This is the second part of a two-part program MIT News article examining the creation of new jobs in the United States since 1940, based on new research by Ford economics professor David Autor. The first part is available here.
Ever since the Luddites destroyed power looms, it has been clear that new technologies can wipe out jobs. But technical innovations also create new jobs: think of a computer programmer or someone installing solar panels on a roof.
Overall, is technology replacing more jobs than it creates? What is the net balance between these two things? So far this has not been measured. But a new research project led by MIT economist David Autor has developed an answer, at least for U.S. history since 1940.
The study uses new methods to examine how many jobs have been lost to machine automation and how many have been generated through “augment,” in which technology creates new tasks. In fact, the study reveals that, particularly since 1980, technology has replaced more jobs in the United States than it has created.
“There appears to have been a faster rate of automation and a slower rate of increase over the last four decades, from 1980 to today, than in the previous four decades,” says Autor, co- author of a recently published article. detailing the results.
However, this discovery constitutes only one of the advances of the study. The researchers also developed an entirely new method for studying the question, based on an analysis of tens of thousands of U.S. Census job categories against a comprehensive review of the text of U.S. patents over the last century. This allowed them, for the first time, to quantify the effects of technology on job loss and creation.
Previously, researchers had largely succeeded in quantifying job losses caused by new technologies, not job gains.
“I feel like a paleontologist who was looking for dinosaur bones that we thought must exist, but that we hadn't been able to find until now,” Autor says. “I think this research opens up things that we suspected were true, but we didn't have direct evidence before this study.”
The paper, “New frontiers: origins and content of new works, 1940-2018”, appears in the Quarterly Journal of Economics. Co-authors are Autor, Ford Professor of Economics; Caroline Chin, doctoral student in economics at MIT; Anna Salomons, professor at the Faculty of Economics at Utrecht University; and Bryan Seegmiller SM '20, PhD '22, assistant professor at the Kellogg School at Northwestern University.
Automation versus augmentation
The study finds that overall, about 60 percent of jobs in the United States represent new types of work, created since 1940. A century ago, that computer programmer may have worked on a farm.
To determine this, Autor and his colleagues combed through about 35,000 job categories listed in U.S. Census Bureau reports, tracking their emergence over time. They also used natural language processing tools to analyze the text of every U.S. patent filed since 1920. The research examined how words were “embedded” in census and patent documents to discover related passages of text. This allowed them to determine the links between new technologies and their effects on employment.
“You can think of automation as a machine that takes data from a job and does it for the worker,” says Autor. “We think of augmentation as technology that increases the variety of things people can do, the quality of things people can do, or their productivity.”
Between about 1940 and 1980, for example, jobs like elevator operator and composer tended to be automated. But at the same time, more workers have filled positions such as shipping and receiving clerks, buyers and department managers, as well as civil and aeronautical engineers, where technology has created a need for more employees.
From 1980 to 2018, the ranks of cabinetmakers and machinists, among others, were reduced by automation, while, for example, industrial engineers and operations and systems researchers and analysts experienced growth.
Ultimately, research suggests that the negative effects of automation on employment were more than twice as large over the period 1980-2018 as over the period 1940-1980. There was a more modest and positive change in the effect of the increase on employment between 1980 and 2018, compared to the period 1940-1980.
“There is no law, these things must be balanced one for one, although there has been no period where we have not also created new works,” Autor observes.
What will the AI do?
Research also reveals many nuances in this process, as automation and augmentation often occur within the same industries. It's not just that technology is decimating the ranks of farmers while creating air traffic controllers. Within the same large manufacturing company, for example, there may be fewer machinists but more systems analysts.
Similarly, over the past 40 years, technological trends have exacerbated the wage gap in the United States, with highly skilled professionals more likely to work in new fields, themselves split between high-paying jobs and jobs low income.
“The new work is divided,” says Autor. “As old works were erased in the middle, new works developed on either side.”
As the study also shows, technology is not the only source of new work. Demographic changes are also driving growth in many sectors of the service industries. Intriguingly, the new research also suggests that large-scale consumer demand also drives technological innovation. Inventions are not just the result of brilliant people thinking outside the box, but they respond to clear societal needs.
The 80 years of data also suggest that future innovation pathways and employment implications are difficult to predict. Consider the possible uses of AI in the workplace.
“AI is really different,” says Autor. “It can replace highly skilled expertise, but can complement decision-making tasks. I think we're in a time where we have this new tool and we don't know what it's for. New technologies have strengths and weaknesses and it takes time to understand them. GPS was invented for military purposes, and it took decades for it to be integrated into smartphones.
He adds: “We hope that our research approach will allow us to say more in the future. »
As Autor acknowledges, the research team's methods can still be refined. For now, he believes the research opens up new areas of study.
“The missing link was documenting and quantifying the extent to which technology improves people's employment,” says Autor. “All previous measurements have simply shown automation and its effects on worker displacement. We were surprised to be able to identify, classify and quantify the increase. That in itself, to me, is pretty fundamental.
Research support was provided, in part, by The Carnegie Corporation; Google; Gak Institute; the MIT Futures Works Task Force; Schmidt Futures Markets; the Smith Richardson Foundation; and the Washington Center for Equitable Growth.