Shocking UN Report Reveals AI Is Learning Sexism, Racism From Humans

We keep hearing that AI is going to be smarter than humans, people like Elon Musk say it already is in some ways. But a new UN warning suggests AI may have inherited some of humanity's ugliest biases.

Researchers studied 133 AI systems and nearly half (44%) showed gender bias, while more than a quarter showed both gender and racial bias.

"Large language models (LLMs) have repeatedly associated women with the home, family and childcare, while linking men to business, leadership and career success." 

"In some cases, AI systems have generated responses portraying women as sexual objects or as subordinate to men," the UN said.

According to UN Women, when researchers asked large language models to simply complete a sentence that began with a person's gender, about one in five responses came back sexist or misogynistic. 

Some even described women as property, as objects, according to the research report.

The UN pointed out that as AI is increasingly being used to write emails, create presentations, generate content and answer questions for billions of people, the inequalities are also being reinforced through discriminatory algorithms.

According to UN Women, this isn't a bug - it's because AI learns from the internet which contains decades of stereotypes, prejudice and discrimination. So when AI trains on human history, it can end up reproducing human bias.

Jayathma Wickramanayake, UN Women Lead on Digital Technologies, said that AI models "pull bias from decades of text written by people, about people, in a world where women were filed under home and family, and men were filed under business and career". 

Women already face disproportionate levels of abuse online, and AI is making some forms of violence easier to create and spread, the report highlighted.

According to UN Women data, nearly one in four surveyed women human rights defenders, activists and journalists reported experiencing AI-assisted online violence; 12% said personal images had been shared without their consent, while 6% reported being targeted by deepfakes or manipulated images and videos.

UN Women is batting for gender equality - and the rights and experiences of women and girls - to be integrated at every stage of the AI lifecycle, from development through to deployment and governance.

Researchers also found racial bias across many systems, raising concerns that AI could reinforce stereotypes about multiple groups, not just women.

We often hear about AI replacing human intelligence, but perhaps the bigger challenge is making sure it doesn't inherit humanity's worst habits and biases.