Harvard: men claim to know fake AI execs Susan Milner, Craig Dalton

The report, published on Wednesday, looks past the headline finding of male overconfidence to the mechanics behind it. Men and women got AI factual questions wrong at roughly the same rate, but women were far more willing to say they did not know — 3.3 out of 12 questions on average, against 2.3 for men. That difference in admitted uncertainty, not in actual accuracy, produced the confidence gap.

The survey then traces that gap into public conversation. Over a year of AI coverage, men appeared in headlines more than 14 times as often as women. Even when researchers narrowed the sample to people quoted as subject experts, three-quarters were men — a share above the roughly 29 percent of the AI talent pool who are women, suggesting the imbalance in coverage is not simply a reflection of who works in the field.

On emotion, the two groups were closer than on knowledge. Both men and women ranked concern as their top response to AI. Women reported more fear and worry; men rated every positive feeling, including curiosity, enthusiasm, and excitement, higher than women did. The report's concluding note, from Shorenstein Center executive director Laura Manley, is that confidence in either doom or abundance crowds out the voices willing to say they are not sure.