Is AI really pushing older workers out of jobs like programming and accounting?
The evidence suggests it can, though the effect concentrates in specific occupations rather than falling evenly across every desk job. In a June 2026 issue brief (26-13) by Geoffrey Sanzenbacher, the Center for Retirement Research at Boston College found that workers ages 55 and older in AI-exposed occupations became more likely to leave their jobs after the launch of ChatGPT. For computer programmers, the predicted increase in exits from work was over 25 percent, rising from 8.7 to 11.1 percent. Accountants and auditors saw a similar pattern, with predicted exits climbing about 22 percent, from 9.9 to 12.1 percent. By contrast, painters, at the low-exposure end of the scale, saw only about a 2 percent increase.
That last comparison matters because it overturns a common assumption. We tend to imagine that physically demanding jobs push people out first, but the Boston College research found that the workers most exposed to AI tend to be higher-paid and college-educated. If you are 55 or older in a knowledge role, the risk is not that you cannot keep up physically; it is that the analytical core of your work is being partly automated, and that pressure can accelerate a departure you did not plan.
Does this mean I should start planning my exit?
Not necessarily. The same research that flagged higher exit rates also points to a second path, and the reporting around it reflects genuine uncertainty about how this plays out person by person. When CNBC covered the Boston College brief on July 13, 2026, it noted that AI may either prompt some older workers to leave their jobs or make their roles more efficient. Those are two very different outcomes, and which one you experience depends heavily on how you position yourself now.
Older workers themselves are far from uniformly pessimistic. In a March 2026 AARP survey of 1,015 U.S. adults aged 50 and older who were in the labor force, about 24 percent described AI as a threat to their line of work, but 19 percent called it an opportunity, and a larger group, 37 percent, said it was both a threat and an opportunity. In other words, most people over 50 in the workforce are not treating AI as a simple death sentence for their careers. They are reading it as a mixed signal, which is the more accurate reading. The question is what you do with that ambiguity.
What actually keeps me employable when the tools are changing?
The short answer: move toward the parts of your job that require judgment, accountability, and relationships, and become fluent in the tools handling the rest. Automation tends to absorb the repeatable, well-specified portions of work first. It is far slower to replace the person who decides which output to trust, who signs off on a financial statement, who reviews code for security and business risk, or who translates a client's messy real-world problem into something the tools can even address.
For a programmer, that might mean deliberately spending less energy defending your speed at writing routine functions and more energy on architecture, code review, debugging systems that AI-generated code introduces, and mentoring. For an accountant or auditor, it means leaning into interpretation, exception handling, controls, and client advisory work, the judgment layer that sits above the calculations. In both cases the goal is the same: make sure your value is described in terms of decisions and outcomes, not tasks a tool can now perform in seconds.
Just as important is direct fluency with the systems reshaping your field. The person who can say "I use these tools daily and here is how I verify their output" is in a stronger position than either the resister who avoids them or the true believer who trusts them blindly. Learn the specific platforms your industry is adopting. You do not need to become an engineer; you need to be the experienced professional who knows where the tools are reliable and where they quietly fail, because that knowledge is exactly what younger, faster adopters often lack.
How do I make sure the choice to leave stays mine?
Protect your options by staying visible and keeping your record current, so that if change comes, you are choosing your next step rather than reacting to a surprise. One risk the Boston College findings hint at is that higher-paid, experienced workers can drift toward the exit through a series of small discouragements: a reorganization, a role that shrinks, a sense that the ground is shifting. The antidote is to stay demonstrably useful and demonstrably present.
Start by documenting outcomes in concrete terms. Instead of listing responsibilities, capture what changed because of your work: errors caught, hours saved, systems stabilized, junior colleagues brought up to speed. Keep a running record so your resume and profile never lag months behind reality. Next, rebuild external relationships if they have gone quiet. Experienced workers often let their networks atrophy because they have been secure for years; that security is exactly what AI-driven change can erode. Reconnect with former colleagues, participate in professional communities, and let people see that you are engaged with where the field is heading.
Finally, treat learning as ongoing rather than a one-time scramble. You do not need to reinvent yourself. You need to add the specific, current skills that keep your judgment relevant to how the work is actually done now. The workers who stay employable through this transition are rarely the ones who out-typed the machine. They are the ones who became the trusted human in the loop, and who made that value legible to their employer and to the wider market.