A new study finds that while AI is theoretically capable of automating many professional tasks, real-world adoption remains limited and has not yet caused widespread job losses — though early signs suggest hiring in highly exposed fields may be slowing. (Source: Image by RR)

Computer Programmers Rank Among Most Exposed Occupations

A new study examining artificial intelligence and employment has introduced a measurement called “observed exposure,” a framework designed to track how AI may influence jobs in the real economy. The metric combines theoretical AI capabilities with real-world usage data to identify which occupations are most vulnerable to automation. By focusing on tasks that AI systems are actually performing — particularly automated work-related tasks — researchers hope to better detect early signs of labor market disruption.

The analysis, as noted in an article at anthropic.com, reveals that AI adoption remains far below its theoretical potential. While large language models could theoretically assist with many professional tasks, actual usage currently covers only a fraction of that capacity. For instance, even in highly AI-relevant sectors like computer and mathematical occupations, current AI usage accounts for roughly one-third of potential tasks that AI could theoretically perform.

Researchers also found that occupations with higher AI exposure tend to have weaker projected job growth through 2034, according to forecasts from the U.S. Bureau of Labor Statistics. Workers in these higher-exposure professions are more likely to be older, female, more educated, and higher paid than those in less exposed roles. Jobs such as computer programmers, customer service representatives, and data entry clerks rank among the most exposed categories.

Despite these indicators, the study found little evidence so far that AI has significantly increased unemployment. Since the release of ChatGPT in late 2022, unemployment rates for highly exposed workers have remained statistically similar to those in less exposed occupations. However, there are early signs that hiring into exposed professions — particularly among younger workers aged 22 to 25 — may be slowing, suggesting that AI’s first labor-market effects could emerge through reduced hiring rather than immediate layoffs.

read more at anthropic.com