AI Advice Works Differently for Women in Business

In a study that feels ripped from a business school case — if the case starred AI — researchers from the National University of Singapore and the University of Pennsylvania tested whether AI pricing advice helps or hurts in real markets. What they found surprised them: AI advice only worked when the sellers were all women.

They set up 91 mini-markets with three sellers each, running 30 rounds of pricing games. In markets made up entirely of women, AI-recommended prices pushed final prices up by 29% and profits up by 39%. In male-only or mixed-gender markets, AI had no meaningful effect. The researchers dug deeper and found that in women-only markets, sellers who saw profits rise started trusting the AI more — and followed its advice more closely. In other markets, the opposite happened: when profits went up, sellers trusted the AI less and ignored it more.

The team calls this “composition-specific dynamics” — basically, AI advice changes behavior only when the group using it is mostly or all women. The study flips the common idea that women distrust AI more than men. Instead, it shows trust depends on the group’s gender mix and how the AI actually performs. The authors caution that algorithms aren’t neutral; they reflect how humans respond to them.

They suggest companies and regulators shouldn’t just tweak algorithms — they should study who’s using AI and how. If your team is mostly women, AI pricing tools might be worth trying. If it’s mixed or mostly men, results may vary. The takeaway isn’t that AI is better for women — it’s that AI’s success depends on who’s in the room.