The AI Boom Is Putting The Career Ladder Into Question. It Is Also Leading Older Workers To Early Retirement
Artificial intelligence may be shortening the careers of older Americans, particularly those working in well-paid, white-collar occupations that previously allowed employees to remain on the job longer, according to a new study.
In a study titled “Are the Careers of Older Workers Being Cut Short by AI?,” the Center for Retirement Research at Boston College found that workers age 55 and older in occupations with greater exposure to AI became more likely to leave employment after ChatGPT was released in November 2022.
The increase was driven partly by workers moving into unemployment rather than voluntarily retiring, suggesting that some late-career employees may be losing their jobs before they are financially ready to stop working.
“The analysis here suggests that older workers in jobs with greater exposure to AI have seen relatively large increases in job exit,” researcher Geoffrey T. Sanzenbacher wrote. Sanzenbacher, a Boston College economics professor and research fellow at the Center for Retirement Research, analyzed data from the Current Population Survey and combined it with an AI Exposure Index developed by the Digital Planet Initiative at Tufts University.
The index measures how effectively AI can perform the tasks associated with different occupations. It compared employment outcomes before and after ChatGPT’s launch, focusing on whether workers remained employed one year after they were first observed in the survey. Researchers examined transitions into unemployment, retirement or departure from the labor force for another reason.
The occupations with the greatest AI exposure were largely technology and data jobs. Web and digital interface designers received the highest possible exposure score of 100, followed by web developers, database architects, computer programmers, and data scientists. Jobs with the lowest exposure included mining workers, orderlies, painting/spraying workers, and fiberglass fabricators.
Before ChatGPT, older workers in highly exposed occupations were generally less likely to leave employment than people in jobs with lower AI exposure. Those careers often required more education, offered higher wages and involved less physical labor, making them easier to continue later in life.
After ChatGPT’s debut, however, much of that advantage disappeared. “The types of jobs exposed to AI used to have a relative advantage with respect to career longevity,” the study said.
Computer programmers showed one of the most dramatic changes. The predicted rate at which older programmers moved out of employment increased from 8.7% before ChatGPT to 11.1% afterward, a jump of more than 25%. By comparison, the projected exit rate among painters increased only about 2%, from 13.5% to 13.7%.
Accountants and auditors also experienced an increase, with their predicted transition out of work rising from 9.9% to 12.1%. For management analysts, the rate increased from 10.5% to 12.2%. The findings do not prove that AI directly caused older workers to lose their jobs. Sanzenbacher warned that the results are preliminary and could be influenced by other economic developments, including federal research cuts or rapid hiring by AI startups.
AI could also help some employees remain productive longer by automating repetitive duties and allowing them to focus on more valuable or engaging work. The outcome may depend on whether workers receive training and are able or willing to adapt to the technology.
Still, the research raises concerns for policymakers who increasingly view longer careers as part of the solution to Social Security‘s financial challenges. Workers who unexpectedly lose their jobs late in life may have fewer opportunities to find comparable employment, rebuild their savings or delay claiming retirement benefits. The report warned that increased job exits could leave affected workers with “shorter careers if they cannot find new work.”
