Women perceive AI as riskier than men do, study finds

An online survey found that women consistently perceive AI to be riskier than men. The key drivers behind this view are women’s higher general risk aversion and their greater exposure to AI-related risks. The paper was published in PNAS Nexus.

Artificial intelligence (AI) is a broad term for computer systems that can perform tasks that normally require human intelligence, such as recognizing patterns, making predictions, understanding language, or supporting decisions. AI is used in companies to automate routine work, analyze large amounts of data, improve customer service, and support faster decision-making. For example, firms use AI in chatbots, fraud detection, demand forecasting, recommendation systems, quality control, document processing, and an increasing number of other areas.

The main benefits of AI for companies include greater efficiency, lower costs, faster processing, improved consistency, and the ability to extract useful insights from complex data. AI can also help companies become more competitive by improving products, responding more quickly to customers, and identifying opportunities earlier.

However, AI also brings risks, including inaccurate outputs, biased decisions, privacy problems, security vulnerabilities, and overreliance on automated recommendations. Another important risk is that employees or managers may trust AI too much even when it makes confident but incorrect judgments. Companies also face legal and reputational risks if AI is used in ways that are unfair, nontransparent, or harmful to customers or workers.

Study authors Sophie Borwein and her colleagues note that existing evidence suggests that men are better positioned than women to benefit from the more widespread use of AI, in spite of the fact that women are rapidly overtaking men in post-secondary educational attainment. They believe that this is because men dominate technology-oriented fields and occupations. These authors conducted an online survey hypothesizing that women perceive AI to be riskier than men. They believed that this happens because women are more averse to risk than men and are also more exposed to AI-related economic displacement.

Study participants included 3,049 American and Canadian residents. The participants were recruited through YouGov’s opt-in panel. The survey contained assessments of individual risk orientation, perceived riskiness of AI, and exposure to technological change. Risk orientation was assessed by asking participants whether they would prefer a guaranteed win of $1,000 or a 50% chance of winning $2,000. The study authors considered education level and occupational field to be proxies for exposure to AI-related risk.

Participants also completed an experiment in which study authors manipulated the perceived riskiness of AI by having participants read a scenario about a company adopting generative AI. The researchers varied the probability that the adoption would have a net positive effect on the company’s workforce.

As expected, results showed that women tended to perceive AI as riskier than men did. The study authors concluded that the key drivers of this gap are women’s higher risk aversion and their greater exposure to AI-related risks.

The results of the experiment showed that, as the probability of net positive effects of AI on the workforce decreased, women’s support for the adoption of AI fell more sharply than men’s. Women also tended to express greater uncertainty about the benefits of AI and more frequently expressed views that AI provides little to no benefit in their open-ended answers.

“Our research suggests that the adoption of AI could exacerbate existing gender inequalities if women remain more skeptical and less supportive than men. These emerging patterns of differential adoption of AI technologies may create new forms of occupational segregation that will persist regardless of whether AI ultimately delivers its promised benefits or creates unforeseen consequences,” the study authors concluded.

The study contributes to the scientific knowledge about attitudes toward AI use. However, it should be noted that AI technology is advancing rapidly, and people’s attitudes toward AI are evolving quickly alongside these changes. As AI further develops and the general population gains more experience with it, attitudes may change in line with these experiences.

The paper, “Explaining women’s skepticism toward artificial intelligence: The role of risk orientation and risk exposure,” was authored by Sophie Borwein, Beatrice Magistro, R. Michael Alvarez, Bart Bonikowski, and Peter J. Loewen.