Men dominate AI headlines as confidence outpaces accuracy

Confidence is not the same as accuracy

The report found that men use AI more often and answer factual questions about it better on average. But the error rate was roughly the same for both groups. The difference was how often respondents said they were unsure: women hesitated on an average of 3.3 out of 12 questions, compared with 2.3 for men.

One example makes the distinction vivid. Among women surveyed, 40% said they knew Milner and Dalton, who are fictional AI-company leaders; 38% said they recognized Mira Murati, who is real, formerly OpenAI’s chief technology officer and now leads Thinking Machine Labs.

Laura Manley, executive director of Harvard Kennedy School’s Shorenstein Center, says men tend to be more confident and optimistic about AI. That confidence can become overconfidence, she says, and men also tend to show more overconfidence in surveys generally.

Who gets called an expert

The report looked beyond individual responses to how AI is covered in the media. Men appeared in AI headlines more than 14 times as often as women over the course of a year. Since major AI companies are led predominantly by men, the researchers also examined stories quoting people as experts: three-quarters of those experts were men, a higher share than men’s representation among AI specialists, where women make up about 29%.

Manley argues that women experts should be given more prominent platforms, including space to say openly when they are unsure.

The emotional responses in the report complicate any simple story about men embracing AI while women reject it. Americans of both sexes most often named worry as their leading reaction. Women were more likely than men to report anxiety, fear and apprehension; men rated every positive emotion higher, including curiosity, enthusiasm and excitement.

The value of saying “I don’t know”

Manley says public arguments about AI are too confident both in predicting catastrophe and in promising abundance. Those voices can drown out people who might say, “I don’t know.” The report’s numbers do not show that women are more accurate; they show that women are more likely to acknowledge uncertainty even when men and women make mistakes at similar rates.

I think the sharper issue is not who sounds most certain about AI, but whose uncertainty gets treated as useful rather than as a reason to stop listening. If expert coverage continues to favor confidence over calibration, the people best placed to describe what remains unknown may keep getting less room to speak.