Harvard study finds men more confident about AI—even when the executives are fictional

A Harvard study released October 14 found a revealing gap in Americans’ AI confidence: men were more likely than women to claim they recognized executives who do not exist. The survey of more than 5,400 Americans also found men answered more factual questions correctly—but men and women gave wrong answers at roughly the same rate.

The fictional names were Susan Milner and Craig Dalton. According to WIRED, 53% of male respondents claimed familiarity with those names, as well as real AI executive Mira Murati. Among women, 40% claimed to recognize the fictional executives, while 38% recognized Murati. The invented names drew recognition claims from substantial shares of both groups, not just men.

For comparison, 73% of male respondents said they knew OpenAI CEO Sam Altman. These questions measured claimed familiarity, rather than asking respondents to demonstrate what they knew about each person. A claim to recognize a fictional name should not, by itself, be treated as proof of deliberate deception.

The factual questions offer a different measure. Men reported using AI more frequently and gave more correct answers on average. Yet the genders had similar rates of incorrect answers. The difference lay in uncertainty: women answered “not sure” on an average of 3.3 questions out of 12, compared with 2.3 for men.

That distinction keeps the finding from being a simple ranking of who knows more. Correct answers, wrong answers and acknowledged uncertainty tell different stories. Laura Manley, executive director of Harvard Kennedy School’s Shorenstein Center, told WIRED that men were more optimistic and confident about AI, sometimes to the point of overconfidence.

It just drowns out all of the voices that should be saying ‘not sure.’

Laura Manley, speaking to WIRED about confident predictions of AI doom or abundance

The study also examined a year of AI news coverage. Men were named in headlines more than 14 times as often as women. Because top AI executives skew heavily male, researchers narrowed their examination to people quoted as subject experts—a way to look beyond the leaders whose positions already attract attention.

WIRED puts men’s share of quoted experts at three-quarters, compared with an AI talent pool that the report describes as roughly 29% women. Mother Jones gives a different measure: men appeared 4.3 times as often as women among well over 100,000 expert citations in domestic AI news from July 2025 through June 2026.

Those figures are not interchangeable, and the precise expert share remains unresolved between the two accounts. Both describe coverage dominated by men. Manley’s response was to suggest giving more women a platform as experts, including room to express uncertainty rather than requiring a confident prediction about AI’s future.

The survey’s differences extended beyond knowledge questions. Women were less likely than men to trust AI tools and information about them. They were also more likely to report insufficient access to information and training about how AI tools were used in their workplaces. That is a reported access gap, alongside the confidence gap.

Concern was the top emotional response to AI for both men and women. Women reported more concern, fear and worry; men rated positive feelings, including curiosity, enthusiasm and excitement, higher. The contrast was therefore not between a wholly enthusiastic group and a wholly fearful one: concern led in both.

Men and women also agreed that men’s needs and perspectives were better represented in AI development and AI-generated content. The researchers said their gender analysis centered on men and women because most respondents identified with those groups, while explicitly recognizing that not everyone identifies within a gender binary.