The phantom executives
In a survey of more than 5,400 Americans, Harvard researchers asked a simple question: do you recognize these names? Sam Altman, the CEO of OpenAI. Mira Murati, the former CTO of OpenAI and current CEO of Thinking Machine Labs. Susan Milner. Craig Dalton.
Seventy-three percent of male respondents said they knew who Altman was. Fifty-three percent said they recognized Murati, Milner, and Dalton. There is a problem buried in that second number. Milner and Dalton do not exist. They were invented for the survey, and more than half of the men who answered claimed to know them anyway.
This is the moment where a tool designed to make people smarter reveals something uncomfortable about how people already think. The survey did not measure AI literacy. It measured the willingness to assert knowledge that was never there. And it found that the willingness is not evenly distributed.
Laura Manley, executive director of the Shorenstein Center at Harvard Kennedy School, told WIRED that men are more optimistic and more confident across the board when it comes to AI. [1] Sometimes, she noted, that confidence tips into something else entirely — like claiming to recognize executives who were never born. The pattern is not new. But the AI moment has given that tendency a stage, a microphone, and a business model.
The gap is not in the answers
Here is the finding that should stop you. Men and women answer factual questions about AI incorrectly at roughly the same rate. The difference is not accuracy. The difference is admission.
Women, on average, said they were unsure of an answer to 3.3 out of 12 questions. Men said the same for 2.3 questions. One full question of difference. That single question — the one where a woman says “I don’t know” and a man guesses — is where the entire confidence gap lives. It is not that men are more willing to perform knowing when they do not. It is that the cost of admitting uncertainty falls unevenly — and men are simply less likely to pay it.
The survey also found that men use AI more frequently and are, on average, better able to answer factual questions correctly. Both things can be true at once. Men can be more fluent with the technology and more likely to bluff about it. The two facts do not cancel each other out. They compound.
What does it mean when the people most confident about a technology are also the people most likely to claim familiarity with things that do not exist? It means the discourse around AI is being shaped by a group that has learned to treat uncertainty as a weakness to be hidden. And it means the people who say “I am not sure” are being drowned out before they can finish the sentence.
The echo chamber of expertise
Researchers at Harvard wondered whether this confidence gap shapes what the public reads and hears about AI. So they looked at a year of media coverage. Men were named in headlines about AI more than 14 times as often as women. That number is staggering on its own. But the researchers did not stop there. They narrowed their lens to people quoted as subject experts — not executives, not founders, just the voices brought in to explain.
Three-quarters of those quoted experts were men. The actual AI talent pool is roughly 29 percent women. The coverage was not reflecting the field. It was amplifying a version of the field that does not match the people doing the work.
Manley put it plainly: “Maybe there should be more women prominently featured as experts or given the platform to express: not sure,” she says. [1] The sentence sounds almost too modest. But it is not modest at all. It is a proposal to change what authority looks like. Right now, authority looks like certainty. The survey suggests that certainty is often a performance, and the performers are disproportionately male.
The emotional ledger
Americans across genders agree on one thing: AI is freaking them out. Both men and women rated concern as their top emotional response to the technology. But the agreement ends there.
Women reported higher levels of fear and worry. Men rated every positive feeling — curiosity, enthusiasm, excitement — higher than women did. The same technology, the same moment in history, two different emotional ledgers. One group is more excited. One group is more afraid. And the group that is more excited is also the group more likely to claim it knows things it does not.
That phrase — not sure — is the one the survey keeps circling. Women use it more often, and it is the one answer the current discourse has the least room for. As Manley puts it, “there’s so much confidence in either doom or abundance” that it “just drowns out all of the voices that should be saying ‘not sure.’” [1].
What gets lost when nobody hesitates
There are many unanswered questions about proper uses for AI and about where the technology is going. That is not a controversial statement. It is the baseline condition of the field. How this ends is unknown. What the labor market will look like in a decade is unknown. Whether the safety research will keep pace with the capabilities is unknown.
In a room full of people who are certain, the person who says “I am not sure” is not treated as rigorous. They are treated as unprepared. The survey suggests that this dynamic is not gender-neutral — and that the people most willing to admit uncertainty are the ones least likely to be handed the microphone.
It is a small sentence. It asks for almost nothing. And yet the survey shows how far we are from it. More than half of the men who took the survey could not say “not sure” about a name that did not exist. They said they knew. They said it confidently. And in doing so, they demonstrated exactly what the technology is up against — not the machines, but the humans who refuse to admit what they do not know.
The room after the survey is quiet. The data is on the table. The phantom executives are still there, waiting for someone to admit they never existed.