The AI glass ceiling

Why women may avoid AI and what it means for tech diversity

THERE’S A COMMON thread that links many of the enthusiastic endorsements, promotions and celebrations of the latest in AI technology made over the last few years: it’s a lot of men doing the talking. And researchers believe they have started to uncover the source of what some are calling an ‘AI gender gap’ — specifically, that women as a group see the risks differently than men.

Click here to view this article in the London Inc. Worklife newsletter

Looking at survey data from 8,000 workers in the UK, a pair of British researchers recently argued in a paper that the source of the AI gender gap (studies have shown that the share of women adopting AI tools lags men by somewhere between 10 and 40 per cent) may come down to differences in the way the two groups perceive risk.

Story Continues Below

 

Across almost all demographics, they found women were more attuned to the various risks AI presents —job loss, energy usage, mental health concerns and so on — than men and are more likely to shy away from adoption because of it.

“Risk perceptions play a central role in shaping the gender gap in AI use,” wrote Fabian Stephany and Jedrzej Duszynski, a pair of researchers at the University of Oxford’s Oxford Internet Institute. “These wider gaps are driven not by increased use among men but by substantial decreases in AI use among women.”

“It doesn’t surprise me that women notice stories [about these risks] and think, ‘that’s horrible, that’s dystopian, I don’t want any part of that,’” noted McMaster University business professor Catherine Connelly, speaking to HR Reporter. “I think that level of skepticism is very understandable.”

Story Continues Below

 

Some worry that this skepticism (and the lower adoption rates that come along with it) comes at a cost. “These gaps are bad for women because they’re not being as productive as they could be,” reasoned Harvard Business School professor Rembrand Koning, who has also studied the implications of the AI gender gap. “But they’re also bad for the economy because we’re losing out on economic growth we could have had.”

If AI use is to become more equal, the University of Oxford team argues that now is the best time to start addressing risk, in part because their research also showed that addressing the concerns women have with AI can reverse the gap.

“When individuals become more optimistic about AI’s broader impacts,” they wrote, “women — especially younger women — exhibit the largest behavioural change. Policies that address these societal concerns through credible oversight, stronger privacy and accountability mechanisms, or clearer evidence on environmental and labour impacts, may yield more equitable update.” Kieran Delamont