A European perspective on how artificial intelligence is changing the fortunes of older workers is offered by a study by Pablo Casas and Concepción Román in the journal Empirica. The authors drew on the extensive SHARE panel, which tracks population health and aging and covers twenty-six European countries, including Czechia, Poland, Germany, and Spain. They matched it with data on advances in AI and the extent to which the technology affects individual occupations.
At first glance, their findings seem reassuring. In Europe, the likelihood of early exit from the workforce is falling among employees whose occupations are already heavily affected by artificial intelligence, as well as in those where its use is expected to expand further. But there is an important condition. This protective effect applies only to people with a university education.
Education level matters
Education acts as a dividing line. Those who know how to work with the technology and can add value where machines cannot yet reach remain in employment longer. Those without these skills find themselves at risk.
Casas and Román highlight this contradiction throughout their work. In a previous study from 2023, they found that automation, by contrast, pushes older employees out of work, with the impact varying by gender, education, and type of employment. They themselves draw attention to an uncomfortable tension. Governments are trying to delay retirement and raise the retirement age, while the labor market is pushing experienced people out earlier. Their recommendation is therefore clear: increases in the retirement age must be accompanied by retraining and support for those at risk of seeing their careers end prematurely.
British companies are planning layoffs, with office workers bearing the brunt
The United Kingdom offers a more concrete picture. The CIPD, an association of HR professionals, surveyed more than two thousand British employers as part of its regular poll. One in six, or seventeen percent, expects AI to reduce its headcount within a year. Among large private companies, the figure is more than a quarter.
What is interesting is who will be affected. Nearly two-thirds of those companies said that administrative, clerical, and professional roles were the most at risk. In other words, precisely the kind of office work also highlighted by US research. The pressure is enormous. In the first two months of 2026, British companies filed collective redundancy notices covering more than fifty-six thousand jobs, nine percent more than in the same period of the previous year.
Careers that are being cut short
The most striking figures came from US research by Geoffrey Sanzenbacher of the Center for Retirement Research at Boston College. He compared employment data from before and after the launch of ChatGPT in November 2022. He assigned each occupation an artificial intelligence exposure index developed by Tufts University, which measures how well the technology can perform the tasks involved in a given job.
The result turned the conventional view of automation on its head. Before the arrival of ChatGPT, older employees in occupations threatened by AI left their jobs less often than their peers in other fields. That made sense. Office work is less physically demanding, better paid, and easier to continue doing for longer. But that advantage has now virtually disappeared.
The model calculated how exit rates shifted across individual occupations. For house painters at the bottom of the ranking, the rate rose by roughly two percent. For computer programmers, it jumped by more than a quarter, from 8.7 to 11.1 percent. Accountants and auditors saw an increase of around twenty-two percent.
It is not retirement, but unemployment
The crucial detail is where these people are going. They are not settling into a comfortable retirement. Statistics classify them as unemployed—out of work but still looking. Losing a job at fifty-eight is not the same as losing one at twenty-eight. An involuntary exit near the end of a career often turns into an early retirement that no one chose.
The difference in technology use offers a clue as to the path older employees are taking. Only around eighteen percent of people aged fifty to sixty-four use generative AI, significantly fewer than those in their thirties and forties. Eighteen percent of people over fifty-five see AI purely as an opportunity, while twenty-eight percent see it only as a threat.
“I don't believe in it. It's a bubble”
Jennifer Kerns left in March after more than thirty years at technology companies. Her most recent role was as a program manager at Microsoft-owned GitHub. She was sixty and left earlier than she had originally intended. One of the main reasons was artificial intelligence, which had become the only topic at the company. “That was it for me,” she told Fortune. “I don't believe in it. I think it's a bubble that's going to burst.” She added that it was not about being afraid of the technology. “It offends me.”
Steve McConnell, a retirement planning adviser and founder of Rain Dog Financial, has been observing an increase in early retirements since the pandemic. According to him, work at technology companies changes exceptionally frequently. Over the past thirty years, desktop computers, the internet, mobile phones, the cloud, and now AI have all arrived. “One common trait among people in the industry is that the learning curve for new technology can consume a great deal of effort,” McConnell said.
The loss of experienced workers could also harm AI itself
Something else worries McConnell. The departure of an entire cohort of experienced engineers could be bad news for the future of artificial intelligence itself. Companies are losing the judgment and knowledge accumulated over years of practice, precisely when the technology is still in its infancy and needs safeguards. “I'm concerned about the loss of judgment if we lose a generation of experienced engineers while AI is still in its infancy,” he added.
Kerns takes a similar view. She points out that companies such as IBM want to triple the number of entry-level positions, but without experienced colleagues, there will be no one to guide these newcomers. Mentorship—the kind of resource that cannot be acquired overnight—will be missing.
