How to Stop Worrying and Love the Robot Apocalypse

In our previous episode, we looked at the benefits and costs of one of the most contentious economic policies around. 

Jacob VIGDOR: Just these really bare-knuckled arguments.

Bare-knuckled arguments about whether the U.S. should raise the federal minimum wage to $15 an hour.

Betsey STEVENSON: There is a constant back and forth where Democrats point to the fact that people simply cannot live on the minimum wage and Republicans point out that all sorts of people will lose their jobs because employers won’t want to hire people at that higher wage. 

Even economists don’t know the right answer.

David NEUMARK: Well, I think we know a lot, actually. We don’t all agree.  

After digging into the issue as deeply as we could, we reached a surprising conclusion.

VIGDOR: It’s like a big fight over what is really not that transformative a policy. 

So a $15 wage is just “not that transformative a policy”? Thanks for wasting our time, economists! But this digging did lead us to something that may be truly transformative.

James ROSENMAN: We are in our infancy of adopting certain robots.

So let’s have that conversation: about robot adoption and the labor markets. We might as well start with an economist.

David AUTOR: No, no, I’m not even a real economist, I just play one at M.I.T. 

That’s David Autor.

AUTOR: I started as an undergraduate at Columbia. I dropped out after three semesters. I worked. I rode a motorcycle. I went back and completed my undergraduate degree at Tufts a couple years later. I studied psychology with a concentration in computer science, and I really didn’t know what to do with myself. 

So, he did some temping, he did construction, he worked at McDonald’s. Then he went back to school again, and got a Ph.D. — in public policy. So, not the typical path for a labor economist at M.I.T. And that real-world experience is reflected in David Autor’s work.

AUTOR: My work is very concrete. I’m not a high theorist; I’m very much driven by practical problems. A lot of the questions I’ve studied are related to things I worked on and saw firsthand, working in poor communities, working in places undergoing political upheaval, watching the gulf of inequality expand in the information age. 

“Watching the gulf of inequality expand in the information age.” Yes, that does sound like a transformative idea. And it leads to a large question: Will new technologies make that inequality gulf bigger — or smaller? You could see it going in either way, right? On the one hand, technology democratizes. Many of us are now rich enough to afford what is essentially a butler. Amazon.com, for instance, will bring you whatever you’d like, quite quickly, at the push of a button. On the other hand: much of the wealth produced by this kind of technology flows way up to the tippity-top of the income ladder. So who are the winners and who are the losers when there’s such a transformative shift in the global economy? Think about one of the last big shifts we lived through: the massive expansion of global trade, during which the U.S. intentionally sent millions of jobs to China. We actually had David Autor on the show a few years back to talk about that — Episode No. 274, if you want to listen, it’s called “Did China Eat America’s Jobs?” So, Autor has done a lot of thinking about these issues.

AUTOR: No country has experienced the extremes of rising inequality that the United States has. And there’s no evidence that the U.S. has gained much from it. We haven’t grown faster than other countries; we don’t have higher labor-force participation rates; we don’t have higher social mobility of people going from rags to riches. 

If you wanted a SparkNotes version of the U.S. economy over the past few decades, it would be this: rising productivity, though not as fast a rise as the post-war era; and stagnant median wages, with the productivity gains largely benefiting the top of the income distribution.

AUTOR: Yeah, it’s just incredibly skewed, and so as far as we can measure it, the median is barely budging. 

And now, after all that, it’s time to consider another very, very large disruption. Because you know that robot future you’ve been hearing about?

DAVE (2001: A Space Odyssey): Open the pod bay doors, HAL. 

HAL (2001: A Space Odyssey): I’m sorry, Dave. I’m afraid I can’t do that.

Yeah, well, the future got here yesterday.

FAMILUSI: Good to see you again. I like your shirt.

PATIENT: Thank you.

FAMILUSI: So tell me, how you feeling today? 

PATIENT: I’m feeling pretty good.

FAMILUSI: You’re welcome. 

*      *      *

In the spring of 2018, David Autor was asked to co-chair an M.I.T. task force called The Work of the Future. It included researchers from a variety of disciplines — economics, engineering, political science, anthropology even. The mission was to explore how new technologies like robotics and automation will affect labor markets, especially whether certain groups of workers would be left behind. Keep in mind that this sort of prediction is really hard — as evidenced by the predictions that economists made about globalization. They predicted that when the U.S. offshored manufacturing jobs to China, that Americans who worked in manufacturing would be made better off, since they’d theoretically be “reallocated” into better jobs. But as David Autor told us in that earlier episode, this didn’t happen.

AUTOR: Some people are leaving the labor market, some people are going into unemployment, some people are going on to disability, and so the reallocation process seems to be slow, frictional, and scarring. The real differentiator is the skill level of the worker. So, higher-paid and more highly educated workers, they seemed to reallocate successfully out of manufacturing into other jobs.

DUBNER: So, the H.R. person at a big textile firm gets an H.R. job elsewhere and the manufacturers on the line are probably not.

AUTOR: The line workers are much less likely to do so, exactly.

So, considering the difficulty of making predictions about “the future of work,” the M.I.T. task force started with one thing they were pretty sure about.

AUTOR: The one thing we were confident in was that the U.S. would keep generating lots of low-wage jobs. 

DUBNER: Too many even, yes? 

AUTOR: Well, actually, too many is better than too few. When there’s too many, at least they’re competing hard for workers. When there are too few, workers are competing for them and that means those jobs will get worse. And so, the one positive thing you could say about the U.S. workforce? Well, we had a lot of crappy jobs. 

The task force began its work before the pandemic. What kind of damage has been done by the pandemic recession?

AUTOR: In the short run, it’s just done enormous damage to most of the in-person service jobs. The ones that were absolutely necessary, like in grocery stores and healthcare, have persisted. But many of the jobs in retail, in restaurants, and hospitality have not. 

A lot of those jobs paid only the minimum wage. Even so, many of them won’t be coming back.

AUTOR: I actually think the biggest change, most obvious, is telepresence, that we are just doing more things remotely. We’ve kind of broken the space-time barrier in that we can’t be in two places at once, but we can get to any two places instantly.

No one can say how fully the old normal will return. But consider just one pandemic change: the massive drop in business travel. Let’s say business travel never returns to pre-pandemic levels. What kind of downstream labor effects does that have?

AUTOR: It’s not just airplanes, right? It’s Ubers and limos, it’s expensive hotels that people pay full freight on weeknights and then go out to marquee restaurants and then go have their shoes shined, and dry cleaners. And so, I think that’s the real challenge. 

The Work of the Future task force took the pandemic into account as best as they could. And they recently published their report. It tries to answer three main questions. The first one: “How are emerging technologies transforming the nature of human work and the set of skills that enable humans to thrive in the digital economy?”

AUTOR: You know, technology is always eliminating work and creating work simultaneously. We tend to focus on what is automated away, and that’s completely reasonable. Simultaneously, new areas of expertise, new luxuries, new services, new demands are constantly being created, and that process, that kind of turnover, is highly productive. 

Consider, for example, how modern medicine is practiced.

AUTOR: There’s hundreds of medical specialties — way, way more than there used to be — and it’s not because doctors have become narrower and narrower and they know less and less. It’s that they know more and more in-depth rather than breadth. The extent of expertise required is just extraordinary, and humans have finite capacity. Where did all that need for expertise come? Well, it came from research and technology and so on. So often we’re broadening expertise. But it’s not just in the high-tech professions. You will find patents emerging for new ways of hardening nails — fingernails, I mean, not the nails you pound into wood. Patents for solar-voltaic electricians, people who install solar cells. There’s a lot of skilled work that’s done hands-on — being an electrician, being a plumber, building a home or repairing an engine.

And much of that work requires a combination of dexterity and flexibility and problem-solving and also knowledge, knowledge on demand. A lot of people today consult YouTube when they want to learn how to sweat a pipe. We can augment people’s capability to do that work by giving them V.R. tools, giving them information on demand. People could be much more effective in that work and more productive, and therefore paid more, if they were augmented in these ways. And so, you can see in those examples how you could use the technology to not make people less necessary, but to make them more effective. 

That said, not every profession benefits from this kind of tech augmentation.

AUTOR: If you’re doing one of those things that all of a sudden a machine can do better than you, your opportunity set contracts and usually the people who are on the one end of that, seeing their work disappear, are not the same people who are getting new opportunities. 

We saw this vividly when the U.S. offshored manufacturing jobs, and we’re seeing it now in other sectors.

AUTOR: For the people who have been working in clerical jobs or many production jobs, what automation has done is made their work unnecessary. 

It’s tempting to think that automation will replace only the simpler jobs that don’t require heavy cognitive input. But that’s not the case. Autor has seen this for himself as some of the firms he’s visited.

AUTOR: One of them was a big insurance company and they do an enormous amount of claims adjudication and claims assessment, and they have these floors of I guess you call them forensic accountants, and they go through a lot of material looking for anomalies, looking for fraud, looking for overpayment and so on. 

It is true that forensic accounting requires a high level of expertise. But combing through these files in search of anomalies is also a tedious task.

AUTOR: And the automation has really accelerated that discovery work. Machines can actually do reasonably well with this, and simultaneously they never run out of attention. They never run out of energy. 

DUBNER: Let’s say machine learning and artificial intelligence can be used to find these anomalies. Does that mean that the people who used to find the anomalies are out of work or they have a different style of work, a different amount of work? 

AUTOR: So definitely the total headcount of people who need to do this work is shrinking. Now, they’re mostly not firing people, but they slow down hiring. The work I think that remains is quite interesting. There’s less tedium and more action. But it does ultimately mean, I think, reduction in the number of people doing that work.

The automating of work is itself big business, even if it hasn’t gotten much attention yet. One of the major firms in this space, UiPath, practices what’s called robotic process automation. UiPath is currently valued at around $26 billion and plans to go public soon. Companies like Microsoft have also bet big on automating software. If you are the kind of person who hears this and shudders at the thought that technology is destroying our way of life — well, there is a long history of such thought. Aristotle had the same concern, and in ancient Rome, some technologies were outlawed because of the expected job loss. In the most recent century, if you’ve ever watched a movie, you’ve likely come across at least one fever dream of technology run amok.

FRANKENSTEIN: It’s alive, it’s alive, it’s alive!

And fears of a robot apocalypse.

RICK DECKARD (Blade Runner): A blade runner’s job is to hunt down replicants, manufactured humans you can’t tell from the real thing.

THE TERMINATOR (Terminator 2): Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14 am Eastern Time August 29th. 

MORPHEUS (The Matrix): We don’t know who struck first, us or them. 

But if the overall goal is to make good policy and economic decisions about our collective future, we probably shouldn’t base that policy on movie plots. As history has shown — again and again and again — the fear of new technologies tends to be overstated, and the gains from technology make most people better off. But maybe, you’re thinking, maybe this time is different. In the old days, when the automobile replaced the horse and carriage, if you lost your job as a carriage maker or a stable hand, you could probably find work in an auto plant. What about today? A recent paper by Daron Acemoglu and Pascual Restrepo found that a single industrial robot will typically reduce employment by as many as six human workers. David Autor again:

AUTOR: And I can understand why companies would do that, it makes a lot of sense. Labor is a cost. No one hires workers for the fun of hiring workers. They hire workers because they need things done. If they could have machines that did it without complaining and cost less, that’s what they would do. But we have a public interest in something more than that. We’re going to have lots of people. The machines ultimately work for the people; we want to augment the people. And there are many highly valuable social problems that could use automation, could use investment, and we underinvest in. For example, healthcare. 

Consider this healthcare checkup:

NURSE: Hi, I have Dr. Familusi, he’s going to be evaluating you today. Okay? 

PATIENT: Okay. 

FAMILUSI: Hello, my dear. How are you? 

PATIENT: Hi, I’m okay. How are you?

FAMILUSI: Good to see you again. I like your shirt! So tell me, how you feeling today?

PATIENT: Yeah, I’m feeling pretty good. No complaints today, really. 

Abiola Familusi is a doctor who works with a nursing home in Westchester County, just outside of New York City. It’s called Andrus on Hudson.

FAMILUSI: Can you open your mouth for me? Say “ah.”

PATIENT: Ahhh. 

FAMILUSI: Okay. Can you lift both arms up for me? Lift your arms up for me, both of them. Excellent. 

But here’s the thing: Dr. Familusi isn’t at the nursing home today. He’s examining the patient remotely.

ROSENMAN: Yeah, we are in our infancy of adopting certain robots.

That’s James Rosenman. He’s the C.E.O. of Andrus on Hudson.

ROSENMAN: We have two robots: one for the purposes of telemedicine, so that physicians can go into patient rooms with the assistance of a nurse when they can’t be physically available on site. 

This telemedicine robot doesn’t look like much — or at least not like what you may think a robot should look like.

ROSENMAN: Yeah, an iPad that is on a base that has wheels that can move to various areas. And we also have another robot that is a social robot to visit certain residents that may be less able to get up and walk around.

DUBNER: I understand you had eight robotic dogs and 11 robotic cats. Did you have to pull them, then, because of Covid concerns? 

ROSENMAN: They’ve been put in the kennel for a little while. The problem with the robots in the environment we’re in right now is that you can’t have them just roaming about. So infection control has added this other layer of complexity to robotics.

So the pandemic is the reason the robotic dogs and cats had to be sidelined. But the pandemic was also the reason that Andrus got the telemedicine robot. Nursing homes were a hot zone for Covid transmission, so Rosenman wanted to minimize face-to-face contact.

DUBNER: Are you pretty typical as far as a nursing facility with the amount of robots you have? Are you at the leading edge or are you lagging? 

ROSENMAN: It’s hard to know where we stand in comparison to other providers because this isn’t a topic that comes up very often. But we do know that a lot of the people that we talk to don’t utilize those in their facilities. 

Also: James Rosenman is a self-proclaimed robot nerd.

ROSENMAN: I think I watched Short Circuit when I was little, the movie.  

JOHNNY FIVE (Short Circuit): Oh, I get it!

ROSENMAN: Johnny Five was a big inspiration for me. 

But there are other, non-pandemic reasons a nursing home or hospital might want to use robots.

ROSENMAN: Yeah. We work very hard on staff retention and we do have a good retention rate.

But—.

ROSENMAN: But we also have people, you know, they retire. We would love them to work there forever and ever. And I’d love to clone people, but we can’t — maybe that’s for another show. But we have a labor shortage in the market of nurses and of CNA’s.

A CNA is a certified nursing assistant. In the U.S. today, there are roughly 4 million RN’s, or registered nurses. A recent study in the American Journal of Medical Quality estimated that by 2030, less than a decade from now, there will be a shortage of half a million RN’s. This gap is driven by both demand — we have a large population of elderly and sick people — and supply — there are more nurses aging out of the workforce than entering it.

ROSENMAN: I’ve continued to see this labor shortage get worse and worse.  

DUBNER: How hard is it for you to hire already?

ROSENMAN: It’s incredibly difficult. It is a very difficult and demanding job. There is a critical shortage of those individuals.

Andrus is a pretty typical nursing home, aside from their new robot workers. They’ve got about 190 residents and nearly 250 employees. The typical resident is over 70, and has a variety of conditions.

ROSENMAN: Respiratory conditions, C.O.P.D., general chronic respiratory failure, congestive heart failure, cancer. 

The nursing assistants manage a lot of the moment-by-moment care; they earn around $20 an hour. Nurses earn around $30 to $40 an hour. Forty dollars an hour works out to around $83,000 a year. And what do these robots cost?

ROSENMAN: It was $4,000 for one of the robots that we are using for socialization. And then for the medical robot, we lease that. We pay about $2,000 a month because it has all the equipment.

DUBNER: Equipment meaning EKG possibility?

ROSENMAN: Exactly. With the telemedicine robot, one of the key components is not just that the clinician can look at the patient and assess them, but it has an array of tools connected with it. So you have what they call a smart stethoscope. So that directly feeds into what the physician can see on their end. You know, an EKG onsite and an ultrasound is something that we’re looking on adding. 

FAMILUSI: Your pulse rate is very good. 

NURSE: Your oxygen saturation is 98 percent. 

James Rosenman says the robots have increased productivity at the nursing home — and, better yet, they’ve helped improve patient outcomes.

ROSENMAN: You know, one area that is always of concern, individuals who come to us for short-term rehabilitation, and then something happens medically with them. Then we have to send them back out to the hospital. It’s called a readmission. And so, we realized that by adding the robot and having faster access to clinicians, to be able to view something in real time, assess it, we were able to fairly significantly reduce readmission rates to the hospital, just through that alone. 

For David Autor, the M.I.T. labor economist, these nursing-home robots can help answer the second question that his Work of the Future task force asked: “How can we shape and catalyze technological innovation to complement and augment human potential?”

AUTOR: You could introduce so much technology into healthcare without reducing employment and yet expanding the quality of care and the quantity of care. And, of course, you’ll need tons and tons of people to actually do the hands-on care work. 

But is that reading of the situation too optimistic? Coming up after the break: a fascinating new study about Japanese nursing homes:

Karen EGGLESTON: What we’re really worried about are the lower-skilled workers that might be completely replaced.

And: Why is the Andrus nursing home an outlier? Why is the U.S. a laggard when it comes to healthcare robots?

ROSENMAN: Sort of hard to understand. 

*      *      *

The M.I.T. labor economist David Autor was co-chair of a task force on the future of work — specifically, how the U.S. workforce is integrating and adapting to new technologies. The task force found that the U.S. is not nearly as adept as one might hope in this regard. Here’s what they wrote in their final report: “Institutional changes and policy choices … failed to blunt, and in some cases magnified, the consequences of these pressures on the U.S. labor market.”

DUBNER: So David, of all the rich countries in the world, how would you rank the U.S. in terms of successfully adapting to the future of work? And assuming that we are not in, let’s say, the 90th percentile or above, why are we trailing? 

AUTOR: I would put the U.S. maybe at the bottom of the top dozen. On the plus side, let’s give the U.S. a little bit of credit, it’s incredibly creative and entrepreneurial. A lot of a lot of the technologies originate here. But in terms of dealing with the consequences as opposed to the opportunities, that’s where we have been extremely poor. Low-wage workers in Canada make 25 percent more per hour than low-wage workers in the United States. It’s hard to believe that Canadian workers are actually 25 percent more productive per hour at McDonald’s than U.S. workers. That seems very unlikely. 

DUBNER: How are those wages so much higher in Canada? 

AUTOR: There are minimum wages and then there are just norms about what is acceptable. And the U.S. has thrown away those norms. To a substantial extent, we’ve convinced ourselves that those norms are the problem, not the solution.  

Aside from those norms, there is also the fear that new technologies will destroy more good jobs than they create, or at least that the productivity tradeoff won’t be worth it. But not all countries feel that way, especially when it comes to robots.

EGGLESTON: I think a lot of people just weren’t aware that Japan’s been subsidizing robot adoption since 2015.

Karen Eggleston is an economist at Stanford.

EGGLESTON: It’s beautiful. You can hear the birds chirping.

A lot of Eggleston’s research looks at healthcare and technology in Asia. Why that focus?

EGGLESTON: Well, Asia is a very important part of the world, and a part of the global economy. I also have family connections to Asia.

When it comes to robots in the workforce, Japan is No. 2 in the world on a per-capita basis. No. 1, by a long shot, is South Korea. Most of those are industrial robots used in the production of automobiles and electronics. The countries with a lot of robots tend to be high-wage countries — which makes sense, since higher wages create more incentives to replace human workers. Germany, Sweden, and the U.S. all have a high share of industrial robots. China, meanwhile, is well down the list — which also makes sense, since labor there is a lot cheaper, at least for now. When it comes to Japan, Karen Eggleston says that robots have been embraced for several reasons.

EGGLESTON: First of all, we know Japan is a very developed economy and invests a lot in many kinds of new technologies, from so-so technologies to brilliant technologies. So investing in robots was natural in that context. 

A “so-so technology” is economist-speak for something that just doesn’t perform very well, especially when it’s new. Think of automated phone services and self-checkouts in grocery stores.

EGGLESTON: Second, and more related to what I usually study, is that the population age structure in Japan is such that it’s leading the world in the demographic transition. And so, therefore has an overall declining and a declining working-age population. 

Japan in fact has the oldest population in the world.

EGGLESTON: So you have an increasing demand for long-term care and a declining supply of workers to staff that long-term care. 

This is the same dynamic that James Rosenman of the Andrus nursing home told us about. But it’s even more pronounced in Japan. A lot of countries ease the burden of an aging population by importing labor.

EGGLESTON: But as many people know, Japan is less welcoming of immigrant labor than many other countries in the world and has actually had a long-standing acceptance of robots.  

DUBNER: I feel like I read that a few years ago, Japan had finally started to loosen up some of the immigration. Is that right? 

EGGLESTON: Japan does continue to loosen immigration. So it’s certainly not a black or white thing. But it’s just relative to many other countries, where the labor market conditions might be different.

In other words, Japan might have opted for more immigrant labor to help care for its aging population but instead, it invested heavily in robots.

EGGLESTON: So they don’t all look like R2D2 or C3PO, but they have functionality that enables them to take actions based on what they’re monitoring. And a “cobot” is a term that’s developed for robots that work alongside humans. 

“Cobot” as in a collaborative robot. It is a very different machine than the kind of robots used in something like auto manufacturing.

EGGLESTON: Correct, yeah. Those robots can swing their arms without worrying that they’re going to knock over a human and damage them. 

DUBNER: And then a cobot is defined as necessarily working alongside humans, is that right? 

EGGLESTON: That’s the idea, is that they can work alongside. They’re not only aware, physically, of a human’s presence, but they can productively interact with a human.

In Japanese nursing homes, there are a variety of cobots designed to accomplish a variety of tasks. One type, for instance, is designed to monitor patients.

EGGLESTON: So these can help both the caregivers and the people themselves to avoid falls, particularly if they roll out of bed at night or they get up and then trip on something. 

There are also cobots to help the nursing-home staff move their patients. 

EGGLESTON: They have these big robots with big arms that help to pick people up. Others that actually are worn by the caregiver really need to strap onto the body when they’re trying to move someone from the bed to a chair or back again. So they’re not shaped like a human, but to fit onto a human body. And these robots are trying to address the issue of back pain that caregivers often experience, and leads to turnover and therefore poor outcomes for long-term care. Other robots help with other activities of the individual, such as being able to move directly themselves, and to function independently, to help with taking a bath or walking around. 

So, unlike the typical robot, a cobot is designed to compliment human labor rather than replace it. That, at least, is the theory. Karen Eggleston, being an economist, wanted to test this theory. She and two colleagues — Yong Lee and Toshiaki Iizuka — set out to gather and analyze data from 860 nursing homes in Japan.

EGGLESTON: We focused on nursing homes partly because that’s where this population-aging question is really most manifest, and also because the huge debate about technologies is — yes, we know that surgeons’ jobs will be affected by technology. But what we’re really worried about are the lower-skilled workers that might be completely replaced. A lot of the research in manufacturing has shown that to be certainly a worry that has foundation.

Eggleston and her co-authors were able to collect a variety of data for this study. First: wage and employment data from these nursing homes. This included whether a given employee was a so-called “regular worker,” which was usually a full-time position and paid fairly well; or a lower-paid “non-regular,” meaning a part-time or flex worker. The researchers also measured the degree of cobot adoption in a given nursing home. But they needed to introduce a random variable to prove causality between the adoption of robots and the effects on staffing. Luckily for them, different prefectures across Japan subsidize cobots at different rates, some as high as 50 percent. This variation in subsidies gave the researchers a nice natural experiment.

EGGLESTON: And we use the variation in those subsidies to help figure out which way the causality arrow goes.

Eggleston and her colleagues have written a working paper called “Robots and Labor in the Service Sector: Evidence from Nursing Homes.” What’d they find?

EGGLESTON: What we find is that robot adoption is strongly correlated with having a much larger nursing home. And it appears to be a causal impact, that adopting robots is associated with more care workers rather than fewer. But these additional care workers are the non-regular type, on more flexible contracts.

DUBNER: So, that sounds as if it could mean that robots are bad for the upper end of that employment spectrum, considering that this is relatively low-paid work anyway. It sounds like it would promote more human workers, but at a lower wage. Is that about right? 

EGGLESTON: Well, yes, it is possible, although we also know that the most commonly adopted robot is the monitoring robots we were talking about. And they are helping to reduce the long night shifts that nurses and care workers have to do. So we think that part of the effect is that the workers have a reduced burden of care. And yes, we do find a lower wage — of a modest amount — for the regular nurses. But if the case is that they have shorter work days, then it’s not clear that that’s actually a welfare loss.

DUBNER: When I first read your paper, the sort of sunny headline that I wrote in my head was: “We thought robots were the enemy of workers and now it looks like they’re best friends.” That’s a little bit too sunny, isn’t it? 

EGGLESTON: Yeah, I think it is a little sunny, although it is a little bit surprising. And depending on how they’re adapted, this automation, yes, it will replace some of the tasks that care workers do. But the ones that do end up staying in this profession, maybe they will have more support, less back pain, have the education to work alongside robots, and may find that a more enjoyable experience, as well as better for the people they serve. A lot of the workforce feels burned out, not necessarily because they don’t like doing what they do, but they don’t like doing all that paperwork and all that other stuff. And they want to interact one-on-one with the people they care for. And cobots, if they work properly, will enable that. Humans have these qualities of being very dexterous and being able to care directly to the patient and communicate and have compassion with them.  

You could argue that healthcare is the ideal scenario for the blending of human and robot labor. There are countless tasks and procedures where technology can plainly be helpful; but the human appetite for compassion also seems boundless, and for now at least humans are better at compassion. You could see cobots helping mightily not just in hospitals and nursing homes, but in at-home care as well. A recent study from the Journal of the American Medical Association found that some five million older adults in the U.S. need help with bathing or using the bathroom. In Japan and elsewhere in Asia, and also in Europe, it is increasingly possible for a robot to assist with such tasks. That’s not just because robots have been subsidized; they’ve also been deregulated.

ROSENMAN: In the United States, we don’t have access to a lot of these types of robots. 

That, again, is James Rosenman, C.E.O. of the Andrus on Hudson nursing home. And why don’t we have more access to these types of robots?

ROSENMAN: It’s a good question. When I look at a lot of these things, or I’ll find something and my eyes will get huge. You know, I’ll do some research on that late at night. And then I find out it’s only available in Japan or the E.U. Actually, in many, many markets. And the glimmer goes out of my eye because I know that we can’t, legally, import that to the United States. The other day, I was just looking at — for example, to reduce the incidence of individuals developing pressure ulcers, for people who are more bed-bound, the current thinking is that you rotate people so that you can increase blood flow and reduce pressure on one given part of the body. So the idea that I was thinking about was maybe there are beds, beyond just the mattresses that are pressure-relieving, a robotic bed to literally move people. Right now that’s being done by humans. It’s not available in the United States. 

DUBNER: So, is it regulation that’s preventing this right now? And if so, what kind of regulation — is it technical regulation? Is it medical regulation, etc.? 

ROSENMAN: Sort of hard to understand. I think that some of it is like a pie chart, if you will, of different reasons. I don’t think there’s one sort of smoking gun or people in the back room saying, “All right, let’s not get these things rolled out, because it goes against our interests.” It’s just very fragmented. And so, you have these different regulatory authorities. You have, who’s going to pay for it? How is it going to be used? You know, you can have it approved, but then you have, how is it used in practical terms on site? I think that first and foremost, there need to be more pilots, studies, models. There are pilots going on every day. Medicare funds those, or they’re funded by other agencies of the federal government. But there haven’t been a lot of pilots that include robotics in our settings. 

So, if you’re thinking big-picture about the future of work, one of the most compelling questions is the degree to which robotics will complement human labor versus replace it. 

LEE: One example that I’ve encountered is in a construction company.

That’s Yong Lee, one of Karen Eggleston’s co-authors on the Japanese nursing-home paper. He too is an economist at Stanford.

LEE: They initially created robots so that they could replace workers — for instance, digging out certain parts of the land to lay the foundation. But they needed people who had years of experience, more than 10 or 20 years of experience. And it was just difficult to find that labor anymore. So what they decided to do is to create a robot where an individual with maybe only one year of experience could operate a machine that could perform the tasks that a skilled laborer with 20 years of experience could perform. So in this sense, they were designing a robot not to replace the skilled individual, but actually to augment an individual with less skill. 

In another study, Lee looked at robots in the manufacturing sector — a study that covered 11 years. There too, he found that robots at first were replacing workers but later, as the technology matured, the robots became more collaborative.

LEE: Robots 10 years ago that did welding and robots 10 years later will likely be different. 

So, how do economists see this relationship unfolding, between human workers and smart machines? How can that relationship be optimized? Karen Eggleston again:

EGGLESTON: It won’t surprise you to know as an educator and a researcher that I believe that investment in human capital is really, really important. And we need to be investing in young people and everyone else to enable them to be lifelong learners and to be adaptable. If we give support to people to be adaptable to changes in the labor markets, there really is a possibility that it will work on behalf of a very broad spectrum of society.

DUBNER: In other words, every piece of technology, in a way, could become a cobot if we humans are skilled enough to collaborate with them. 

EGGLESTON: Yes, yes. I think there really is a potential for technology to make our lives better. But I’m not of that opinion that it’s going to automatically happen. I think it comes down to the choices that we make, particularly in policy, on behalf of the most vulnerable in our society.

AUTOR: We have time to adapt — our institutions, our educational systems, and the way we work. 

And that, again, is the M.I.T. economist David Autor. The third and final question from his task force on the future of work was this: “How can our civic institutions ensure that the gains from these emerging innovations contribute to equality of opportunity, social inclusion, and shared prosperity?”

DUBNER: The problem strikes me, as a layperson, is maybe a gigantic coordination problem. Because we look to our governments to coordinate the way jobs and the economy will flow and take care of everybody, but in fact, governments aren’t really very equipped to do that, whereas firms have a different set of incentives. So, can you just describe how that will unfold in a way that leaves people not either out of work or grotesquely underpaid or working in an economy where the gap between the high and low just gets bigger and bigger?

AUTOR: So first, I want to argue that the government actually can do a lot and that and we in America tend to deride our government and assume it can’t be effective. But in many ways, history demonstrates just the opposite. You don’t have to look very far back in history, just look back a year when the government passed the CARES Act, and overnight essentially took 10 percent of G.D.P. and said, “Hey, we’re going to send this to households, to businesses and to the unemployed, to keep this pandemic from turning into an economic catastrophe.” And it was highly effective. And the government, similarly, has been effective in shaping technology over many generations. The U.S. had a leading patent system. It’s in our Constitution. But the U.S. has also invested in R&D through our universities in health development and so on. So, it actually plays a big role in even setting the rules of the road.  

To that end, the M.I.T. Work of the Future task force had some concrete recommendations. They include: heavy investment in education and job training, both in schools and through private firms; improving the quality of existing jobs via policies like a higher minimum wage and labor-organizing protections; and reforming the tax incentives that privilege capital investments over labor. If you think all that sounds a lot like the recommendations we’ve been hearing about for a few decades now — I agree. So you might be forgiven for thinking these adjustments won’t happen, at least not in time to deal with the robotic revolution. But David Autor isn’t panicking. The revolution may be inevitable, but it’s not instantaneous.

AUTOR: The technology is spectacular and it’s going to have momentous impacts, but they’re unfolding gradually. They often take years to decades — think about the gap between the hype about driverless cars and the number that you don’t yet see on the roads. And many of the things are still a ways off. I mean, these things will happen, but they take time. 

Let me ask you to cast your mind forward, let’s say, between 10 and 20 years. It’s pretty easy to foresee that a lot of low-skill jobs will be replaced or very much amended. But let’s say even a lot of medium- and high-skill ones, let’s say economists, and, writers, and podcasters, and forensic insurance agents. Let’s say that many, many, many of those get essentially wiped out by some combination of robots, and cobots, and artificial intelligence, and machine learning. Wouldn’t that mostly be a wonderful thing?

AUTOR: So it’s wonderful in one sense: It means we are now much richer. We can do everything we are doing and yet not using labor to do it — so we have incredible leisure opportunities, therefore we have incredible productivity, incredible wealth. The problem that creates is twofold. One is a huge distributional challenge. Our main method of income distribution in this country and in most industrialized economies is ownership of labor. You have some labor, you invest in your skills, and then you sell those skills and labor to the market for 30, 35 years, you save up some money, you retire. If labor is no longer scarce, what claim do you have on the assets of that society? So, I worry about that problem, the problem of abundance, actually, the problem of lack of labor scarcity. The other is I do think work, you know, one can oversell it, but work should be venerated to some degree. It gives people identity, gives them structure, it gives them purpose.

I mean, this is what the Calvinists have always told us, but how do we know this is true?

AUTOR: Well, we know when people lose work, they are miserable. So, if we’re going to have less work, I’d like to see everybody have a little bit less rather than many people not working at all. 

David Autor is a lot smarter than me, so I’m inclined to believe him when he says that people are miserable when they lose work. On the other hand, could it be that people who’ve lost work in the past have been miserable because our civilization is built around work as the primary means to satisfy your basic needs? If the assets of society, as Autor puts it, are so bountiful at some point in the future, shouldn’t there be a way to share in those assets while our robot and cobot friends do most of the work? Some people are lucky enough to love their work. I’ll be honest, that describes me, most days at least; and I’m guessing it describes David Autor too. But many, many, many people have jobs they do not love, and which keep them from what they do love. Economists are pretty good at measuring utility; but they’re not very good yet at measuring things like love. Maybe, if the robots and cobots are really smart, they can teach the economists how to do that.

*      *      *

Freakonomics Radio is produced by Stitcher and Renbud Radio. This episode was produced by Zack Lapinski. Our staff also includes Alison Craiglow, Greg Rippin, Mark McClusky, Matt Hickey, Mary Diduch, Emma Tyrrell, Lyric Bowditch, and Jacob Clemente. We had help this week from Jasmin Klinger. Our theme song is “Mr. Fortune,” by the Hitchhikers; the rest of the music was composed by Luis Guerra. You can subscribe to Freakonomics Radio on Apple Podcasts, Stitcher, or wherever you get your podcasts.