Gender Confidence Gap in AI Knowledge Reveals Industry Blind Spots

A new Harvard Kennedy School survey of more than 5,400 U.S. adults exposes a striking confidence gap in how men and women assess their own AI literacy. Seventy-three percent of male respondents said they recognized OpenAI CEO Sam Altman, while 53 percent claimed familiarity with three names — Mira Murati, Susan Milner, and Craig Dalton. Only Murati is real; the other two were invented by researchers.

Women were less likely to claim recognition of the fake executives (40 percent) and of Murati (38 percent). Yet both genders answered factual AI questions incorrectly at similar rates. The difference: women selected "not sure" on 3.3 of 12 questions on average, versus 2.3 for men. This pattern suggests men default to certainty, women to caution.

Media amplification of the gap

Analysis of a year of AI news coverage found men named in headlines 14 times more often than women. Among quoted subject-matter experts, three-quarters were men — well above the roughly 29 percent female share of the AI talent pool. The result is a public discourse that rewards bluster over intellectual honesty.

Both genders ranked "concern" as their top emotional response to AI, but women reported higher fear and worry, while men reported more curiosity and enthusiasm. As Shorenstein Center director Laura Manley notes, the dominance of confident voices — whether predicting doom or abundance — drowns out the "not sure" perspective that rigorous engineering demands.

Why it matters for GPU / AI infrastructure: Procurement and architecture decisions rely on accurate risk assessment. Overconfidence can lead to under-provisioned clusters, skipped validation steps, or vendor lock-in driven by hype rather than benchmarks. Teams that normalize uncertainty — and diversify the voices shaping evaluation criteria — make more resilient infrastructure choices.