I can barely get through a conversation without talking about AI.
Whether it’s someone talking about it with an almost religious fanaticism or simply just using it for task planning and questions, AI’s reach feels ubiquitous.
And while that’s certainly my impression here at BU, AI has varying levels of popularity, especially when looking at the gender divide.
In a 2025 preprint paper titled “Global Evidence on Gender Gaps and Generative AI,” Nicholas Otis and Solene Delecourt of University of California, Berkeley, Katelyn Cranney of Stanford and Rembrand Koning of Harvard Business School investigate the gender gap in AI usage both through a meta analysis of aggregated observational studies and their own experimental study.
The authors found women are roughly 20% less likely than men to use AI, a gap they say “is not merely the result of gendered survey response bias.” The authors also report that “even when opportunities to access generative AI were equalized, the gender gap in generative AI use persists.”
I’ll reiterate: even when controlling for equal access to technology, we still have lower rates of usage from women.
Koning says in a discussion with the Harvard Business School magazine this phenomenon may be a result of how women fear they are perceived by AI.
“Women face greater penalties in being judged as not having expertise in different fields,” he said. “They might be worried that someone would think [that] even though they got the answer right, they ‘cheated’ by using ChatGPT.”
Considering historic stereotypes regarding women’s diminished capabilities, leadership skills and overall intelligence, the need to prove competency and shirk the short cut of AI is another barrier that men might not have to face.
The AI gender gap can lead to pernicious ramifications. For instance, AI has the potential to boost productivity. But without equal adoption, we see an imbalance of productivity gains skewed towards men. This worsens the existing gender inequality by reinforcing beliefs that men are superior to women in the workforce, according to the preprint paper.
Furthermore, machine learning and AI models work by taking in user input and improving its responses based on the feedback, according to an article from The Wall Street Journal.
Thus, as stated in “Global Evidence on Gender Gaps and Generative AI,” AI models with a disproportionate amount of male input — or input that generally doesn’t reflect the true makeup of society — will simply amplify biases.
Generative AI users also report that there is a difference in companies’ promotion of AI tools between employees of different genders. Deloitte reports in 2024 that 49% of women users say their company invests in training employees on using generative AI as compared to 79% of men.
Out of the 18 studies observed in the Otis, Delecourt, Cranney and Konning paper, there is only one that demonstrates an approximately equal usage of generative AI. It’s from a Boston Consulting Group study of tech employees, where there was actually a 3% increase in female usage over men.
But this makes sense: of course the numbers are going to look more equal in a technological industry where everyone presumably has a similar education and, more importantly, is similarly excited to adopt new technologies.
We’ve essentially controlled for societal gender bias by leveling the playing field in terms of background experience and attitude towards technology.
Now we just need to do that everywhere.
AI is the future. While I don’t believe in the usage of AI for every task — and I strongly believe that we need more rigorous ethical guidelines on its worldwide implementation — it would be naive to suggest that the increase in AI and machine learning hasn’t led to powerful advances in fields like medicine and computer science.
We’ve already seen AI becoming more and more integrated into everyday life — looking at you, Google AI overview — and that will only continue.
Of course, this research suggests that we need to improve the rates of women in tech. But we also need companies to step up and equally train their employees in AI methods. We need universities to do their part by teaching fair and equitable ways across majors to implement AI.
We can’t shirk it altogether and avoid the inevitably of this new technology.