At any given time, technology does two things to employment: It replaces traditional jobs, and it creates recent lines of labor. Machines replace farmers, but enable, say, aeronautical engineers to exist. So, if tech creates recent jobs, who gets them? How well do they pay? How long do recent jobs remain recent, before they grow to be just one other common task any employee can do?
A brand new study of U.S. employment led by MIT labor economist David Autor sheds light on all these matters. Within the postwar U.S., as Autor and his colleagues show in granular detail, recent types of work have tended to profit college graduates under 30 greater than anyone else.
“We had never before seen exactly who’s doing recent work,” Autor says. “It’s done more by young and educated people, in urban settings.”
The study also incorporates a robust large-scale insight: Loads of innovation-based recent work is driven by demand. Government-backed expansion of research and manufacturing within the Forties, in response to World War II, accounted for an enormous amount of latest work, and recent forms of experience.
“This says that wherever we make recent investments, we find yourself getting recent specializations,” Autor says. “For those who create a large-scale activity, there’s all the time going to be a chance for brand spanking new specialized knowledge that’s relevant for it. We thought that was exciting to see.”
The paper, “What Makes Recent Work Different from More Work?” is forthcoming within the Annual Review of Economics. The authors are Autor; Caroline Chin, a doctoral student in MIT’s Department of Economics; Anna M. Salomons, a professor at Tilburg University’s Department of Economics and Utrecht University’s School of Economics; and Bryan Seegmiller PhD ’22, an assistant professor at Northwestern University’s Kellogg School of Management.
And yes, learning about recent work, and the sorts of employees who obtain it, is likely to be relevant to the spread of artificial intelligence — although, in Autor’s estimation, it is just too soon to inform just how AI will affect the workplace.
“Individuals are really nervous that AI-based automation goes to erode specific tasks more rapidly,” Autor observes. “Eroding tasks isn’t the identical thing as eroding jobs, since many roles involve a number of tasks. But we’re all saying: Where is the brand new work going to come back from? It’s so vital, and we all know little about it. We don’t know what it should be, what it should seem like, and who will give you the chance to do it.”
“If everyone seems to be an authority, then nobody is an authority”
The 4 co-authors also collaborated on a previous major study of latest work, published in 2024, which found that about six out of 10 jobs within the U.S. from 1940 to 2018 were in recent specialties that had only developed broadly since 1940. The brand new study extends that line of research by looking more precisely at who fills the brand new lines of labor.
To do this, the researchers used U.S. Census Bureau data from 1940 through 1950, in addition to the Census Bureau’s American Community Survey (ACS) data from 2011 to 2023. In the primary case, because Census Bureau records grow to be wholly public after about 70 years, the students could examine individual-level data about occupations, salaries, and more, and will track the identical employees as they modified jobs between the 1940 and 1950 Census enumerations.
Through a collaborative research arrangement with the U.S. Census Bureau, the authors also gained secure access to person-level ACS records. These data allowed them to investigate the earnings, education, and other demographic characteristics of employees in recent occupational specialties — and to check them with employees in longstanding ones.
Recent work, Autor observes, is all the time tied to recent forms of experience. At first, this expertise is scarce; over time, it might grow to be more common. In any case, expertise is usually linked to recent types of technology.
“It requires mastering some capability,” Autor says. “What makes labor invaluable isn’t simply the power to do stuff, but specialized knowledge. And that always differentiates high-paid work from low-paid work.” Furthermore, he adds, “It must be scarce. If everyone seems to be an authority, then nobody is an authority.”
By examining the census data, the students found that back in 1950, about 7 percent of employees had jobs in forms of work that had emerged since 1930. More recently, about 18 percent of employees within the 2011-2023 period were in lines of labor introduced since 1970. (That happens to be roughly the identical portion of latest jobs per decade, although Autor doesn’t think it is a hard-and-fast trend.)
In these time periods, recent work has emerged more often in urban areas, with people under 30 benefitting greater than some other age category. Getting a job in a line of latest work seems to have a long-lasting effect: People employed in recent work in 1940 were 2.5 times as more likely to be in recent work in 1950, in comparison with the final population. College graduates were 2.9 percentage points more likely than highschool graduates to be engaged in recent work.
Recent work also has a wage premium, that’s, higher salaries on aggregate than in already-existing types of work. Yet because the study shows, that wage premium also fades over time, as the actual expertise in lots of forms of latest work becomes way more widely grasped.
“The scarcity value erodes,” Autor says. “It becomes common knowledge. It itself gets automated. Recent work gets old.”
In any case, Autor points out, driving a automobile was once a scarce form of experience. For that matter, so was with the ability to use word-processing programs similar to WordPerfect or Microsoft Word, well into the Nineteen Nineties. After some time, though, with the ability to handle word-processing tools became essentially the most elementary a part of using a pc.
Back to AI for a minute
Studying who gets recent jobs led the students to striking conclusions about how recent work is created. Examining county-level data from the World War II era, when the federal government was backing recent manufacturing in public-private partnerships throughout the U.S., the study shows that counties with recent factories had more recent work, and that 85 to 90 percent of latest work from 1940 to 1950 was technology-driven.
On this sense there was an excellent deal of demand-driven innovation on the time. Today, public discourse about innovation often focuses on the availability side, namely, the innovators and entrepreneurs attempting to create recent products. However the study shows that the demand side can significantly influence progressive activity.
“Technology isn’t like, ‘Eureka!’ where it just happens,” Autor says. “Innovation is a purposive activity. And innovation is cumulative. For those who get far enough, it should have its own momentum. But in case you don’t, it’ll never get there.”
Which brings us back to AI, the subject so many individuals are focused on in 2026. Will AI create good recent jobs, or will it take work away? Well, it likely depends how we implement it, Autor thinks. Consider the large health care sector, where there may very well be a number of forms of tech-driven recent work, if individuals are occupied with creating jobs.
“There are other ways we could use AI in health care,” Autor says. “One is simply to automate people’s jobs away. The opposite is to permit individuals with different levels of experience to do different tasks. I might say the latter is more socially useful. But it surely’s not clear that’s where the market will go.”
Alternatively, perhaps with government-driven demand in various forms, AI could get applied in ways in which find yourself boosting health care-sector productivity, creating recent jobs in consequence.
“Greater than half the dollars in health care within the U.S. are public dollars,” Autor observes. “We now have a number of leverage there, we are able to push things in that direction. There are other ways to make use of this.”
This research was supported, partially, 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 Instituut Gak.