Women 20% Less Likely to Use Generative AI Than Men, Study Finds

Women are about 20% less likely than men to use generative artificial intelligence, according to a new study, both at work and in everyday life. They account for 42% of global ChatGPT website users, 42.4% of Perplexity users, and only 31.2% of Anthropic’s Claude users.

Gender gap spans regions and sectors

Researchers from the Haas School of Business analysed data from 18 studies, encompassing more than 140,000 individuals worldwide, and found that the gender gap persists in nearly all regions, sectors, and occupations. Across 12 studies, women had 22% lower odds of using generative AI than men.

The study also took into account data from SimilarWeb, which draws on first-party analytics shared by websites and apps, as well as other sources. This showed that only 27.2% of global ChatGPT app downloads were from women, and it’s 25.2% in just the US.

Survey in Kenya shows persistent gap

The researchers, who also hail from Stanford University and Harvard Business School, conducted a proprietary survey tool. In a poll of more than 17,000 adults in Kenya were asked whether they wanted guidance on downloading or accessing ChatGPT. The results found that women were still 13.1% less likely to adopt the tool, even when offered support.

“If this global disparity persists, it risks creating a self-reinforcing cycle,” the authors wrote. “Women’s underrepresentation in generative AI usage would lead to systems trained on data that inadequately sample women’s preferences and needs, ultimately widening existing gender disparities in technology adoption and economic opportunity.”

Drivers behind women’s lower adoption

The paper highlights several drivers of the gender gap. Women report lower awareness of AI tools and less understanding of how to use them, particularly because they may work in roles less exposed to technology. They also express less confidence when prompting chatbots and are less persistent when outputs are unsatisfactory.

Finally, women are more likely to view AI use as “cheating” in academic or work settings, and perceive that they see less productivity or fewer educational benefits than men. The latter could be related to how many models are trained on data that is biased toward men, making them less useful for women. 

Biases in AI systems reinforce inequality

For example, studies have shown that AI tools used in hiring decisions are more likely to recommend men over women, especially for higher-paying jobs, because they are trained on job adverts that have gendered stereotypes embedded in them. More than half of England’s local councils use AI systems that may downplay women’s physical and mental health needs.

The researchers suggest that the original biases in AI models can be self-reinforcing. If the technology proves more useful for men, more men will adopt it, and the systems will continue to learn from male-dominated data. 

Evidence of systemic sexism in AI

Evidence that AI is already embedded with sexism includes the fact that the default voice options on most voice assistants, including Alexa, Siri, and Cortana, are female. Microsoft made Cortana female because it was seen as the “stronger choice” for a “helpful, supportive, trustworthy assistant,” according to the Wall Street Journal, a rationale that reinforces stereotypes by equating servitude with femininity.

There will also be financial consequences of biased AI, given how AI is predicted to boost global economies. In the US alone, a persistent 25% usage gap could translate into hundreds of billions of dollars in lost productivity, the paper’s authors wrote. 

As the Kenya-based study shows, the gender gap persists even when access is equalised, indicating that more work needs to be done to identify the root cause. “Without such efforts, generative AI’s potential to drive economic growth and improve productivity will almost surely disproportionately benefit male users, not only entrenching existing gender gaps, but also opening the possibility that society will miss out on the work, inventions, and ideas women would have produced with this new technology,” they wrote.

Explore how bias can creep into AI recruitment tools and what companies can do to keep hiring both smart and fair.