An economist explains why education won’t save us from being replaced by robots

Education was the key to prosperity in the industrial economy, but it won’t work in the same way as artificial intelligence and robots remodel the job landscape.
Education was the key to prosperity in the industrial economy, but it won’t work in the same way as artificial intelligence and robots remodel the job landscape.
Image: AP Photo/Matt Dunham
We may earn a commission from links on this page.

In 1964, a group of professors, activists, and scientists put together a report that warned US President Lyndon B. Johnson of three impending threats, including a “Cybernation Revolution” that would put massive numbers of people out of work. The committee included a list of recommendations for staving off each crisis.

Number one was “a massive program to build up our educational system.”

Education has for decades been the proposed solution to the threat of technology that replaces human work, as people in diverse occupations from paralegal to fast food worker contemplate the possibility that machines will soon do their work faster and more accurately. In recent months, reforming education and training to meet these future challenges has been recommended by the World Economic Forum, McKinsey, and former New York City mayor Michael Bloomberg.

One reason so many hang their hopes on education is that the strategy has worked in the past. The Industrial Revolution created skilled work not only in factories, where workers ran machines, but also inside of offices. Education became more important than it had been on rural farms, and improved learning helped lift the standard of living for working class people.

But the same strategy won’t work today, writes Ryan Avent in his recently released book, The Wealth of Humans. Avent is a former economic consultant and a current senior editor at The Economist. He spoke with Quartz about why he doesn’t think education will solve the problems he believes increasing automation will cause in the future.

Quartz: You write that “the future of work is either one where employment grows while pay stagnates, or in which the work becomes more productive…and employment stagnates.” What do you mean by that?

Ryan Avent: Improvements in technology over the last couple of decades, and I think this will probably be true in the future as well, have been more about finding ways to do things in place of human beings rather than alongside them. They’re technologies that are substitutes for human work rather than compliments, and I think that the better artificial intelligence gets, the more dexterous robots get, we’ll see more of those kinds of things.

Because machines are increasingly good at doing what humans will do, the only jobs that the average human will really be able to get are those where it’s attractive to hire them because they’re really cheap. It might be that you will be an extra person working in home care, making not much above the minimum wage, taking care of someone’s sick mother. Or will be part of someone’s household staff. You will be a greeter in a hotel. These are jobs where it’s nice to have the people, but only if they don’t earn that much. Because people don’t have the option of dropping out of the labor force altogether, they have to take those jobs.

The other version would be one where, as people are displaced, they’re able to work less, maybe because we’re able to have our welfare benefits be much more generous, or have something in the future like a basic income, which is just a payment the government makes to everyone. For whatever reason…as machines do more and more stuff, we’re able to allow human beings to work fewer hours, or for fewer humans to work.

Education and training were successful strategies for adapting to this type of large-scale technological shift in the industrial economy. Why?

I think there were two reasons it worked before. One, to go back to the first question, was that the technology was a little different. You had big powerful machines, so you needed to have human controllers. And there was a lot of work like that. Henry Ford’s factories didn’t work without thousands and thousands of people.

So the technologies were more likely to work with humans than replace humans. But that’s less true today.

The biggest thing that was different is that, at the beginning of the Industrial Revolution, most people had no education at all. They couldn’t read, they couldn’t write. They basically had no science training. So you could make a huge impact on how much education the population had just by getting people into primary schools, or secondary schools. And teaching them basic math and reading and science. You could double or triple or quadruple the educational attainment of the population. Now, if you’re in an advanced economy, somewhere around 90% of working adults have gone to high school, and around 40% have gotten some sort of post-secondary degree. So it’s much harder to say, we’re going to have the same sort of increase in the educational attainment of the population.

I could be wrong, but I don’t think it’s reasonable to expect that you can take most of those 40% to 50% who are graduating college and then have them go on to get an advanced degree. Because advanced degrees are really hard. I just think the lift that education would have to do to be as big as part of the solution this time around is just way too big. 

Manufacturers often talk about needing people who can use a computer or fill skilled trade jobs. Is it necessary that people have advanced degrees for the kind of retraining they would need?

Not all of the new jobs that are being created will require an advanced degree in computer science. But my view is, the number of jobs that falls into [the category manufacturers describe] is actually fairly small. We’re talking about thousands, maybe tens of thousands, maybe a little more, but we’re not talking about millions and tens of millions of jobs. Whereas when we’re talking about job losses in retail or transportation we are talking about millions and tens of millions of jobs.

The other thing that people talk about when they talk about the future of work are care jobs. It is true that humans are well-suited to do those jobs in some cases, but I think in fewer cases than people think. A lot of people think as therapy, for instance, as something that only a human could do. But that’s not necessarily true. I think that in a lot of cases it might be easier for people to share what they’re feeling with a machine rather than a human who they might feel is judging them.

And so I’m not sure there is going to be as much employment in healthcare as people think. The other thing is that the more employment growth there is in healthcare, the more it means we’re not getting productivity increases in healthcare that would make it cheaper and more accessible. And that would be bad. Ultimately we want to have people get affordable healthcare. And part of what that entails is reducing how labor intensive it is.

Another interesting point you made in the book is that if we train more people to be doctors, doctors will be paid less.

The amount of gain from growth that people are able to capture has a lot to do with how scarce their skills are. It means that you can bargain for a pretty high income.

Over the last two or three decades, wages have been pretty stagnant for most people, but not for skilled professionals. I think that is going to change, increasingly, because the effective supply of people who can do those things will grow. Maybe because we’re able to take a scan from a hospital in America and send it to a doctor in India, and have them look at it, and that’s much cheaper…than bringing more people into the US. Or maybe it’s because we have something like Watson that can do it.

Machines that can do what doctors can do is a lot like bringing in a lot more doctors. It has the same effect.

So if education is not going to save us, and training isn’t going to save us, and apprenticeship programs are not going to save us, and we aren’t necessarily going to be able to count on a universal basic income coming through politically, is there anything that you think would be effective in avoiding this disaster you’ve described?

I think in the short run, about the next ten to twenty years, the problem is actually relatively manageable. We’re close to having driverless cars, but there will still be a decade before there are large numbers of driverless cars on the road. I think there are a lot of straight-forward things that can be done now to make conditions better for workers, and some of that is make all the investments we haven’t been making in infrastructure and research, making it much easier to get preschool education, to get a good primary or secondary education, to go to college without taking on a ton of debt. So there are a lot of kind of common sense stuff that would improve things that we haven’t been doing.

 I think that over a slightly longer time horizon, you don’t have to jump straight to a technological utopia, where everyone is getting a basic income and living in virtual reality worlds all day. There’s a subtle sort of utopia we can reach. If you think about what happened in the early 20th century, we had a bunch of regulations that made life better for workers, so they weren’t getting mauled by machines and children weren’t allowed to work in factories, and there were only so many hours of week they were allowed to work. There were changes made to increase worker bargaining power. And there were all sorts of changes made to the safety net to get people pensions and supplement their income in all sorts of ways.

What happened was that incomes went up a lot, and working hours for most people fell a lot. People were still working full-time, but the idea of working full time in 1955 was a lot different than working full-time in 1890.

And so maybe we sort of move in that direction. We give small extensions to the welfare state, maybe free college and healthcare, a more generous earned income tax credit, which subsidizes people’s wages, and then we also get regulations that reduce the working week, that make it easier for people to take sick leave when they have a child or need to care for a parent.

So in 30 or 40 years, where we end up is most people work 20 hours per week. And they get a little bit more support from the government in terms of what they own.

That’s a massive change, but it’s not necessarily a world that looks completely different than where we are.

Do you think there is any reason to hope that change will materialize?

I think the reason for optimism is that when we faced challenges like this in the past, eventually, we worked it out.