A study reveals that AI is shortening the careers of older workers.

      The discussion surrounding AI and employment has primarily centered on graduates. However, new research indicates that attention should also be directed towards individuals in their late fifties, according to CNBC.

      Employees aged 55 and older in professions susceptible to AI are now leaving their jobs at increased rates since the launch of ChatGPT. This observation comes from Geoffrey Sanzenbacher at Boston College's Center for Retirement Research.

      The key point is where these workers end up. The rise in exits manifests as unemployment rather than retirement, indicating that these individuals are out of work and actively seeking employment.

      How the research is conducted

      The study combines data from the US Current Population Survey with an AI exposure index developed by Tufts University's Digital Planet Initiative. This index evaluates how effectively AI can perform various job tasks, rather than merely forecasting if a job is at risk.

      The analysis reveals significant insights. Occupations such as web designers, web developers, database architects, computer programmers, and data scientists rank highest in terms of AI exposure, while jobs like mining operators, orderlies, and painters rank lowest.

      The comparison focuses on the timeframe before and after ChatGPT's November 2022 launch, which is important because jobs with varying levels of AI exposure already demonstrated different trends.

      The turnaround

      What makes this result noteworthy is the shift in patterns. Before ChatGPT, older workers in AI-exposed positions were considerably less inclined to leave their jobs than their counterparts in less exposed roles. This trend was anticipated, as desk jobs tend to be less physically demanding, better compensated, and more conducive to longer tenures.

      Since the launch of ChatGPT, however, this advantage has significantly eroded. The rate at which exposed workers are exiting their positions has risen enough to negate the benefits they previously enjoyed.

      The figures, job by job

      The model estimates how much the exit rate for each occupation has changed. Painters, positioned at the lower end of AI exposure, experienced an exit rate increase of around 2%. Conversely, computer programmers saw their exit rate rise by more than 25%, increasing from 8.7% to 11.1%. Similarly, accountants and auditors experienced an increase of about 22%, from 9.9% to 12.1%.

      As a result, the impact is most pronounced among those who are educated and well-compensated. Workers in exposed roles are more likely to be white, over twice as likely to have a degree, and earn approximately $1,410 weekly, compared to $869 for those in less exposed roles. This upends the conventional narrative about automation, which typically begins on the factory floor.

      Three mechanisms at play

      Sanzenbacher outlines three ways this is happening. First, automation can directly replace workers, leading them to unemployment or to exit the labor force altogether. Second, the pressure to integrate new technologies can encourage some workers to leave instead of learning a new system late in their careers— a trend previously noted during the widespread adoption of personal computers.

      The third pathway is more positive. Increased productivity could boost wages and allow workers to engage in more fulfilling tasks, potentially extending their careers rather than cutting them short.

      The gap in technology adoption indicates older workers may be hesitant. Roughly 18% of those aged 50 to 64 utilize generative AI, significantly lower than the usage among workers in their thirties and forties, with children adopting it at three times the rate of adults.

      Attitudes reflect this divergence. An AARP survey found that 18% of individuals over 55 viewed AI solely as an opportunity, while 28% regarded it purely as a threat.

      What the research does not address

      The overall effects, while significant, are modest in absolute terms, and the author is cautious in presenting these findings. He emphasizes that this is an initial analysis, urging for careful interpretation.

      Two confounding factors might influence the results in unpredictable ways. Cuts to government research funding might have significantly affected AI-exposed roles, thereby exaggerating the observed impact, while the current boom in AI startups could bolster demand for data scientists, masking the effects that may arise when this surge subsides.

      Moreover, jobs exposed to AI remain comparatively safer overall. Even with the increases, programmers leave their jobs less frequently than painters because the former role is less physically demanding and better compensated.

      The accurate interpretation indicates a narrowing gap, not a collapse. AI is diminishing the disparity in career longevity between high-paying and low-paying roles from the top down.

      Why this is more significant than it seems

      Losing a job at age 58 carries different implications than losing one at 28. Involuntary job loss later in life often leads to unplanned early retirement, accompanied by an unanticipated pension.

      The policy implications are direct. Governments are advocating for longer working lives to strengthen retirement systems, while technology may be subtly pushing people towards earlier exits.

      This situation also complicates the industry's assurances, including Sam Altman’s assertion that