Women appear more cautious about artificial intelligence than men, according to a new study that links the difference to risk tolerance and exposure. Researchers at Northeastern University in Boston analysed surveys of around 3,000 Canadians and Americans to understand why women are slower to adopt AI tools and more hesitant about its workplace applications. Their findings were published this month in PNAS Nexus.
The study found that women are generally more risk-averse than men. In survey scenarios involving financial choices, women were more likely to select a guaranteed $1,000 (€842) rather than take a 50 percent chance of receiving $2,000 (€1,684) or nothing. This cautious approach extended to AI, with female respondents about 11 percent more likely than men to say the risks of AI outweighed its benefits.
When asked open-ended questions about AI, women expressed more uncertainty and scepticism than men. However, the researchers discovered that the gender gap largely disappeared when outcomes were certain. If AI-driven job gains were guaranteed, both men and women responded positively, and women who were less risk-averse in the survey expressed similar levels of confidence as men.
“Basically, when women are certain about the employment effects, the gender gap in support for AI disappears,” said Beatrice Magistro, assistant professor of AI governance at Northeastern University and co-author of the study. “So it really seems to be about aversion to uncertainty.”
The study also highlighted that women are more exposed to the economic risks posed by AI. Many hold roles in sectors where AI could either complement or replace human work, increasing both potential benefits and vulnerabilities. This dual exposure amplifies uncertainty and contributes to more cautious attitudes toward the technology.
Women currently adopt AI tools at a rate 25 percent lower than men and represent less than one in four AI professionals globally. The researchers suggest that these patterns should inform policymaking, particularly in crafting regulations that protect workers from the disruptive effects of AI.
Potential measures include policies to reduce the risk of job displacement, compensatory schemes for workers affected by AI, and strategies to address gender bias in AI systems. The goal, researchers said, is to ensure AI adoption does not leave women behind as the technology reshapes workplaces.
The findings underscore broader societal challenges as AI expands. From deepfake technologies to automation, some of the most harmful effects of AI have disproportionately impacted women. By understanding the drivers of scepticism, experts hope to create a more inclusive environment for AI adoption that balances opportunity with safeguards.
