AI hiring systems show gender bias in pay despite preferring women candidates

A study found that when AI language models like ChatGPT evaluate job candidates, they recommend hiring women more often than men and rate them as more qualified — but still suggest paying them less. This means the same AI system that appears to favor women on paper actually perpetuates pay inequality in practice.

Companies are beginning to use AI systems to screen resumes and make hiring recommendations at scale. This research reveals that AI doesn't magically remove human bias — it can shuffle bias around (preferring women on paper while underpaying them) in ways that are harder to spot and challenge than old-fashioned discrimination. If AI hiring tools are deployed widely without understanding these patterns, they could lock unequal pay into automated systems that nobody thinks to question because the surface-level metrics look fair.