AI Is Pushing Older Workers Out of Jobs Earlier Than Expected, New Research Finds
A Boston College study shows workers aged 55 and over in AI-exposed roles are leaving the workforce sooner, and the jobs most at risk are not the ones most people expect.
Key points
- Workers aged 55 and over in AI-exposed industries showed a statistically significant rise in job exits after the launch of ChatGPT in late 2022, according to a new Boston College study.
- The five jobs with the highest AI exposure include web and digital interface designers, web developers, and database architects, all higher-paid white-collar roles.
- 42% of 1,504 workers surveyed by Monster in December 2024 said they do not use AI tools at all.
- AARP and LinkedIn research found 49.4% of workers over 50 hold roles partly insulated from AI disruption, compared with 42.2% of younger workers.
- Social Security's main trust fund may run out in late 2032, which could push high-income workers to stay employed longer, just as AI threatens their roles.
AI was supposed to be a young person's problem. Older workers, the thinking went, were safely out of reach, either winding down careers in physical jobs or coasting on decades of hard-won expertise.
New research says that picture is wrong.
A paper from the Center for Retirement Research at Boston College found that workers aged 55 and over in jobs with high AI exposure, meaning roles where AI tools can already perform many of the tasks, are leaving work more frequently than before. Exits are split equally between involuntary job losses and workers choosing to quit rather than adapt.
"It's a statistically significant effect," said Geoffrey Sanzenbacher, the study's author and an economics professor at Boston College. "For some occupations, it can be quite large."
The research draws on U.S. Labour survey data and AI exposure scores from Tufts University's Digital Planet initiative, a programme tracking how digital technology reshapes different jobs. The turning point in the data is November 2022, when OpenAI released ChatGPT and made AI tools suddenly visible to employers at scale.
Before ChatGPT, older workers in AI-exposed roles were actually less likely to leave their jobs. After it launched, that flipped.
What does this mean for people approaching retirement?
It creates a double squeeze for high earners. The jobs most exposed to AI include web and digital interface designers, web developers, database architects, computer programmers, and data scientists: exactly the well-paid office roles that white-collar workers spend decades building towards. Those same workers face the biggest risk from any Social Security reform, because the programme's trust fund is projected to run out in late 2032 and trimming payments to higher earners is among the most discussed fixes.
"There's a high probability that higher-income people see a bigger benefit cut than lower-income people from whatever happens with Social Security next," Sanzenbacher told CNBC. "These are the very people who therefore need to work longer."
The workers most likely to need to keep working could be the first to find their roles disrupted.
At the other end sit excavating workers, roof bolters, painting and spraying workers, and fiberglass fabricators: hands-on physical roles AI cannot easily replicate. These have historically meant shorter careers because they are hard on the body. AI may close that gap from the other direction, shortening high-earners' careers rather than lengthening lower-income ones.
The picture is not purely bleak. Sanzenbacher notes AI could also let some people work longer by handling repetitive tasks and freeing time for higher-level thinking. Separate AARP and LinkedIn research found experienced workers are more likely to hold roles requiring collaboration, judgment, and leadership, skills AI struggles to replace.
For anyone over 50 who has not yet touched an AI tool, Monster career expert Vicki Salemi, writing by email to CNBC, suggests a two-part approach: learn the AI tools your employer already pays for, then make sure your CV and conversations highlight communication, problem-solving, and relationship-building skills that no chatbot can convincingly replicate.
Watch the Social Security timeline. If trust-fund projections hold and benefit cuts land on high earners around 2032, the financial pressure to keep working will collide directly with the AI disruption this research is already measuring. That intersection is the number to keep an eye on.
