Subscribe
UN Women has warned that AI systems are reinforcing gender stereotypes and racial bias.
A study of 133 AI systems found 44% showed gender bias, while 26% showed both gender and racial bias.
The agency says AI is also intensifying online violence against women and girls, including through deepfakes and image-based abuse.
UN Women has called for gender equality to be embedded across AI development, deployment and governance.
Artificial intelligence is already shaping how billions of people see the world, but it is getting women wrong, UN Women has warned. Ahead of two major UN-linked AI meetings in Geneva in July, the agency said gender equality must be built into the design, deployment and governance of AI systems.
UN Women cited a study of 133 AI systems which found that 44 per cent showed gender bias, while 26 per cent showed both gender and racial bias. It said the risks were growing as generative AI became widely used in marketing, communications and media production.
In the UK, 88 per cent of advertising and media agencies are already using generative AI in some form, according to the agency. But only 51 per cent of marketers currently use human oversight to test AI-generated creative work before it is released.
UN Women said large language models had been found to associate women with terms such as “home”, “family” and “children”, while linking men with “business”, “executive”, “salary” and “career”. In some cases, it said, AI-generated responses reflected sexist and misogynistic attitudes.
The agency also warned that AI was worsening online violence against women and girls. It said almost one in four surveyed women human rights defenders, activists and journalists had experienced AI-assisted online violence. Some reported non-consensual sharing of personal images, deepfakes or manipulated videos, and unsolicited sexual advances through digital messaging.
Women also remain underrepresented in the development of AI systems, making up about 30 per cent of the global AI workforce, UN Women said. It warned that this lack of representation meant the people designing these tools did not reflect the wider population they were meant to serve.
The agency said women outside the AI sector were also nearly twice as likely as men to hold jobs at high risk of automation, with the greatest impact likely to fall on groups already underrepresented in media and labour markets.
UN Women said inclusive AI was not only a rights issue but also a commercial concern. It cited research by the Unstereotype Alliance showing that advertising free of gender stereotypes could improve sales, brand value and customer loyalty.
The agency said AI could help detect stereotypes and improve representation, but only if women and girls were included in decisions about how the technology is built and used.
