 1:46
  Speaker 2: like PREAI computers, is that they followed rules and they
• 1:51
  Speaker 2: could you could write a program to accomplish a task
• 1:56
  Speaker 2: that you would previously have thought of as a cognitive procedure.
• 2:00
  Speaker 3: Right.
• 2:00
  Speaker 2: It required all these ifens, all these steps gathering information,
• 2:06
  Speaker 2: doing calculation and making things that we thought of were
• 2:09
  Speaker 2: you know, exclusively in the minds dominion, right, And so
• 2:13
  Speaker 2: that that codification might took out a lot of the
• 2:17
  Speaker 2: routine tasks in the work that people did. How that
• 2:20
  Speaker 2: affected work was really quite nuanced, right, So people who
• 2:25
  Speaker 2: did bookkeeping and accounting, right, they spent much much less
• 2:28
  Speaker 2: of their time on data you know, tabulation, data calculation,
• 2:33
  Speaker 2: and much more on scenario planning right on forecasting, on
• 2:37
  Speaker 2: you know, forensics, trying to you know, find errors and
• 2:40
  Speaker 2: so among technical, professional managerial work, it made work more
• 2:44
  Speaker 2: productive and it made more specialized. It eliminated a lot
• 2:46
  Speaker 2: of the grunt work and enabled people to focus on
• 2:50
  Speaker 2: the abstract problems. But for the other group of people
• 2:53
  Speaker 2: whose work was simplified, whether it was in the kind
• 2:56
  Speaker 2: of you know, uh, stock and inventory clerks, whether it's
• 2:59
  Speaker 2: people who were doing who were working as cashiers who
• 3:01
  Speaker 2: used to have to know, you know, how to be
• 3:04
  Speaker 2: able to make change and you know, keep traffing numbers on.
• 3:07
  Speaker 3: It made the work simpler and less specialized.
• 3:10
  Speaker 2: And in the case of production work, and you know,
• 3:13
  Speaker 2: it eliminated certain types of production jobs, certain types office
• 3:16
  Speaker 2: jobs because they could be more or less fully automated.
• 3:19
  Speaker 2: And that had the effect of taking people with you know,
• 3:23
  Speaker 2: high school degrees and some college and making their skills
• 3:26
  Speaker 2: less valuable in offices, less valuable in you know, in
• 3:29
  Speaker 2: factories and assembly lines. And so it shifted the weight
• 3:33
  Speaker 2: of employment for people without college degrees away from offices,
• 3:37
  Speaker 2: away from factories, and towards a lot of services food service, cleaning, security, entertainment, recreation,
• 3:44
  Speaker 2: home health aids. And that's socially valuable work, but it's
• 3:48
  Speaker 2: poorly paid. And there's a number of reasons why it's
• 3:50
  Speaker 2: poorly paid on oversimplified, but one of them is it's
• 3:53
  Speaker 2: not specialized. Right, people can do it without training your certification.
• 3:59
  Speaker 1: So one example that I find helpful in thinking about
• 4:03
  Speaker 1: the kind of dynamics you're talking about is looking at
• 4:05
  Speaker 1: accountants versus bookkeepers. Accountants focus more on the big pick
• 4:09
  Speaker 1: sure typically have a college degree, often a special license.
• 4:12
  Speaker 1: Bookkeepers focus more on the day to day look at
• 4:14
  Speaker 1: you know, money coming in and out of a business.
• 4:16
  Speaker 1: They don't typically have as much training as accountants, and
• 4:20
  Speaker 1: so I mean, can you just talk about how computers
• 4:23
  Speaker 1: affected accountants versus bookkeepers.
• 4:26
  Speaker 2: So for accountants, it was the story of essentially eliminating
• 4:33
  Speaker 2: the supporting work, the inexpert work, the drudge work, and
• 4:37
  Speaker 2: allowing them to focus on the cognitively demanding stuff right,
• 4:40
  Speaker 2: to do the problem solving, to do the forecasting, to
• 4:45
  Speaker 2: do the scenario planning, to do the forensics of finding
• 4:48
  Speaker 2: you know, errors and issues and so on.
• 4:51
  Speaker 1: Accountants can be better accountants because of computers. They can
• 4:54
  Speaker 1: do figure out more complicated ways to save people money
• 4:57
  Speaker 1: on their taxes or to find errors or whatever. And
• 5:00
  Speaker 1: so it's good for accountants yep, and other similar you
• 5:05
  Speaker 1: know sort of professionals with a high amount of expertise. Right,
• 5:08
  Speaker 1: and then the as a tool that's making them better
• 5:10
  Speaker 1: at their jobs. What is it doing to the bookkeepers?
• 5:13
  Speaker 2: So it's making your work like specialized less specialized, right,
• 5:16
  Speaker 2: and so many more people can do it, and therefore
• 5:21
  Speaker 2: it doesn't have to pay as well. It may grow
• 5:24
  Speaker 2: right because firms like well, look it's cheaper than it
• 5:26
  Speaker 2: used to be. It's more productive than it used to
• 5:27
  Speaker 2: be and lots of people can do it h and
• 5:29
  Speaker 2: we see this often, right, So like another nice another
• 5:32
  Speaker 2: example this is ride hailing hailing software the you know
• 5:36
  Speaker 2: the phone apps right, uber.
• 5:38
  Speaker 3: Uber, uberlyft Right.
• 5:40
  Speaker 2: They changed the work of taxi and chauffeur service in
• 5:42
  Speaker 2: number of ways. They made it much easier to enterance on.
• 5:44
  Speaker 2: But one thing they did is they entirely eliminated the
• 5:47
  Speaker 2: expertise that was required to drive, because you used to
• 5:50
  Speaker 2: have to know your way around.
• 5:51
  Speaker 1: A place to drive, famously in London, Right, there's a
• 5:54
  Speaker 1: test that you had to know every street in London
• 5:57
  Speaker 1: or something to be a cap driver in to get it.
• 5:59
  Speaker 2: Yeah, take about three years of memorization. It was an
• 6:02
  Speaker 2: incredibly rigorous feed Yeah.
• 6:04
  Speaker 1: And then you can now you can just get a
• 6:06
  Speaker 1: map on your phone. Like now, technoledge is just profoundly obsolete.
• 6:09
  Speaker 1: It's such a wild show, exactly.
• 6:11
  Speaker 2: It's it's totally economically stranded, right. It's it's still out there,
• 6:14
  Speaker 2: but it isn't useful. It has no scarcity value, right
• 6:17
  Speaker 2: because everybody else gets it on their phone.
• 6:19
  Speaker 3: So what did this do? It eliminated the.
• 6:23
  Speaker 2: Expertise requirement to do this work, but it made the
• 6:25
  Speaker 2: work very productive. Right, So we have you know, several
• 6:29
  Speaker 2: hundred percent more people in this walk of life in
• 6:32
  Speaker 2: driving show for service than we did twenty years ago,
• 6:36
  Speaker 2: but their pay, the pay of a taxi driver's showpers
• 6:40
  Speaker 2: has fallen relative to the average.
• 6:42
  Speaker 1: Right.
• 6:42
  Speaker 2: So this so this dynamic of simplifying work enabling more
• 6:46
  Speaker 2: people to do it but at lower rates of pay.
• 6:48
  Speaker 2: That's one of the kind of directions that automation can push,
• 6:52
  Speaker 2: but it can push in the other just like those accountants.
• 6:54
  Speaker 1: So you have this phrase, at least I associate the
• 6:58
  Speaker 1: phrase with you that describes this phenomenon, the hollowing out
• 7:03
  Speaker 1: of the middle. And it seems like this is roughly
• 7:07
  Speaker 1: as a sort of crude sure, college degree not college degree,
• 7:11
  Speaker 1: seems like a way to draw the line for who
• 7:14
  Speaker 1: on average benefited from the arrival of computers and who
• 7:17
  Speaker 1: on average was hurt by It is that about, right?
• 7:19
  Speaker 3: Absolutely?
• 7:19
  Speaker 2: I mean that has been over the last forty years
• 7:21
  Speaker 2: has been the huge economic dividing line.
• 7:23
  Speaker 3: Right.
• 7:23
  Speaker 2: The earnings differential between college and on college you know,
• 7:28
  Speaker 2: was always you know, substantial, and then around nineteen eighty
• 7:31
  Speaker 2: started rising and rising, and it rose for thirty years
• 7:34
  Speaker 2: and then has plateaued at that point. So now the
• 7:36
  Speaker 2: college high school differential on average and is about seventy
• 7:39
  Speaker 2: five percent.
• 7:40
  Speaker 1: So people with college degrees make seventy five percent more
• 7:43
  Speaker 1: for our people.
• 7:44
  Speaker 2: On average, Yeah, on average, so huge, huge difference.
• 7:48
  Speaker 1: So okay, now we have set the table for the
• 7:51
  Speaker 1: arrival of AI. Right, this is this is a story,
• 7:54
  Speaker 1: a true story of what's happened over the last several
• 7:57
  Speaker 1: decades because of computers. Now we have this new kind
• 7:59
  Speaker 1: of technology showing up at our doorstep. You've written a
• 8:05
  Speaker 1: story about how AI might arrive quite differently in the
• 8:08
  Speaker 1: labor force than computers did, right, with different distributional effects.
• 8:14
  Speaker 1: Tell me about that.
• 8:15
  Speaker 2: So over the last you know, really over the last
• 8:19
  Speaker 2: four decades, we've become a society dominated by services. You know,
• 8:25
  Speaker 2: we spend twenty percent of GDP on health care and education.
• 8:29
  Speaker 2: It's like, you know, so back in the you know,
• 8:32
  Speaker 2: nineteen forty, right, we spent a lot more money on
• 8:35
  Speaker 2: you know, cars and appliances and food and clothing.
• 8:38
  Speaker 1: Food. Food was a huge at the household budget in
• 8:40
  Speaker 1: nineteen forty and now it's ten percent instead of thirty
• 8:43
  Speaker 1: or forty or something.
• 8:44
  Speaker 2: Right, But why so the mixed news in that is
• 8:47
  Speaker 2: why do we spend less on those things? Partly because
• 8:49
  Speaker 2: we've gotten really productive with them. They've gotten really cheap, right,
• 8:53
  Speaker 2: all these services like healthcare, like education, like law, they've
• 8:58
  Speaker 2: not We've not gotten more productive with them, right, we
• 9:01
  Speaker 2: just have expensive people doing them. And those people have
• 9:04
  Speaker 2: gotten more expensive as the services become more necessary.
• 9:07
  Speaker 3: They've not gotten faster with them.
• 9:08
  Speaker 2: And so this created it's a distribution income that very
• 9:11
  Speaker 2: much favors the professional class and the guilds of people,
• 9:14
  Speaker 2: be they lawyers, be they doctors, they be they professors, right,
• 9:18
  Speaker 2: you know, engineers, architects, you know, contractors, And so that's
• 9:24
  Speaker 2: not a great scenario for most people. You know, if
• 9:27
  Speaker 2: you're a professional, you're on both sides that transaction. Like
• 9:29
  Speaker 2: you know, yes, you have to pay a ton to
• 9:31
  Speaker 2: a lawyer and to send your kid to college. But
• 9:33
  Speaker 2: on the other hand, you're paid a lot to you know,
• 9:35
  Speaker 2: do marketing, whatever it is that you do. But for
• 9:37
  Speaker 2: most people, most people are not in that line of work, right.
• 9:39
  Speaker 2: Only you know, forty percent of us workers have a
• 9:42
  Speaker 2: college degree. Most of them are not highly paid professionals.
• 9:45
  Speaker 2: So most people are only on the side of the
• 9:47
  Speaker 2: transaction where they pay a lot.
• 9:48
  Speaker 3: Of money for those services.
• 9:51
  Speaker 2: So the potential for AI is to enable more people
• 9:56
  Speaker 2: to do that expert work and bring down its price
• 10:01
  Speaker 2: while increasing the set of jobs available to people.
• 10:05
  Speaker 1: Right, So there's two sides of the coin you're sort
• 10:08
  Speaker 1: of packing into that last half of the sentence. Right,
• 10:10
  Speaker 1: So you're saying more people could do jobs that are
• 10:14
  Speaker 1: now done by a small guild of expects. So that's
• 10:16
  Speaker 1: good for the people doing the jobs. And also if
• 10:19
  Speaker 1: that happened, it would bring down the price of those services,
• 10:22
  Speaker 1: So that's good forever.
• 10:23
  Speaker 3: That's good for customers.
• 10:24
  Speaker 1: Like that sounds like a lot to ask for, Like
• 10:27
  Speaker 1: what what is an example?
• 10:29
  Speaker 3: Sure?
• 10:29
  Speaker 1: Like what do you think? Sure?
• 10:30
  Speaker 2: But let me also say, like, not everybody loves that story, right,
• 10:33
  Speaker 2: Who doesn't love that story?
• 10:35
  Speaker 3: The people at the top of that pyramid.
• 10:37
  Speaker 1: Right, right, you know, the people in the guilds. That's
• 10:39
  Speaker 1: why they're in a gilt.
• 10:39
  Speaker 2: Right.
• 10:40
  Speaker 1: But so let's do an example because it'll become much clearer.
• 10:43
  Speaker 1: Like you you've written about doctors and nurse practitioners right
• 10:45
  Speaker 1: in this context, So you.
• 10:46
  Speaker 2: Know nurse practitioners, you know, they're registered nurses who have
• 10:49
  Speaker 2: an additional master's degree, and that enables them to diagnose.
• 10:55
  Speaker 3: Treat and prescribe.
• 10:57
  Speaker 2: In many states they can debscribe, not all of them,
• 10:59
  Speaker 2: it depends, you know, there's different rules, and they now
• 11:02
  Speaker 2: that really encroaches on the territory that used to be
• 11:05
  Speaker 2: wholly owned by physicians, right, and you know this, they
• 11:11
  Speaker 2: have not been welcomed by the American Medical Association, not all.
• 11:15
  Speaker 2: They fought very hard to limit their scope of practice.
• 11:19
  Speaker 2: And in fact, you know, the neurors practitioners basically it's
• 11:21
  Speaker 2: been a you know, they started sixty years ago, like
• 11:25
  Speaker 2: trying to create a credential, a curriculum of scope of practice.
• 11:28
  Speaker 2: They you know, fought their way and that it's only
• 11:30
  Speaker 2: in the last twenty years they've grown, and they expected
• 11:31
  Speaker 2: to grow a lot, but always facing great pushback from
• 11:36
  Speaker 2: the medical establishment. Why because they're encroaching on their expert
• 11:39
  Speaker 2: territory that's very threatened. So I think like the you know,
• 11:43
  Speaker 2: the growth of mirrors practitioners has in many ways been
• 11:46
  Speaker 2: good for the labor market and good for patients.
• 11:48
  Speaker 3: Right, It's brought down costs.
• 11:49
  Speaker 1: And is the evidence I haven't looked at the evidence
• 11:51
  Speaker 1: on this in particular.
• 11:52
  Speaker 2: The evidence is not there's there isn't enough evidence on this, Okay,
• 11:56
  Speaker 2: So I mean I.
• 11:57
  Speaker 1: If there isn't a clear evidence one way or the other,
• 11:59
  Speaker 1: that's right.
• 12:00
  Speaker 2: Okay, there's one paper I know, but it's in a
• 12:02
  Speaker 2: very very limited scope, so I would know it's in
• 12:05
  Speaker 2: the Veterans Administration. But I think like from a from
• 12:08
  Speaker 2: a like, like I brought my kids to the pediatrician.
• 12:12
  Speaker 3: And my began bringing my kids to pediatrician.
• 12:14
  Speaker 2: There were no open nurse practitioners and I would wait
• 12:16
  Speaker 2: hours to see the pediatrician. He was terrific, but then
• 12:18
  Speaker 2: it was like a half day lost by the time
• 12:20
  Speaker 2: my kids were ten, Right, I would just go into
• 12:23
  Speaker 2: the p diatrican's office and boom, I would see a
• 12:25
  Speaker 2: nurse practitioner and it was so great. And they were attentive, right,
• 12:28
  Speaker 2: and they were knowledgeable, they're up to date, and they
• 12:30
  Speaker 2: were accessible.
• 12:31
  Speaker 3: Uh.
• 12:31
  Speaker 2: And of course and that's a that's a high pay
• 12:34
  Speaker 2: job and it's growing rapidly. So now that's not a
• 12:37
  Speaker 2: function of AI clearly, right.
• 12:39
  Speaker 1: So so okay, so now we have a we have
• 12:41
  Speaker 1: this structure, bring AI into the story. Now, how does
• 12:45
  Speaker 1: how does AI bear our story?
• 12:46
  Speaker 2: So? AI gives people tools to accomplish complicated things that
• 12:53
  Speaker 2: might be out of reach without those tools. But let
• 12:55
  Speaker 2: me be clear, it doesn't mean anyone can do anything.
• 12:57
  Speaker 2: Doesn't mean I can just go into met doctors often
• 12:59
  Speaker 2: be a nurse practitioner.
• 12:59
  Speaker 3: Right, you need to have domain knowledge.
• 13:01
  Speaker 2: Right, I have to understand, like what do I do
• 13:03
  Speaker 2: with someone starts bleeding or has a heart attack.
• 13:05
  Speaker 3: I need to.
• 13:06
  Speaker 2: Understand, you know, a ton of things. But given that,
• 13:08
  Speaker 2: with better tools, I can do more.
• 13:11
  Speaker 3: Right.
• 13:11
  Speaker 2: So AI can help with diagnosis, right, it can help
• 13:16
  Speaker 2: figure out how a medical practitioner should allocate their time
• 13:20
  Speaker 2: and attention. Appointment with a nurse practitioner, even in a clinic,
• 13:23
  Speaker 2: maybe twelve minutes, maybe twenty minutes, there's a ton that
• 13:27
  Speaker 2: could happen.
• 13:28
  Speaker 3: Right.
• 13:28
  Speaker 2: Having a tool that says, hey, here's the medical record, right,
• 13:31
  Speaker 2: this is what we saw last time. You know, here's
• 13:33
  Speaker 2: your high priority item, and here's it you know, and
• 13:36
  Speaker 2: here's the right resources.
• 13:37
  Speaker 3: And then you, between the AI.
• 13:39
  Speaker 2: And yourself, you can say, okay, now I can triage
• 13:41
  Speaker 2: this and decide should I send this person home?
• 13:43
  Speaker 3: Should I refer them upward? Can I care for them now?
• 13:46
  Speaker 2: And so you are it's a force multiplier for your
• 13:49
  Speaker 2: expertise right enables you to go further, and that would
• 13:52
  Speaker 2: be true. You know, whether you're if you're doing legal work,
• 13:56
  Speaker 2: if you're doing accounting work, if you're developing software. Software
• 13:59
  Speaker 2: is an area where AI is going to allow lots.
• 14:01
  Speaker 3: And lots of entry.
• 14:03
  Speaker 2: A lot more people can do software using AI because
• 14:06
  Speaker 2: they don't have to understand a much of the kind
• 14:10
  Speaker 2: of engineering of coding. Now as you hear software is
• 14:14
  Speaker 2: going to you knows exist at many levels. Right the
• 14:16
  Speaker 2: people who are training AI models, running data centers, doing
• 14:21
  Speaker 2: enterprise software, they will be very specialized, very highly paid.
• 14:25
  Speaker 2: But there will be a lot more software that's like
• 14:28
  Speaker 2: Uber drivers.
• 14:29
  Speaker 3: Right.
• 14:29
  Speaker 2: Oh, I want some to vibe code, you know, an
• 14:32
  Speaker 2: app for my home to do X, Y and Z.
• 14:35
  Speaker 2: I can easily find someone who can write that software
• 14:37
  Speaker 2: for me because they'll just use an AI. I probably
• 14:39
  Speaker 2: probably you won't do it yourself. Probably you'll still hire someone.
• 14:42
  Speaker 1: I mean, that's not obvious to me. Why do you
• 14:45
  Speaker 1: say probably you won't do it yourself.
• 14:47
  Speaker 2: There's you know, for example, you can easily build a
• 14:50
  Speaker 2: website right using wicx or squarespace things, But most people
• 14:54
  Speaker 2: who have a business, they hire someone to build a
• 14:56
  Speaker 2: website for them. Why, Well, One, there's a lot of
• 14:59
  Speaker 2: design that goes into it. It's not just a matter
• 15:01
  Speaker 2: of getting it running right. And two there's a lot
• 15:05
  Speaker 2: of back end behind it in terms of oh it's
• 15:07
  Speaker 2: going to take payments, Oh it's gonna you know.
• 15:10
  Speaker 1: But these things seem aliable so that.
• 15:13
  Speaker 2: You know it could change. But there are many many
• 15:15
  Speaker 2: things that you could in theory do for yourself.
• 15:17
  Speaker 1: I would not want to be in the website building
• 15:20
  Speaker 1: business right now. I don't know, right, does that not
• 15:24
  Speaker 1: like I'm not sure you can definitely do design and
• 15:26
  Speaker 1: back end work with whatever claude We'll see.
• 15:31
  Speaker 3: Yeah, we'll say.
• 15:32
  Speaker 2: The likelihood is there will be more entry. It will
• 15:34
  Speaker 2: come in at lower cost, but more people will be
• 15:36
  Speaker 2: able to do it, and you know, just like the
• 15:39
  Speaker 2: Uber example, right, So I think they will be more
• 15:42
  Speaker 2: demand for software because it's cheaper. There'll be more people
• 15:45
  Speaker 2: who can produce good software at lower cost. It will
• 15:48
  Speaker 2: be less specialized, and therefore I suspect it it'll be
• 15:52
  Speaker 2: numerically large, but it won't be as well paid as
• 15:55
  Speaker 2: the typical software job.
• 15:57
  Speaker 1: So you'll have more software engineers and the media software
• 16:00
  Speaker 1: engineer will make less than they make today.
• 16:02
  Speaker 3: Yeah, and they won't. They won't have a degree in engineering, right.
• 16:05
  Speaker 1: Right, they'll just be somebody who's good a viating.
• 16:07
  Speaker 2: I mean, like you know, website design now is primarily
• 16:09
  Speaker 2: done by people who are designers, not programmers.
• 16:13
  Speaker 1: Yes, but design is is this is something that AI
• 16:17
  Speaker 1: is getting getting better.
• 16:18
  Speaker 3: Absolutely, that's right.
• 16:20
  Speaker 1: So let what are some other domains. I mean you
• 16:22
  Speaker 1: mentioned the law.
• 16:23
  Speaker 2: Law is a good example, but again laws guilt.
• 16:25
  Speaker 1: I could see law being more hollowing out of the
• 16:27
  Speaker 1: middle though, right, like more, you know, you don't need
• 16:31
  Speaker 1: to hire the junior lawyer. You don't need to hire
• 16:33
  Speaker 1: or at least as many junior lawyers. You don't need
• 16:35
  Speaker 1: to hire as many paralegals. You can have one lawyer
• 16:38
  Speaker 1: with an army of agents.
• 16:41
  Speaker 2: Uh, it's possible. But I think in many cases you
• 16:47
  Speaker 2: want a a knowledgeable, reliable, trustworthy person working with you.
• 16:55
  Speaker 1: Yeah, you the you, the CLO, the person who knows
• 16:57
  Speaker 1: a lawyer. But you would like them to.
• 16:58
  Speaker 3: Be efficient and cheap.
• 17:00
  Speaker 2: Yeah, right, so you'd like them to be able to
• 17:01
  Speaker 2: write a contract for you in an hour rather than
• 17:03
  Speaker 2: a day.
• 17:04
  Speaker 1: Yeah. There are the sort of software first versions of
• 17:07
  Speaker 1: this already. Right, there's legal Zoom, right has existed since before.
• 17:12
  Speaker 1: It's right actually for kind of routine things like a
• 17:15
  Speaker 1: will or whatever.
• 17:16
  Speaker 2: I mean, So in many cases something is not legal
• 17:20
  Speaker 2: unless it's done by a lawyer, Like if a lawyer
• 17:22
  Speaker 2: doesn't sign it, it doesn't count.
• 17:24
  Speaker 1: This is the guild part, right.
• 17:25
  Speaker 3: That's right.
• 17:26
  Speaker 2: But even so, you could say, well, look, you know
• 17:28
  Speaker 2: more lawyers can offer some of these services at lower
• 17:31
  Speaker 2: prices if they can do them effectively.
• 17:33
  Speaker 1: Yeah, or you could have one lawyer with one hundred
• 17:38
  Speaker 1: call them paralegals plus ais, and the lawyer is signing
• 17:41
  Speaker 1: it and the paralegal is talking to the customer and marketing. Okay,
• 17:44
  Speaker 1: But that.
• 17:44
  Speaker 2: Even so, that would in many ways would be a
• 17:46
  Speaker 2: better world. Right, legal services be cheaper, and there'd be
• 17:49
  Speaker 2: this whole layer of people who are doing that intermediate role, right,
• 17:54
  Speaker 2: as opposed to going to one very expensive person. And
• 17:56
  Speaker 2: hopefully in the good scenario, there's a layer of people
• 18:00
  Speaker 2: who are not just doing grunt work. They have some
• 18:02
  Speaker 2: specialized knowledge, they're.
• 18:03
  Speaker 3: Using the tools. Right often to use these tools, well,
• 18:05
  Speaker 3: you have to know something.
• 18:07
  Speaker 1: So this is an important point, right, because it's easy
• 18:11
  Speaker 1: for me to imagine in the world where you actually
• 18:13
  Speaker 1: don't have to know much to use the tool well, right,
• 18:16
  Speaker 1: and that this sort of current moment we're in is
• 18:18
  Speaker 1: just the you know, the foothills of real AI, and
• 18:22
  Speaker 1: it'll look like some terrible scratchy TV broadcasts from ninety
• 18:27
  Speaker 1: years ago in ten years, do you know what I mean?
• 18:30
  Speaker 1: And so it seems totally likely that the AI could
• 18:34
  Speaker 1: get good enough that you wouldn't need a paralegal like
• 18:38
  Speaker 1: that the paralegal wouldn't add value because the AI would
• 18:41
  Speaker 1: just figure out what you needed better than a paralegal would.
• 18:44
  Speaker 3: Uh, I don't you know.
• 18:47
  Speaker 2: I don't want to dismiss that possibility, but I do
• 18:51
  Speaker 2: think in many cases, you're ultimately going to want another
• 18:54
  Speaker 2: person involved in what you're doing, whether for its medical care,
• 18:58
  Speaker 2: whether it's for law, whether design, and so you will
• 19:02
  Speaker 2: want a person who has better tools and some specialized knowledge,
• 19:08
  Speaker 2: but they will not be as expensive, they will be
• 19:10
  Speaker 2: more productive. And let me say that, you know, and
• 19:13
  Speaker 2: there's something else to bear in mind, right, many of
• 19:15
  Speaker 2: these services that we're talking about about law, design, contracting,
• 19:20
  Speaker 2: like you know, even like restaurants, right, they don't lend
• 19:24
  Speaker 2: themselves to a superstar type model where one person does
• 19:27
  Speaker 2: all the work, right, like in entertainment. Right, we have
• 19:29
  Speaker 2: you know, Taylor swim Right, she can entertain two billion
• 19:31
  Speaker 2: people at a time, and so it actually crowds out
• 19:34
  Speaker 2: like there, it's hard you know, in movies, in music, right,
• 19:38
  Speaker 2: the superstars crowd out everybody else. Right, And that didn't
• 19:42
  Speaker 2: used to be true in a time when we didn't
• 19:43
  Speaker 2: have broadcasts, we didn't have television.
• 19:45
  Speaker 1: Right, there was an opera house in every face in
• 19:47
  Speaker 1: your town.
• 19:48
  Speaker 2: Right. Yeah, but this can't be true in medicine. It
• 19:51
  Speaker 2: can't be true in law, It can't be true in restaurants.
• 19:53
  Speaker 2: It can't be true in contracting, you know, in many
• 19:55
  Speaker 2: of these services because they're congestible people, people have finite capacity. Right,
• 20:00
  Speaker 2: there's no doctor who can serve two billion people simultaneously.
• 20:03
  Speaker 1: Did you see the movie Her? You're describing the part
• 20:08
  Speaker 1: right before big reveal in that movie.
• 20:11
  Speaker 3: The I'm Meanna.
• 20:13
  Speaker 2: There's a lot of hands on work, and there's a
• 20:14
  Speaker 2: lot of a lot of communication. And I think, so
• 20:17
  Speaker 2: someone's going to see the best doctor, someone's gonna see
• 20:19
  Speaker 2: the median doctor, someone's gonna see the twenty fifth percentile doctor, right,
• 20:23
  Speaker 2: and if those doctors have better tools and better you know,
• 20:27
  Speaker 2: supports and by and decision supports they will be better doctors.
• 20:31
  Speaker 1: Well should That part is the easy part, But I
• 20:33
  Speaker 1: feel like that's neither here nor there with respect to
• 20:35
  Speaker 1: what we're talking. No, no, no, that's.
• 20:36
  Speaker 2: That's very critical because it means that you know, it
• 20:39
  Speaker 2: means it just because some doctor has AI, right, that
• 20:43
  Speaker 2: doctor does not crowd out all the other workers.
• 20:46
  Speaker 1: Yeah, I mean doctor is like a weird case that
• 20:48
  Speaker 1: is not like most things, right, Like doctor, like you
• 20:51
  Speaker 1: care about having a person in the room at least
• 20:53
  Speaker 1: for now when it's a doctor, more than you care
• 20:55
  Speaker 1: about it with almost anything else, right Like like with
• 21:00
  Speaker 1: most legal stuff. I mean, unless it was some huge thing,
• 21:03
  Speaker 1: as long as you, like everybody use the AI and
• 21:06
  Speaker 1: you know it was fine, you would just use the AI,
• 21:07
  Speaker 1: do you know what I mean? Like for sure at
• 21:10
  Speaker 1: least in a lot of cases. Yes, some people want
• 21:13
  Speaker 1: a person, but.
• 21:13
  Speaker 3: Do you know what I mean, I'm following what you're saying.
• 21:15
  Speaker 1: Somebody who's a teenager now, like they don't want to
• 21:18
  Speaker 1: talk to a person, Like they would rather not talk
• 21:21
  Speaker 1: to a person if they could. I know teenagers, Yeah,
• 21:24
  Speaker 1: I mean live in my house.
• 21:28
  Speaker 2: Yeah, I think I think medicine is maybe is the
• 21:31
  Speaker 2: strongest example.
• 21:33
  Speaker 1: So, yeah, medicine is the strongest example. Yes, you want
• 21:35
  Speaker 1: you want a person, you want a doctor or a nurse,
• 21:37
  Speaker 1: and it's you're scared.
• 21:39
  Speaker 2: And well, I mean, let's first let's remember that, you know,
• 21:41
  Speaker 2: medicine is a huge part of employment. It's also about
• 21:44
  Speaker 2: twenty percent of US employment healthcare, and it's growing.
• 21:47
  Speaker 3: Right.
• 21:48
  Speaker 2: Health is sort of the ultimate luxury good. Right, It's
• 21:51
  Speaker 2: the one thing that as you get more and more affluent,
• 21:54
  Speaker 2: you want to spend more and more on, right, because
• 21:57
  Speaker 2: you know, you don't want more and more food you
• 21:59
  Speaker 2: can have. You you know, some people can have one
• 22:01
  Speaker 2: hundred cars, but most people are good with two. But
• 22:03
  Speaker 2: your longevity and your well being, right, those are kind
• 22:07
  Speaker 2: of fundamental to everything else that you do. And so
• 22:10
  Speaker 2: and you look at all these billionaire bros right whore
• 22:12
  Speaker 2: you know, they have their blood boys and there you know,
• 22:15
  Speaker 2: and they're spending infinite you know why are they doing that?
• 22:17
  Speaker 1: Right?
• 22:17
  Speaker 2: Why is why is healthcare becoming a larger and larger
• 22:20
  Speaker 2: part of expenditure over time?
• 22:22
  Speaker 3: Right?
• 22:22
  Speaker 2: Because we're getting more affluent h and so it matters
• 22:25
  Speaker 2: more h And they're getting more effective.
• 22:29
  Speaker 3: Right.
• 22:29
  Speaker 2: You know, there was a one hundred years ago, it
• 22:31
  Speaker 2: wasn't clear it was good for your health. Nego see
• 22:32
  Speaker 2: a doctor, right, No.
• 22:33
  Speaker 1: It quite possibly bad for you, exactly.
• 22:36
  Speaker 2: Now there's just an amazing number of things and that
• 22:38
  Speaker 2: number is expanded, so we will spend more on health services,
• 22:42
  Speaker 2: it will be a larger part of employment. And of
• 22:44
  Speaker 2: course we're aging, right, I mean, you know our population
• 22:46
  Speaker 2: is getting older, so it's not at all crazy to
• 22:49
  Speaker 2: think this will be remain And I don't care how
• 22:52
  Speaker 2: much AI you throw in there. It's not going to
• 22:55
  Speaker 2: shrink in terms of employment. It's going to be ill,
• 22:58
  Speaker 2: provide more services, ideally at lower cost, and make it
• 23:01
  Speaker 2: more accessible. It's not going to become small. So although
• 23:05
  Speaker 2: it is a special case, it's not like a ledge cases.
• 23:10
  Speaker 2: So it's a big special, big special special exactly.
• 23:17
  Speaker 1: Will be back in just a minute. I want to
• 23:29
  Speaker 1: go back to the macro idea. Right, So you have
• 23:32
  Speaker 1: this argument, this story about how computers before AI essentially
• 23:37
  Speaker 1: contributed to rising at a quality. Right, Like, if you
• 23:39
  Speaker 1: were in the top whatever, twenty percent of the distribution,
• 23:42
  Speaker 1: computers are good for you. If you were in the middle,
• 23:44
  Speaker 1: they were bad for you.
• 23:45
  Speaker 3: That's correct.
• 23:45
  Speaker 1: In short, how what's the macro story for AI.
• 23:52
  Speaker 2: So I don't want to oversimplify. I think it's quite nuanced.
• 23:56
  Speaker 2: It's possible that it will enable war competition in these
• 24:02
  Speaker 2: high end services that we can spend so much money on,
• 24:05
  Speaker 2: and that will could contribute to lowering inequality somewhat and
• 24:09
  Speaker 2: sort of growing back a different middle, right, middle set
• 24:13
  Speaker 2: of software engineers, a middle set of kitchen designers and contractors,
• 24:17
  Speaker 2: you know, middle set of very skilled repair people who
• 24:19
  Speaker 2: you know, there's a ton of knowledge and expertise involved.
• 24:22
  Speaker 2: It's not all you know comes from college degrees, right,
• 24:25
  Speaker 2: people who are in the trades, who do electrical work,
• 24:27
  Speaker 2: who do plumbing, who do you know construction?
• 24:30
  Speaker 1: I mean people the trades have been doing pretty well, right, exactly.
• 24:32
  Speaker 1: I avoided this hollowing out of the middle because a
• 24:34
  Speaker 1: computer can't whatever, fix your fix the wiring in your house.
• 24:38
  Speaker 2: Yeah, absolutely not. That's right, And I think AI can
• 24:41
  Speaker 2: actually be helpful, and we already see examples of this, right,
• 24:43
  Speaker 2: you know, you can solve a harder problem. So that
• 24:46
  Speaker 2: would create a better a better middle, a different middle.
• 24:49
  Speaker 2: It wouldn't be the same middle, and so it's somewhat
• 24:52
  Speaker 2: expertise leveling.
• 24:53
  Speaker 3: Right.
• 24:53
  Speaker 2: It creates competition where we desperately need it and creates
• 24:57
  Speaker 2: opportunity for people who have appropriate skills foundational expertise. Right,
• 25:02
  Speaker 2: it's got to be got to know something about medicine.
• 25:04
  Speaker 2: You gotta know something about the trades, You got to
• 25:05
  Speaker 2: know something about the law, and you need to know
• 25:06
  Speaker 2: something about software.
• 25:07
  Speaker 3: But maybe not as much. Right. You could imagine we
• 25:11
  Speaker 3: have a.
• 25:11
  Speaker 2: Four year you know, healthcare degree, right where you learn
• 25:14
  Speaker 2: a bunch of stuff and then you're ready to do
• 25:16
  Speaker 2: a lot of services, a lot of health services.
• 25:18
  Speaker 3: So that would be that would be.
• 25:20
  Speaker 2: A world where there's more opportunity for middle skilled workers
• 25:23
  Speaker 2: and lower prices, and so that would have less inequality.
• 25:27
  Speaker 2: But I want to say that even if that's true
• 25:29
  Speaker 2: my scenario and there's less, there's gonna be a ton
• 25:32
  Speaker 2: of disruption, right, because we're talking about expertise and sometimes
• 25:35
  Speaker 2: being vacated.
• 25:36
  Speaker 3: Right.
• 25:36
  Speaker 2: If you're a medical transcriptionist, that job is almost gone.
• 25:39
  Speaker 3: Right.
• 25:39
  Speaker 2: If you're a language translator, some will remain and they'll
• 25:42
  Speaker 2: be highly highly skilled, right, but there'll be few of them.
• 25:45
  Speaker 2: I'm very worried about people who work in call centers, right,
• 25:48
  Speaker 2: you know, there's several million of them. It's very disruptive
• 25:54
  Speaker 2: psychically damaging, economically scarring for people to have their expertise
• 26:00
  Speaker 2: suddenly devalued. The thing that you know how to do.
• 26:02
  Speaker 2: For most most people are doing the best paid thing
• 26:04
  Speaker 2: they can do. Not everybody, right, some people are podcasting,
• 26:08
  Speaker 2: but out our professors, right, many professors, yes, could you know.
• 26:13
  Speaker 3: There are higher paying things they could do. They love
• 26:15
  Speaker 3: what they do.
• 26:15
  Speaker 2: So most people are doing you know, they don't have
• 26:18
  Speaker 2: that luxury, right, They're doing the job that pays them
• 26:21
  Speaker 2: the best income they can get. And if that work
• 26:23
  Speaker 2: becomes less valuable, there's not an equally good job, right,
• 26:26
  Speaker 2: They're gonna have to conness like WHI is loss of
• 26:28
  Speaker 2: manufacturing work so devastating makings. It's often specialized work that
• 26:32
  Speaker 2: you know is relatively highly paid, and the alternative work
• 26:36
  Speaker 2: for someone with that education, you know, it's not going
• 26:38
  Speaker 2: to be as good. So there's gonna be a lot
• 26:41
  Speaker 2: of change, and I expect AI to devalue some forms
• 26:44
  Speaker 2: of expertise really rapidly. Others will become more valuable, and
• 26:48
  Speaker 2: it's not the same people who are on the upside
• 26:50
  Speaker 2: and downsid You could say the average is great, but
• 26:52
  Speaker 2: like no one experiences the average, And so even if
• 26:56
  Speaker 2: you say, oh, yeah, that could be a better world.
• 26:57
  Speaker 2: It's like, yeah, but we don't get to that world overnight,
• 27:00
  Speaker 2: and there's a lot of change that happens along the way,
• 27:03
  Speaker 2: and we should be investing and preparing for that.
• 27:05
  Speaker 1: So let's talk about that, I mean, and that we
• 27:08
  Speaker 1: can be like there's the policy WII and then there's
• 27:11
  Speaker 1: like I'm some guy, what should I do? So, like
• 27:14
  Speaker 1: what do we do on both of those levels?
• 27:16
  Speaker 2: Let's start with the policy week? Yeah, okay, So there
• 27:19
  Speaker 2: are three buckets of things that I think are very important.
• 27:22
  Speaker 2: One of them is to help people transition between jobs.
• 27:25
  Speaker 2: So the first of those is wage insurance. Wage insurance
• 27:28
  Speaker 2: is like you're working a manufacturing job. You're making fifty
• 27:31
  Speaker 2: thousand dollars a year, twenty five dollars an hour. You
• 27:34
  Speaker 2: lose that job and you know manufacturing is going down.
• 27:37
  Speaker 2: There's non an equivalent job, and the next job you
• 27:40
  Speaker 2: can get is fifteen bucks an hour, thirty thousand dollars
• 27:42
  Speaker 2: a year, and you say, I'm not gonna take that job.
• 27:45
  Speaker 3: That's like beneath my digity, and so you.
• 27:47
  Speaker 2: Don't, and then you spend a long time looking and
• 27:49
  Speaker 2: that you know it has its own costs, and people
• 27:51
  Speaker 2: the longer they're at a laborports, they're harder, they find Again,
• 27:53
  Speaker 2: so wage Insurance says, hey, you're like, we understand that
• 27:56
  Speaker 2: we're gonna make up half the difference between your twenty
• 27:59
  Speaker 2: five dollars hour job and your fifteen dollar an hour
• 28:01
  Speaker 2: job up to a finite amount of time, up to
• 28:03
  Speaker 2: two years, up to eight thousand dollars total. And we're
• 28:05
  Speaker 2: not telling you stay in a low wage job forever.
• 28:08
  Speaker 2: By all means get a better job, and so. And
• 28:10
  Speaker 2: we have evidence on it, not enough as much as
• 28:12
  Speaker 2: we want. But during the twenty tens, right as part
• 28:15
  Speaker 2: of the Trade Adjestment Act, there was an experiment done
• 28:17
  Speaker 2: with wage insurance. The result of this policy experiment was
• 28:22
  Speaker 2: it got people back to work faster, and it was
• 28:24
  Speaker 2: so effective that it paid for itself, in other words,
• 28:26
  Speaker 2: by the reductions in unemployment insurance payments and the increase
• 28:32
  Speaker 2: in tax revenue from people working. It was self funding. Now,
• 28:36
  Speaker 2: that's on a limited base of evidences. Were just placed
• 28:38
  Speaker 2: manufacturing workers, they were older. But there are now major
• 28:44
  Speaker 2: pilot studies underway working with states. I'm you know, more
• 28:47
  Speaker 2: details will be revealed soon. They're quite expensive and they're
• 28:50
  Speaker 2: you know, they're funded to try this at a larger scale,
• 28:55
  Speaker 2: so wage insurance.
• 28:56
  Speaker 1: This is one idea.
• 28:58
  Speaker 2: And I'll call it's a no regrets idea in the
• 29:00
  Speaker 2: sense that we won't be sorry we did it even
• 29:03
  Speaker 2: if there isn't a jobs apocalypse.
• 29:04
  Speaker 3: Right, It's a good idea. It was a good idea
• 29:07
  Speaker 3: twenty years ago. It's a good idea now. It'll still
• 29:10
  Speaker 3: be a good idea.
• 29:10
  Speaker 2: And the idea is people are making transitions, and those
• 29:14
  Speaker 2: transitions are frictional, they're costly, it's psychically damaging, and when
• 29:17
  Speaker 2: people are out of work for a long time that
• 29:19
  Speaker 2: has its own costs. You would like to facilitate them
• 29:21
  Speaker 2: going back to work, even while they continue to search.
• 29:25
  Speaker 1: And like, I feel like subsidizing work, essentially paying people
• 29:31
  Speaker 1: to work it politically it feels good. It does, and
• 29:33
  Speaker 1: frankly for the people like that. I mean, work is
• 29:36
  Speaker 1: valuable in our society, and like, you can feel different
• 29:39
  Speaker 1: ways about that, but I basically feel good about that,
• 29:41
  Speaker 1: and like subsidizing people that work seems good to me, frankly.
• 29:45
  Speaker 2: And I think that fits with the American psyche right.
• 29:47
  Speaker 2: Our political system is very hostile towards people who are
• 29:51
  Speaker 2: not working. Right, If you're not working and you're not
• 29:54
  Speaker 2: a kid, and you're not elderly, what.
• 29:56
  Speaker 3: Is wrong with you? Right, is what is the kind
• 29:58
  Speaker 3: of message?
• 29:59
  Speaker 2: That's why you're constantly we're always putting these work requirements.
• 30:02
  Speaker 1: And I mean you're not making a normative statement. Just
• 30:04
  Speaker 1: to be clear, you're describing the sort of political vibes.
• 30:06
  Speaker 3: Yeah, no, no, clearly, No, no, I do not. I'm not
• 30:08
  Speaker 3: I'm not endorsed that view at all.
• 30:09
  Speaker 2: I'm just saying that that appears to be the perspective
• 30:11
  Speaker 2: of the politicians, right, That's why they said, oh, you can't
• 30:13
  Speaker 2: get you know, snap, unless you're working, you can't get medical.
• 30:16
  Speaker 2: You know, they're constantly trying to condition things on work.
• 30:19
  Speaker 2: So this says it says, okay, like this is about supporting.
• 30:23
  Speaker 3: People to work. Okay, that's one thing one okay.
• 30:26
  Speaker 1: Two.
• 30:26
  Speaker 2: Two is we need to do retraining at scale much
• 30:31
  Speaker 2: more effectively.
• 30:32
  Speaker 1: Yeah, retraining seems like it never worked.
• 30:35
  Speaker 2: Okay, See that's what everybody says. But in fact that's
• 30:38
  Speaker 2: kind of based on old wisdom.
• 30:39
  Speaker 3: We now know much.
• 30:41
  Speaker 1: More about ai really good at retraining.
• 30:43
  Speaker 2: Oh no, we don't know that, but we do know
• 30:45
  Speaker 2: of really effective training programs. Now, they're small.
• 30:49
  Speaker 3: So like Project.
• 30:50
  Speaker 2: Quest you're a Jewish vocational services goodwill there what are
• 30:55
  Speaker 2: called sexual training programs. They train people for specific occupations
• 30:58
  Speaker 2: and activities. Right, so construction, it could be a medical technical,
• 31:02
  Speaker 2: it could be diesel repair, et cetera. But we don't
• 31:05
  Speaker 2: know how to scale those at a very large scale.
• 31:07
  Speaker 2: So what folks are working on now is trying to
• 31:09
  Speaker 2: evaluate those things in real time using big data without experiments,
• 31:13
  Speaker 2: and so you can actually measure the incremental earnings that
• 31:16
  Speaker 2: are coming out of these programs and then recycle some
• 31:18
  Speaker 2: of that incremental revenue to feedback into these programs.
• 31:22
  Speaker 3: And we because we.
• 31:22
  Speaker 2: Need to have a phollanx of options available to you.
• 31:26
  Speaker 1: Yeah, right, I mean that one. Maybe I just am
• 31:32
  Speaker 1: out of date on it, but you know, like your
• 31:33
  Speaker 1: own work on the effect of competition from China on
• 31:37
  Speaker 1: you know, workers in the US in the first part
• 31:39
  Speaker 1: of the century, Like, it's interesting to me that you
• 31:42
  Speaker 1: found more recently that like the towns that were really
• 31:44
  Speaker 1: impacted by that, the towns have come back, right to
• 31:47
  Speaker 1: some extent, right, but the specific workers who were displaced
• 31:50
  Speaker 1: have not, but that we did.
• 31:52
  Speaker 2: Nothing to retrain them, right, I mean there was no
• 31:54
  Speaker 2: policy in place, Right.
• 31:56
  Speaker 1: So you think it's a policy. You think you think
• 31:58
  Speaker 1: a policy could have helped the person who was somewhat
• 32:00
  Speaker 1: fifty to my approximate age, someone got laid off.
• 32:03
  Speaker 2: Yeah, okay, yeah, I mean it's not a panacea, but
• 32:06
  Speaker 2: it's something. And you know, other countries do this much
• 32:08
  Speaker 2: more effectively than we do, right, you know Denmark.
• 32:11
  Speaker 1: Right, I was afraid you were going to say Denmark,
• 32:13
  Speaker 1: it's always dead Mark.
• 32:14
  Speaker 3: I know, I know everyone.
• 32:15
  Speaker 2: It's like, you know, don't tell me about Scandinavia, like
• 32:17
  Speaker 2: it's not a real places.
• 32:19
  Speaker 1: Doesn't map to the US.
• 32:21
  Speaker 3: Right, Well, I agree, so city, I think that's absolutely right.
• 32:23
  Speaker 2: We can't do the Danish system, but it's an existence
• 32:28
  Speaker 2: proof of you can do it. Well, right, they're affected,
• 32:30
  Speaker 2: they can they can do it, Yell, it can be right.
• 32:32
  Speaker 2: So that's why my main, second major policy is to
• 32:35
  Speaker 2: send all unemployed workers to Denmark where they can be reattrained.
• 32:38
  Speaker 1: I just think that's going to have an incredible safety
• 32:40
  Speaker 1: net and swim in the harbor and co. Exactly, it's
• 32:42
  Speaker 1: gonna what's number three?
• 32:44
  Speaker 3: Okay?
• 32:45
  Speaker 2: Number three is what I call universal basic capital. Other
• 32:48
  Speaker 2: people call it that as well. So let me just
• 32:50
  Speaker 2: same shate universal basic income. It's a notion that you know,
• 32:53
  Speaker 2: whether you're working or not, we just write you check
• 32:54
  Speaker 2: every month from the largest.
• 32:56
  Speaker 3: Of the tech bros. Or whatever.
• 32:58
  Speaker 2: Yeah, I don't think the political economy of that is
• 33:00
  Speaker 2: viable for the same reason we just had the conversation
• 33:02
  Speaker 2: about we don't like to as a country. We're just
• 33:05
  Speaker 2: not supportive of people who are not working, or we
• 33:08
  Speaker 2: don't believe even unconditional transfers. So universal basic capital is
• 33:12
  Speaker 2: a little different idea. It says, look, at the time
• 33:14
  Speaker 2: you were born, we're going to give you an endowment
• 33:16
  Speaker 2: of capital.
• 33:16
  Speaker 3: It doesn't have to be a lot, a couple of
• 33:18
  Speaker 3: that you can call.
• 33:18
  Speaker 1: Them because capital means stock stock and we just say stock,
• 33:21
  Speaker 1: you know, we.
• 33:21
  Speaker 2: Call them Trump accounts, right, which is what exists already, right, Yeah,
• 33:25
  Speaker 2: And it's a it's an asset you can you know,
• 33:29
  Speaker 2: it's you can't spend it until you're of age.
• 33:31
  Speaker 1: You so like, just just to be a little more specific,
• 33:34
  Speaker 1: the notion is like the government will buy whatever a fund,
• 33:38
  Speaker 1: an ETF, a mutual fund that tracks the S and
• 33:39
  Speaker 1: P five hundred. The ob a great way to give
• 33:42
  Speaker 1: one thousand dollars to every baby basically, and you now
• 33:44
  Speaker 1: own a thousand dollars of S and P five hundred
• 33:46
  Speaker 1: stock and you can't touch it till you're eighteen. Is
• 33:48
  Speaker 1: that what you're thinking? Exactly?
• 33:50
  Speaker 2: And you can invest in it further, right, you just
• 33:52
  Speaker 2: like you can you have tax advantage for one K savings, right,
• 33:55
  Speaker 2: you have.
• 33:55
  Speaker 1: To like a five to nine like a college savings
• 33:57
  Speaker 1: stock or something exactly.
• 33:58
  Speaker 3: So, like, what what is the purpose of this? It does?
• 34:01
  Speaker 3: It does three things.
• 34:02
  Speaker 2: One is it it diversifies people, right, So most people,
• 34:06
  Speaker 2: you know, most people's lifetime income is bound up in
• 34:08
  Speaker 2: the value of their skill and expertise. And that's it's
• 34:11
  Speaker 2: it's always vulnerable, it's always risky.
• 34:12
  Speaker 1: It's labor. Most most people make most of their money
• 34:15
  Speaker 1: from working.
• 34:15
  Speaker 2: Exactly from the right, you're saying, it's always that's it's
• 34:18
  Speaker 2: always risky to have all your eggs in one basket, right,
• 34:21
  Speaker 2: especially if you know you do something specialized and all
• 34:22
  Speaker 2: of a sudden that thing becomes obsolete.
• 34:24
  Speaker 3: That's really risky.
• 34:25
  Speaker 2: So one, is it gives you It's not it's by
• 34:27
  Speaker 2: no means a complete diveristication start, but it's a start, right,
• 34:30
  Speaker 2: And especially that's one thing. Two if we think we're
• 34:33
  Speaker 2: entering a world where capital is going to rise in
• 34:35
  Speaker 2: value relative to labor, right, that will be less than
• 34:38
  Speaker 2: you especially want to have people have money in capital, right.
• 34:41
  Speaker 1: Yeah, meaning if if more of the national income is
• 34:44
  Speaker 1: going to go to companies are the owners of shares
• 34:47
  Speaker 1: of stocks, and less is going to go to labor, right,
• 34:50
  Speaker 1: the two places that can go, then you want more
• 34:52
  Speaker 1: people to own stock, if you want everybody to prosperate.
• 34:54
  Speaker 3: It, exactly, that's right.
• 34:56
  Speaker 2: And you know labor is you know, everyone is born
• 34:59
  Speaker 2: in you know the good scenario with one worker themselves, right,
• 35:02
  Speaker 2: who work for them.
• 35:04
  Speaker 3: But capital is not like that, right.
• 35:06
  Speaker 1: Right, some people are born with a lot and most
• 35:08
  Speaker 1: people are.
• 35:08
  Speaker 2: Born people born and zero that's right. And the third
• 35:11
  Speaker 2: thing it does is universal based capital. Again, is it
• 35:15
  Speaker 2: gives them an ownership stake, almost like a voting stake
• 35:18
  Speaker 2: in the economy.
• 35:19
  Speaker 1: Yes, well, I mean most most companies, you have a
• 35:22
  Speaker 1: voting voting rights if you're on stock, although not all.
• 35:24
  Speaker 2: Exactly, And so if you know, if you're I'm very
• 35:27
  Speaker 2: worried about a society where more people are seen as
• 35:31
  Speaker 2: claimants without being seen as contributors.
• 35:33
  Speaker 1: So claimant rather than contributor. That's the universal basic income.
• 35:36
  Speaker 1: That's the world where's the machines take our jobs and
• 35:39
  Speaker 1: people just get a check every month or whatever.
• 35:42
  Speaker 2: I don't morally oppose that, let me be clearer. I
• 35:44
  Speaker 2: just don't think it's our I don't think our body
• 35:47
  Speaker 2: politic tolerates it.
• 35:49
  Speaker 1: Yes well, and are like our structures of meaning and
• 35:53
  Speaker 1: of value are not well set up.
• 35:55
  Speaker 3: That I also agree with. That's even secondary point.
• 35:58
  Speaker 2: So that's so University of the Capital says, you know again,
• 36:01
  Speaker 2: it's a very small step in the way.
• 36:02
  Speaker 1: It says.
• 36:02
  Speaker 2: Look everyone, now, hey, I'm not just a claimant. I'm
• 36:05
  Speaker 2: a contributor. I own part of the capital.
• 36:07
  Speaker 1: And in this universe, is the government just buying shares
• 36:10
  Speaker 1: in the in stock and everything and giving it to people.
• 36:13
  Speaker 3: Sure, but it's not that hard to do.
• 36:15
  Speaker 2: Like, there's only about four million babies born a year
• 36:17
  Speaker 2: in the United States, right, so if you had to
• 36:19
  Speaker 2: write them each a check for a thousand dollars, that's
• 36:21
  Speaker 2: four billion bucks, right, That's less than we spend on
• 36:24
  Speaker 2: You know, I don't know anything you can anything, any
• 36:28
  Speaker 2: bomb you can drop, so it's not.
• 36:32
  Speaker 1: But I means, so one thousand dollars, I suppose compounding
• 36:35
  Speaker 1: it becomes non.
• 36:35
  Speaker 3: True, right, and you can do more than that.
• 36:37
  Speaker 1: Right.
• 36:37
  Speaker 2: You could even say to billionaire and say, hey, you
• 36:39
  Speaker 2: know what, here's a tax advantage way that you could
• 36:41
  Speaker 2: donate to the public. Take your shares and just throw
• 36:44
  Speaker 2: them into the baby bond fund and they will be destribed.
• 36:46
  Speaker 1: That's nice, right, Yeah, that's interesting. I mean it's sort
• 36:48
  Speaker 1: of a distributed sovereign wealth fund. Right, it's sort of
• 36:51
  Speaker 1: rhymes with a sovereign wealth fund.
• 36:53
  Speaker 2: That's that's exactly exactly what it is. Exactly what it is,
• 36:56
  Speaker 2: but a sovereign wealth fund. Right, you don't get to
• 36:58
  Speaker 2: choose your own investments. You can't say I need to
• 37:00
  Speaker 2: take it out today.
• 37:01
  Speaker 1: No, it's not yours in the way, that's what you're proposing.
• 37:05
  Speaker 2: So you know, and again, none of these policies, the
• 37:08
  Speaker 2: three I'm talking about, you know, uh, wage insurance, you know, modernized.
• 37:11
  Speaker 1: Training, they're small ish, right, They're they're small issues, that's right.
• 37:15
  Speaker 2: None of them is that, you know, and I think
• 37:17
  Speaker 2: we need big you know, we face big challenges. These
• 37:19
  Speaker 2: are around the edges. But they're all what I call
• 37:22
  Speaker 2: no regrets policies. Right, we won't be sorry we did them.
• 37:25
  Speaker 2: Even if the worst doesn't.
• 37:26
  Speaker 3: Come to pass.
• 37:27
  Speaker 2: These are things that are just you know, they are
• 37:30
  Speaker 2: wise in terms of helping people make transitions and giving
• 37:33
  Speaker 2: people diversification. You know, in the good scenario, right, if AI,
• 37:37
  Speaker 2: if we use it, well, we're going to become more affluent. Right,
• 37:41
  Speaker 2: but that doesn't mean we're going to have that will
• 37:43
  Speaker 2: be well shared.
• 37:44
  Speaker 3: Right.
• 37:44
  Speaker 2: We need a society which says, look, if there's gonna
• 37:48
  Speaker 2: be an AI dividend, meaning we become more productive, we
• 37:51
  Speaker 2: want to make sure that people are brought along, right,
• 37:53
  Speaker 2: that if you if you're if your expertise becomes devalued,
• 37:56
  Speaker 2: that there's new ways to train that get you back
• 37:58
  Speaker 2: in the labor market, that your portfolio, your income sources
• 38:01
  Speaker 2: are more diversified. You know, people you know, the science
• 38:04
  Speaker 2: fiction writer Ted Shang, you know, said look, when you
• 38:08
  Speaker 2: know people are saying, oh, you know, people are scared
• 38:10
  Speaker 2: of automation. They're scared of machines. They're not scared of automations,
• 38:13
  Speaker 2: they're scared of capitalism. Right, They're scared of their work
• 38:17
  Speaker 2: will be automated and there'll be nothing left for them
• 38:19
  Speaker 2: to get they won't get paid.
• 38:21
  Speaker 3: Right.
• 38:21
  Speaker 2: It's not that people are opposed to higher productivity or
• 38:26
  Speaker 2: you know, things being more convenient.
• 38:27
  Speaker 3: They're opposed to losing their livelihoods. Right, And AI could
• 38:32
  Speaker 3: right produce.
• 38:32
  Speaker 2: A lot of higher productivity and a lot of you know,
• 38:34
  Speaker 2: convenience automation, but put a lot of people out of work,
• 38:36
  Speaker 2: at least temporarily during the transition.
• 38:38
  Speaker 3: Right. So, no wonder people are afraid of it. Right,
• 38:41
  Speaker 3: No wonder people are opposed. No wonder people are angry
• 38:43
  Speaker 3: about data centers. They don't trust our.
• 38:46
  Speaker 2: Social and political systems to ensure them and make sure
• 38:49
  Speaker 2: they're brought along. So it's very understandable.
• 38:52
  Speaker 1: Right.
• 38:52
  Speaker 2: So I would like to take the edge off of
• 38:54
  Speaker 2: some of that, Right, I would like to have policies
• 38:57
  Speaker 2: that help support the challenges you help people through those
• 39:01
  Speaker 2: challenging things that are some of them are likely to
• 39:04
  Speaker 2: ahead even in the good scenarios.
• 39:09
  Speaker 1: We'll be back in a minute with the lightning ground.
• 39:23
  Speaker 1: Let's finish with the lighting rounds. Okay, what's the most
• 39:27
  Speaker 1: peril you've ever been in on a sailboat?
• 39:33
  Speaker 3: Actually?
• 39:33
  Speaker 2: I went, I went, I took my I went sailing
• 39:37
  Speaker 2: during a hurricane.
• 39:39
  Speaker 1: Bad idea.
• 39:40
  Speaker 2: Yeah, and I got knocked over six times on purpose. Well,
• 39:44
  Speaker 2: I know, I tried to stay upright.
• 39:46
  Speaker 1: No, no, did you go sailing at a hurricane?
• 39:48
  Speaker 2: I went sailing and Hurricane Lee a couple of years
• 39:51
  Speaker 2: ago in New Hampshire.
• 39:53
  Speaker 1: On a couple of years ago. I was gonna ask
• 39:55
  Speaker 1: if you were fourteen.
• 39:55
  Speaker 2: No, no, no, I was fifty nine. But I have
• 40:00
  Speaker 2: an ultra light RSA. It's like a laser, and it's
• 40:03
  Speaker 2: super fast and it's very competent even in high winds,
• 40:07
  Speaker 2: and I really want to try it out.
• 40:08
  Speaker 3: It was small lake.
• 40:10
  Speaker 2: People were like my neighbors were all there with the cameras.
• 40:13
  Speaker 2: I assume that if I started to drown, they would have, uh,
• 40:17
  Speaker 2: I don't know, sent the footage to see exactly exactly. So, yeah,
• 40:20
  Speaker 2: that was perilous.
• 40:21
  Speaker 3: That was That was exciting.
• 40:23
  Speaker 1: And did you capsize six times?
• 40:25
  Speaker 3: Six times?
• 40:25
  Speaker 1: Six times?
• 40:26
  Speaker 3: Was exhausting. It was completely exhausting.
• 40:28
  Speaker 1: It's like a little boat and you can kind of
• 40:30
  Speaker 1: lean and ride it is now you have to climb.
• 40:31
  Speaker 3: On the daggerboard and pull it up and then throw.
• 40:34
  Speaker 2: You have to do what's called the walrusk, where you'd
• 40:36
  Speaker 2: like sort of throw yourself out of the water into
• 40:38
  Speaker 2: the center of the boat.
• 40:38
  Speaker 3: Otherwise it capsize back on you. So that was fun.
• 40:43
  Speaker 1: What's it like teaching college students in the age of
• 40:45
  Speaker 1: AI at the dawn of AI?
• 40:49
  Speaker 2: So the students I teach a MIT students and they're
• 40:52
  Speaker 2: it's a great So.
• 40:53
  Speaker 1: They're really good at prompt They're really good, they're very.
• 40:59
  Speaker 2: There, they're super motivated. It's a it's a real luxury.
• 41:02
  Speaker 2: So I don't worry that they great.
• 41:03
  Speaker 1: It's perhaps never been a better moment to be an
• 41:06
  Speaker 1: MIT student than right and you.
• 41:08
  Speaker 2: Were being an MIT per I don't worry that they're
• 41:10
  Speaker 2: going to take all the shortcuts and not learn anything.
• 41:12
  Speaker 2: They're super hard working, so I feel like they're just
• 41:15
  Speaker 2: power to.
• 41:15
  Speaker 1: They're building the tools for everyone else to take the
• 41:17
  Speaker 1: shortcuts and not learn anything, and.
• 41:19
  Speaker 2: They know how to use the tools well, right, So
• 41:21
  Speaker 2: I feel like I can push them harder and say,
• 41:24
  Speaker 2: go figure this out, right.
• 41:26
  Speaker 1: Well, are you building AI into your classes or just
• 41:29
  Speaker 1: assume they're going to use AI to figure things out?
• 41:30
  Speaker 1: And that's good.
• 41:32
  Speaker 2: I my curriculum assumes they're using A like for example,
• 41:36
  Speaker 2: I've downweighted out of class exams. Right, I still get
• 41:38
  Speaker 2: problem sets and they're hard, but they don't count as much.
• 41:41
  Speaker 2: You have to do it on exams or on your papers. Right,
• 41:44
  Speaker 2: So I expect they're using it, but they know that
• 41:47
  Speaker 2: eventually they will have to demonstrate the knowledge. Right.
• 41:49
  Speaker 3: But I can throw the harder stuff with them, say
• 41:51
  Speaker 3: you'll solve this.
• 41:52
  Speaker 2: I can also say, hey, this can't be as badly
• 41:55
  Speaker 2: written as it is right now because you have better tools,
• 41:57
  Speaker 2: so go do it again. And of course when I
• 42:00
  Speaker 2: building my lectures, I can incorporate new material more quickly
• 42:03
  Speaker 2: because I can build slides and you know, master stuff
• 42:06
  Speaker 2: faster myself.
• 42:07
  Speaker 1: Most peril you've ever been in on a motorcycle.
• 42:10
  Speaker 3: But you're always in parallel on a motorcycle.
• 42:13
  Speaker 1: Yes, that is the correct answer. Anything else we should
• 42:19
  Speaker 1: talk about.
• 42:20
  Speaker 2: The uh, you know, I will say, I mean, just
• 42:23
  Speaker 2: like you know, I think a lot of young people
• 42:24
  Speaker 2: are extremely anxious about AI UH and understand I understand why.
• 42:28
  Speaker 2: But I also think they are the group that's gonna
• 42:31
  Speaker 2: use it well better than the rest of us and
• 42:33
  Speaker 2: figure out what to do with it. And so there's
• 42:36
  Speaker 2: never been a technology that's made old people great at
• 42:38
  Speaker 2: the at the cost of young people that I'm aware of. Right,
• 42:41
  Speaker 2: Almost all new technologies are developed and pioneered and ultimately
• 42:45
  Speaker 2: kind of figured out by younger generation that figures out
• 42:49
  Speaker 2: what to do with it. And so I think, you know,
• 42:51
  Speaker 2: businesses are now looking very hard to find, like you know,
• 42:54
  Speaker 2: kind of AI native kids, right, And so I think
• 42:58
  Speaker 2: there's a lot of risk. I understand the uncertainty and
• 43:00
  Speaker 2: the fear. I don't want to disminish that, but I
• 43:02
  Speaker 2: also think there's opportunity.
• 43:10
  Speaker 1: David Otter is a labor economist at MIT. Please let
• 43:15
  Speaker 1: us know what you think of the show, which you
• 43:17
  Speaker 1: want to hear more of, which you want to hear
• 43:19
  Speaker 1: less of particular guest ideas. You can email us at
• 43:24
  Speaker 1: problem at Pushkin dot fm. I read all the emails.
• 43:27
  Speaker 1: You can also find me on x rom LinkedIn.
• 43:31
  Speaker 3: Really do appreciate all the messages that we get.
• 43:34
  Speaker 1: Today's show was produced by Gabriel Hunter Chang and Trina Menino.
• 43:37
  Speaker 1: It was engineered by Hans Dale Sheet and edited by
• 43:40
  Speaker 1: Lydia Jan Kott. I'm Jacob Goldstein, and we'll be back
• 43:43
  Speaker 1: next week with another episode of What's Your Problem.