Gen Z women losing ground to men in entry-level jobs—but AI isn't the main cause

The Canaries Dashboard has emerged as one of the country's most closely watched labor-market trackers since its debut, establishing a clear pattern: workers ages 22–25 in AI-exposed occupations like software development and customer service have seen employment decline sharply since ChatGPT's late-2022 debut, even as overall U.S. employment growth remained healthy. That divergence has not reversed but deepened, growing by roughly half a percentage point per month.

The latest update, released for the first time with gender disaggregation, reveals a paradox that challenges conventional wisdom about AI and gender equity. Prior research from the International Labour Organization and other sources has suggested women's jobs are nearly twice as likely to be exposed to generative AI, and separate estimates have placed women at three times the automation risk of men. Yet the Canaries data—which tracks actual, realized employment outcomes month by month—shows that if AI exposure were the primary driver of the gender gap, the gap should widen sharply as exposure rises. Instead, it remains roughly the same whether a job is barely touched by AI or squarely in its path.

The researchers' conclusion reframes the problem: the real issue is not that AI treats women differently within occupations, but that women are overrepresented in AI-exposed roles as a byproduct of occupational sorting that predates generative AI by decades. Brynjolfsson and Richardson's core argument has been that AI is disrupting tasks before it disrupts jobs—automating the mechanical work (summarizing, formatting, scheduling) typically handed to the newest employees—and early-career roles sit disproportionately in that contracting group. Women's concentration in such roles is a labor-market structure issue, not an AI-specific one, pointing to questions around education mix, industry concentration, hours worked, and return-to-office effects that further research is expected to address.