73 Percent of Men Claimed to Know an AI Executive Who Does Not Exist

Seventy-three percent of male respondents told Harvard researchers they recognized OpenAI CEO Sam Altman. A little over half—53 percent—also claimed to know Mira Murati, Susan Milner, and Craig Dalton.

There is one problem with that answer. Murati is real: she is the former CTO of OpenAI and current CEO of Thinking Machine Labs. Milner and Dalton do not exist.

The finding comes from a new survey of more than 5,400 Americans, published Wednesday by the Harvard Kennedy School's Shorenstein Center. It is the latest piece of evidence that self-reported AI knowledge is unreliable—and that the unreliability is not distributed evenly across genders.

By James Calloway

What did the survey actually find?

The headline number is the phantom-executive test. Researchers included two invented names alongside one real figure. Men fell for it at higher rates than women: 53 percent of men claimed familiarity with all three names, versus 40 percent of women who claimed to know Milner and Dalton and 38 percent who recognized Murati.

Compared to women, "we're seeing that men are more optimistic and more confident" across the board on AI, Shorenstein Center executive director Laura Manley told WIRED. Sometimes that veers into overconfidence—like asserting you recognize executives who were never born.

Manley offers a caveat worth keeping in view: men tend toward overconfidence in surveys generally, not only on AI. The study does not isolate why the gap exists. Takeaway: men report more certainty about AI, and some of that certainty attaches to people who are not real.

Do men actually know more about AI?

Partly. Researchers found that men use AI more frequently and, on average, answer factual questions about AI correctly more often.

But the margin is narrower than the confidence gap suggests. Men and women answer questions incorrectly at roughly the same rate. The real difference sits in the "not sure" column: women said they were unsure of an answer for an average of 3.3 questions out of 12, compared with 2.3 for men.

That distinction matters. The gap in correct answers comes largely from willingness to guess—and guessing sometimes produces false familiarity with fictional executives. The survey design, a forced-recognition test with planted fake names, is a standard technique for catching exactly this behavior. Takeaway: men do somewhat better on AI facts, but they also bluff far more often.

How does the confidence gap show up in media coverage?

Researchers then asked whether this tilt toward certainty colors journalism. It does.

Across a year of media coverage, men appeared in AI headlines more than 14 times as often as women. Because top AI executives skew heavily male, the researchers re-ran the analysis looking only at people quoted as subject experts—removing the excuse that the sourcing simply mirrors who runs the industry.

The result: three-quarters of quoted AI experts were men. That is a higher proportion than the actual AI talent pool, which is roughly 29 percent women.

In other words, coverage amplifies the existing imbalance rather than reflecting it. "Maybe there should be more women prominently featured as experts or given the platform to express: not sure," Manley said.

There is a tension in that quote worth noting. Journalists generally prize experts who speak decisively; a platform for "not sure" runs against both sourcing norms and quote-driven formats. Manley is proposing a change in what counts as quotable. Takeaway: male experts outnumber women in AI coverage at rates beyond even the field's skewed demographics.

Are Americans afraid of AI?

Yes—across genders. Men and women both rated "concern" as their top emotional response to AI.

The divergence is in degree. Female respondents reported more concern, fear, and worry than men. Men rated every positive feeling—curiosity, enthusiasm, excitement—higher than women did.

That pattern tracks with the confidence data: the same disposition that produces false recognition of invented executives also produces higher enthusiasm. Neither emotional register tells you whether a respondent understands the technology.

The stakes are real. Public attitudes shape policy debates, adoption decisions, and the political appetite for regulation. A population split between overconfident enthusiasm and elevated fear leaves little room for the people measuring what is actually known. Takeaway: everyone is worried about AI, but men pair their worry with noticeably more optimism.

What gets lost when nobody says 'not sure'?

Manley's sharpest point concerns the structure of the AI conversation itself. "There's so much confidence in either doom or abundance," she said. "It just drowns out all of the voices that should be saying 'not sure."

The survey's own numbers support that framing. Uncertainty is measurably gendered, with men withholding "not sure" at lower rates than women, and the expert class quoted in coverage skews further male than the industry itself.

The unanswered questions—on proper uses of AI and its future—remain numerous. One gender is simply less likely to acknowledge them. "We have to be able to say 'not sure' more often," Manley said.

The study does not test whether saying "not sure" correlates with better decisions, and it does not claim to. It documents a gap and its footprint in media. But the phantom names do the persuasive work here: a discourse in which 53 percent of men claim familiarity with a nonexistent executive is a discourse with an honesty problem, not just a knowledge problem.