"AI will affect 40% of jobs". We've read that sentence a hundred times, and it is almost useless because "affected" hides two opposite futures. Picture two people who are both told they are "highly exposed to AI:. One is a lawyer. AI becomes her copilot. It drafts the first version, she works faster, and she earns more. The other is a childcare administration worker. AI does not assist her. It absorbs the task. Her role begins to disappear. Same label. Opposite fate.

This study suggests the divide runs clearly along gender lines. By linking O*NET job data, US Census pay figures, and two measures of AI exposure, we asked a sharper question than "men's jobs or women's jobs?". We asked where on the pay ladder the exposure lands and fount that:

• (1) In male dominated work, exposure clusters near the top, in highly paid and highly skilled roles where AI is more likely to assist.
• (2) In female dominated work, exposure is spread across the pay ladder. The lowest paid roles with the least power are hit just as hard, and AI may be more likely to replace workers than assist them.

There is another twist. The AI dominating today's headlines, large language models, skews heavily toward women's jobs, with an exposure score of about 74 compared with 24 for men's jobs. Robotics skews toward men's jobs. So the technology organisations may be rushing to adopt is aimed most directly at roles with the least bargaining power and the least access to reskilling.

Publications

• When AI Enters the Workplace, Who Faces Greater Risks? A Gendered Analysis. AIES 2026. PDF

Code and data

Fill in the form to receive the link to the datasets:

N.B.: If you do not receive the instruction message within a few hours, please check your junk/spam e-mail folder just in case the email was moved there.