Men Can’t Stop Faking How Much They Know About AI

Women are now less likely to be perceived as AI experts because they are more willing to admit uncertainty.

A recent survey of over 5,400 Americans by Harvard researchers revealed a significant gender disparity in AI knowledge and confidence. 73 percent of male respondents claimed to know Sam Altman, and 53 percent recognized fictional AI executives Murati, Milner, and Dalton. In contrast, only 40 percent of women recognized the fictional executives and 38 percent recognized Murati, who is a real former OpenAI CTO. While men used AI more frequently and answered factual questions more confidently, both genders answered incorrectly at similar rates. The difference lies in women more often admitting uncertainty, averaging 3.3 uncertain answers out of 12 compared to men's 2.3. Media coverage also reflects this, with men featured in AI headlines 14 times more often than women, and three-quarters of quoted AI experts being male, despite women comprising roughly 29 percent of the AI talent pool. While both genders expressed concern about AI, women reported higher levels of fear and worry, whereas men reported higher levels of curiosity and excitement.

The survey highlights a concerning trend where overconfidence in AI knowledge among men, to the point of recognizing fictional figures, contrasts with women's greater willingness to admit uncertainty. This gendered difference in expressing confidence, rather than actual knowledge, could have tangible career repercussions in a field that increasingly values perceived expertise. When media outlets and industry platforms amplify voices based on asserted certainty over demonstrated knowledge, women may be further marginalized.

The overrepresentation of men in AI headlines and as quoted experts, even when compared to the actual gender distribution in the field, suggests a systemic bias in how AI expertise is recognized and disseminated. This dynamic risks creating a feedback loop where perceived male dominance in AI reinforces itself, potentially discouraging women's participation and leadership. The focus on confidence over accurate self-assessment could stifle nuanced discussions about AI's complexities, particularly the areas where true uncertainty exists.