Why ChatGPT Thinks Professors Are Men—and So Do Our Kids

A few years ago, my friend Koleen and I had a party. It wasn’t just any party; it’s what we called the “Professor Party,” as it was to celebrate the two of us being promoted to full professor—the highest rank we could attain in academia.

The party was great—people dressed up in their favorite nerdy attire and danced to music from a live band. The party was so much fun that we’ve had it every year since, with this year marking its fourth.

This year, instead of doing the same thing we’ve done for the last three, we thought of something special. Both Koleen and I have become department chairs since our last promotion, and Koleen will be ending her term this summer. Given that the party is always outdoors, we decided to have a bonfire, burning an actual chair to mark the end of this important accomplishment.

ChatGPT output to prompt.

Source: Vanessa LoBue/ChatGPT

Naturally, when making the invitations, I wanted to feature the burning chair, so I asked ChatGPT (as one does) to generate an image for me of “professors having an outdoor party and burning a chair.” I had no words for what it gave me.

In the image (pictured here), there were 10 people depicted celebrating around a burning chair—all of whom were men.

Chat GPT relies on data from Wikipedia, books, news articles, scientific journals, and publicly available data on the internet. This means that according to the internet, and the majority of books, news articles, scientific journals, and general information available on the web, professors are men and only men.

Gender Representation in Academia

You’d think in 2025 this wouldn’t happen—of course there are professors who are women. But statistically, women are still underrepresented in faculty positions at universities, especially at the higher levels.

For example, women currently represent 43 percent of full-time and tenure track professors, while men represent the other 57 percent. When you break this down by rank, women represent 50 percent of professors at the lowest (assistant) rank, 45 percent at the middle (associate) rank, and 33 percent at the highest (full professor) rank, compared to men, who have 67 percent of these top positions.

Why does this happen? There are probably a lot of reasons, for example, a lack of support for women in these positions, particularly lack of support for women who decide to have children over the course of their careers; but research suggests that the problem starts way before that, likely originating in early childhood.

Indeed, research has shown that even children think of scientists as men. In an analysis of over 20,000 children who were simply asked to “draw a scientist,” researchers reported that most children draw a man, and this gets worse as children get older. More specifically, while boys consistently draw men across the lifespan, as girls get older, they become less and less likely to draw a woman (Miller et al., 2018).

Perceptions of Intelligence

Further, being the type of person who is good at science—and even being the type of person who is smart—is also something girls identify with less and less as they get older. One study showed that by the time girls are 6 or 7 years of age, they are becoming less and less interested in games that are designed for “really smart” kids, and they are less likely to associate being smart with being a girl (Bian, Leslie, & Cimpian, 2017).

On top of defaulting to men when asked to draw a scientist, there is also evidence that the default for drawing any person is more likely to be male. More specifically, when children are asked to “draw a person,” kids initially go with their own gender, with girls drawing girls and boys drawing boys.

Gender Essential Reads

The Motherhood Penalty and the Fatherhood Competence Gap

The Evolution of Sex: Gender, Mating, and Social Relations

But as kids get older, this trend remains stable for boys—boys of all ages draw boys—but while younger girls draw their own gender, girls become more and more likely to draw boys as they get older (Lei et al., 2022). In other words, as children get older, they are more likely to not only default to men as scientists, but also as people.

Adult Influence on Gender Perception

Why is this happening?

There is evidence that kids are not just coming up with this on their own; they are most likely learning this from the adults around them. In fact, research shows that when asked to think of a person, even adults (both male and female adults) overwhelmingly default to men. In other words, when asked to describe the most typical person you can imagine, both men and women typically describe a man.

Parents seem to pass down this message to their children. In a very recent study, parents of 4- to 10-year-old children were shown photos of children playing on a playground and were asked to provide a caption. Parents were more likely to use gender-neutral language like “the kid is sliding” when describing boys and gender specific labels like “the girl is sliding” when describing girls (Leshin et al., 2025).

The moral of the story here is that gender bias in higher education is still alive and well, and that this bias is more evident than ever when you rely on AI, which uses the information available on the internet—from us—from which to draw. This means that relying on AI will reflect back the biases that are already evident in our culture, including gender bias, but also other biases based on race, ethnicity, socioeconomic status, and many, many more. Indeed, although the most obvious thing for me in the ChatGPT professor image was the lack of women, the second most obvious should be the lack of racial and ethnic diversity.

In order to tackle this bias for good, we need to start early. Our youngest children are the ones most likely to draw scientists and people more generally as female, and only default to male after years and years of experience with our cultural norms. That means there is hope to change this trend, but it will depend on us—parents, teachers, family, and friends—to make that change happen, both in our everyday lives and on the internet.