Technology often creates jobs for young, skilled workers. Will AI do the same? | MIT News

At any given time, technology does two things in employment: It replaces traditional jobs, and it creates new lines of work. Machines are replacing farmers, but they are making, say, aeronautical engineers possible. So, if technology creates new jobs, who gets them? How well do they pay? How long do new jobs stay new, before they become just another routine job done by any job?

A new study of American employment led by MIT economist David Autor sheds light on all these issues. In the postwar US, as Autor and his colleagues show in great detail, new forms of work tend to benefit college students under 30 more than anyone else.

“We’ve never seen anyone doing new work before,” said Autor. “It’s mostly done by young and educated people, in urban areas.”

The research also contains a powerful insight of a larger scale: Most innovation-based activity is demand-driven. The expansion of government-sponsored research and production in the 1940s, in response to World War II, resulted in a large number of new jobs, as well as new technologies.

“This means that wherever we make new investments, we end up with new skills,” said Autor. “When you create a large-scale work, it’s always going to have the potential for new specialized knowledge that’s important to it. We thought that was exciting to see.”

The paper, “What Makes a New Work Different from the Others?” coming to Annual Review of Economics. The authors are Autor; Caroline Chin, doctoral student in MIT’s Department of Economics; Anna M. Salomons, professor at the Department of Economics of Tilburg University and the School of Economics of Utrecht University; and Bryan Seegmiller PhD ’22, assistant professor at Northwestern University’s Kellogg School of Management.

And yes, learning about the new job, and the types of workers who find it, may be relevant to the deployment of artificial intelligence — although, in Autor’s estimation, it’s too early to tell how AI will affect the workplace.

“People are really worried that AI-based automation will destroy certain jobs very quickly,” Autor noted. “Destroying jobs is not the same as destroying jobs, as many jobs involve many jobs. But we all say: Where will the new job come from? It is very important, and we know little about it. We don’t know what it will be, what it will look like, and who will be able to do it.

“If everyone is an expert, no one is an expert”

The four co-authors also collaborated on a previous large-scale study of new jobs, published in 2024, which found that nearly six out of 10 jobs in the US from 1940 to 2018 were in new jobs that had advanced significantly since 1940.

To do that, the researchers used US Census Bureau data from 1940 to 1950, as well as Census Bureau’s American Community Survey (ACS) data from 2011 to 2023. In the first case, because Census Bureau records become completely public after about 70 years, scholars can examine individual-level data and track individual-level data about occupations, and track many workers. Census figures for 1940 and 1950.

Through a collaborative research program with the US Census Bureau, the authors also obtained secure access to individual-level ACS records. This data allowed them to analyze the salary, education, and other demographic characteristics of workers in new jobs — and compare them to workers in the past.

New work, Autor notes, is always connected to new types of technology. At first, this technology is scarce; over time, it may become more common. In any case, technology is often linked to new types of technology.

“It takes a certain skill,” Autor said. “What makes a job valuable is not just the ability to do things, but specialized knowledge. And that often separates a high-paying job from a low-paying one.” Moreover, he adds, “It should be scarce. If everyone is an expert, no one is an expert.”

By examining census data, scholars found that back in 1950, about 7 percent of workers had jobs in the types of jobs that had emerged since 1930. More recently, about 18 percent of workers in the 2011-2023 period were in occupations introduced since 1970. hard and fast practice.)

In these times, new jobs have emerged more often in urban areas, with people under 30 benefiting more than any other age group. Getting a job in a series of new jobs appears to have a lasting effect: People employed in a new job in 1940 were 2.5 times more likely to find a new job in 1950, compared to the general population. College graduates were 2.9 percent more likely than high school graduates to take up a new job.

The new job also has a premium, that is, the wages are better overall than in the existing jobs. But as research shows, that wage premium also fades over time, as certain technologies in many new job types become more widely understood.

“The deficit is eroding,” Autor said. “It becomes common knowledge. It becomes automatic. New work becomes obsolete.”

After all, Autor points out that driving a car was not a skill that was lacking. For that reason, we were able to use word processing programs like WordPerfect or Microsoft Word, from the 1990s. After a while, however, being able to handle word processing tools became a basic part of using a computer.

Back to AI for a minute

Studying who gets new jobs has led researchers to some surprising conclusions about how new jobs are created. Examining state-level data from the World War II era, when the federal government supported new manufacturing in public-private partnerships across the US, studies show that states with new factories had the most new employment, and that 85 to 90 percent of new employment from 1940 to 1950 was technology-driven.

In this sense there was a great change driven by necessity at that time. Today, the public discourse about innovation tends to focus on the supply side, that is, inventors and entrepreneurs trying to create new products. But research shows that the demand side can greatly influence innovation activity.

“Technology is not like, ‘Eureka!’ where it happens,” said Autor. “Innovation is a purposeful activity. And innovation is cumulative. If you get far enough, it will have momentum. But if you don’t, it will never get there.”

Which brings us back to AI, a topic that many people are focused on in 2026. Will AI create great new jobs, or will it eliminate work? Well, maybe it depends on how we use it, Autor thought. Think of the large health care sector, where there could be many new types of technology-driven jobs, if people were interested in creating jobs.

“There are different ways we can use AI in healthcare,” Autor said. “One is for people to be unemployed. The other is to allow people with different levels of expertise to do different jobs. I would say this is good for society. But it is not clear where the market will go.”

On the other hand, perhaps through government-driven demand in various ways, AI can be used in ways that ultimately increase productivity in the health sector, creating new jobs as a result.

“More than half of US health care dollars are public dollars,” noted Autor. “We have a lot of power there, we can push things in that direction. There are different ways to use this.”

This research was supported, in part, by the Hewlett Foundation, the Google Technology and Society Visiting Fellows Program, the NOMIS Foundation, the Schmidt Sciences AI2050 Fellowship, the Smith Richardson Foundation, the James M. and Cathleen D. Stone Foundation, and the Instituut Gak.