AI, a Mirror of Prejudice: Women on the Front Lines

Women and AI: Unequal Use and Worrying Consequences

A recent study highlights a notable disparity in the use of artificial intelligence (AI) at work between men and women. According to this research, women tend to use these technologies less frequently than their male counterparts and also receive less recognition for their use. The International Labour Organization (ILO) emphasizes that jobs predominantly held by women are nearly twice as exposed to the risks of automation as those dominated by men.

In the field of AI, only 14% of leadership positions are held by women, reflecting a significant imbalance. Furthermore, 33% of men report using AI daily in their work, compared to only 27% of women. This difference is not limited to usage frequency: men are 27% more likely to be praised for their use of AI and 23% more encouraged by their superiors to adopt these tools. For women, the use of AI often occurs without institutional support, and they sometimes fear being perceived as cheating. Indeed, 29% of the women surveyed expressed this concern, compared to 22% of men.

Design Bias: AI Reflects Gender Stereotypes

The study, conducted by the organization Lean In, reveals that biases present in AI are not limited to its use but also extend to its design. Researchers from Stanford University found that ChatGPT, an advanced language model, associates male pronouns with the word "programmer" 83% of the time, and female pronouns with the word "nurse" 91% of the time, even when explicitly asked to avoid these stereotypes.

In 2026, CV screening tools showed a preference for names that sound white 85% of the time and for male names 52% of the time. These AI models learn from data reflecting the world as it is, rather than as it could be, perpetuating existing inequalities.

Automation Poses Greater Threats to Female-Dominated Jobs

Women adopt AI with more reluctance, even as their jobs are the most exposed to automation. An ILO report published on March 5, 2026, covering 84 countries, indicates that 29% of predominantly female professions are exposed to generative AI, compared to only 16% of male professions.

The sectors most affected by this exposure include secretarial work, accounting, data entry, and customer service. Marie-Sophie Zambeaux, co-author of "Artificial Intelligence in HR (2025)," refers to "second-order bias": AI does not create new prejudices but translates existing biases into algorithms, thereby amplifying inequalities without creating new ones.

Female Skepticism Towards AI and Its Implications

Women are 38% more critical than men regarding the ethical implications of AI and are 29% more likely to question its reliability. Although this skepticism is often justified, it can lead to delays in adopting AI, which has repercussions for their careers.

Anna Sotnikova, a researcher at EPFL, emphasizes that attempts to correct biases in AI often involve trade-offs, and new challenges arise as soon as one problem is solved. While recent models are more balanced regarding certain visual stereotypes, subtle distinctions persist.

In France, the low representation of women in digital fields limits their influence on the development of AI applications. According to the Lean In study, 19% of women anticipate that AI will lead to more female layoffs, compared to 8% of men. Meanwhile, 16% of feminized professions are among the most exposed to AI, compared to only 3% of male professions.

It appears that AI, far from correcting gender inequalities, tends to reproduce and even exacerbate them.