A detailed study by MIT economist David Autor and research team suggests that although new technology innovations have created new jobs since 1940, they have also replaced more jobs than created, particularly since 1980s. The study analyzed tens of thousands of U.S. census job categories in combination with an examination of the text of U.S. patents from the last century, to quantify technology’s effect on job loss and job creation.
Until the present study, scholars had primarily been able to quantify job losses due to technology, but not job gains. Autor compared this to a paleontologist searching for previously elusive dinosaur bones, saying that his study breaks new ground and provides direct proof of things researchers suspected were true.
The research found that approximately 60% of jobs in the U.S. represent new types of work created since 1940. To reach this conclusion, the researchers combed through about 35,000 job categories from U.S. Census Bureau reports, tracked their emergence over time, and utilized natural language processing tools to assess the text of every U.S. patent filed since 1920. This meticulous study enabled them to identify the connections between new technologies and their influence on employment.
Furthermore, the study made distinctions between “automation” and “augmentation”. Automation refers to machines taking over a worker’s job, while augmentation describes technology that enhances people’s abilities or productivity. For instance, during 1940-1980, jobs like elevator operators and typesetters were automated, but more professions such as engineers, clerks, buyers, and department heads were created due to technology. However, from 1980 through 2018, automation led to job loss in professions like cabinetmakers and machinists, while jobs like industrial engineers, systems researchers, and analysts experienced growth.
The study indicates that the negative effects of automation on employment were more than twice as dominant during the period from 1980-2018, compared to 1940-1980. However, the effect of augmentation on employment also saw a modest improvement from 1980-2018, as compared to the previous four decades, indicating that new work is continually being created.
Looking ahead, Autor points out the complexity of predicting future pathways for innovation and its effects on employment. The potential use of AI in workplaces might prove to be a game-changer, as it could both replace high-skill expertise and complement decision-making tasks. Hence, he emphasizes the need for more extensive research to explore AI’s impact on employment dynamics. While recognizing the need for further refinement of their research methods, Autor is hopeful that their innovative approach will continue to bring new insights into this intricate area.
