Story Content

Anthropic has published pioneering research on the labor market implications of AI using a new "observed exposure" measure that is both preliminary and based on real-world exposure to the Claude model.

The research finds computer programmers to be the most affected, with 75% of tasks performed by AI. This is followed by customer service representatives, whose primary tasks can increasingly be part of an AI pipeline and API. The third most exposed are data entry keyers (67% exposed), as AI is efficient at analyzing documents and entering information.

Also exposed are financial and investment analysts, market research analysts, medical record keepers, and software QA testers. The study finds AI is impacting information and knowledge workers in white-collar occupations more than physical manual workers such as cooks, mechanics, or construction workers, who have exceedingly low exposure levels.

Although the potential for using AI is still much greater (computer and math jobs can be 94% exposed), actual use is only a small share. The results indicate employment growth is likely to slow in these highly affected jobs, particularly among younger workers.