Experts Warn AI Risks Bias Without Women in Training

Experts warn that artificial intelligence tools risk being unrepresentative and biased without women involved in their training, as predominantly male teams often overlook gender balance in data and outputs.

Karina Gibert, dean of the Official College of Computer Engineering of Catalonia, and Mònica Espinosa, director of the Innovation and Skills Centre at the Catalan Cybersecurity Agency, made the case during an event at the Auditori MoraBanc. They argued that AI trained without diverse input fails to reflect society equitably.

"If you don't train it properly, AI can exhibit non-neutral behaviors and dysfunctions," the speakers noted. Male-dominated teams, they said, frequently neglect to ensure women are adequately represented in training data or that the tools perform equally for men and women, sometimes showing favoritism toward the latter.

The root issue lies in education and early career pipelines, Gibert explained. "You can't place women in environments where none exist," she said, calling for efforts to inspire primary and secondary schoolgirls toward technology careers. Few women are trained as professionals in the field, though Gibert pointed to progress: at her faculty, female enrollment has risen from 5% six years ago to over 22%, thanks to targeted initiatives.

Lack of female role models perpetuates the cycle, with low representation in tech leadership likely to persist without visible examples. Both Gibert and Espinosa embody such figures, having forged paths in male-only settings to reach their senior roles.

Gibert expressed no regrets about her career choice, driven by her strong aptitude for mathematics. Upon entering the workforce, they found themselves as the lone women in decision-making rooms, often viewing issues from fresh perspectives. "You're always the lady who's there and sees things differently," Espinosa described, adding that even debates remain challenging as the only woman present.

For years, they have felt like necessary "dissidents" to counter male-biased influences and deliver fair, ethical technology.