AI is getting women wrong as gender bias persists, data reveals
AI is getting women wrong as gender bias persists, data reveals
AI is getting women wrong as gender - Artificial intelligence is reshaping the way people work, interact, and access information, yet new evidence shows it continues to replicate outdated gender stereotypes. These biases not only deepen societal inequalities but also contribute to online harassment and exclude women from shaping the digital landscape. UN Women has highlighted these concerns, warning that without deliberate efforts to address them, AI will continue to reinforce harmful narratives about women and men.
Technology Reflects Deep-Seated Inequalities
Recent studies reveal that as generative AI becomes more integrated into daily life, it is amplifying existing disparities. From drafting emails to crafting marketing strategies and generating reports, the technology is increasingly being used by industries that have historically marginalized women. In the United Kingdom alone, 88 percent of advertising and media agencies now incorporate AI in their operations, according to data shared with UN News.
However, this widespread adoption does not guarantee fairness. A comprehensive analysis of 133 AI systems found that 44 percent exhibited gender bias, while over a quarter demonstrated discrimination across both gender and race. This indicates a systemic issue rather than isolated mistakes. Large language models, in particular, have shown a consistent pattern of associating women with domestic roles and childcare, while positioning men as central to business and professional success. Such biases are evident in AI-generated responses that depict women as sexual objects or emphasize their subordination to men.
"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," said Jayathma Wickramanayake, UN Women's Lead on Digital Technologies. She emphasized that this is not a technical error but a reflection of long-standing societal norms embedded in the data that trains AI systems.
Wickramanayake further noted that the problem lies in a lack of inclusive policy design. Of 138 countries surveyed globally, only 24 explicitly addressed gender in their national AI strategies, and just 18 included concrete measures to ensure gender-responsive development. "This isn’t a bug waiting to be fixed in the next update," she explained. "It’s a choice we make repeatedly—whether in the data we feed AI, the teams designing it, or the documents that define its governance."
Online Abuse Amplified by AI
For many women and girls, the consequences of AI bias extend beyond perception. The technology is enabling new forms of digital violence that are easier to produce and disseminate. UN Women data shows that nearly one in four women human rights defenders, activists, and journalists reported experiencing AI-assisted online harassment. Twelve percent described situations where personal images were shared without their consent, and six percent encountered deepfakes or manipulated videos targeting their identities.
As AI-generated content becomes the standard, the risk of undetected abuse grows. Harassment, misinformation, and image-based attacks are increasingly automated, making it harder for victims to seek recourse. "The economic impact of AI may also fall unevenly," added Wickramanayake, noting that women are nearly twice as likely as men to hold jobs at high risk of automation, particularly in sectors outside the AI field. These disparities are compounded by factors such as race, disability, income, and geographic location.
Underrepresentation in Tech Development
Another concern is the lack of diversity in the industries creating AI. Despite the technology's pervasive influence, women constitute only 30 percent of the global AI workforce, according to the International Labour Organization (ILO). This underrepresentation risks perpetuating biases, as the people designing systems may not reflect the full spectrum of societal experiences.
Without active inclusion of women and other underrepresented groups, the digital future may be shaped by perspectives that overlook their unique challenges. "The algorithms that govern our online world are learning from the world as it is," said Wickramanayake. "If women are not included in the process, their needs and contributions will be ignored in the blueprint for AI."
The UN Women organization is calling for urgent action before key global events, such as the upcoming United Nations Global Dialogue on Artificial Intelligence Governance and the AI for Good Global Summit in Geneva in early July. These gatherings offer an opportunity to address the gaps in AI policy and ensure that the technology serves as a tool for equity rather than a mechanism for exclusion.
Towards Inclusive AI Design
To combat these issues, experts advocate for embedding gender equality into every stage of AI development—design, deployment, and regulation. This includes diversifying the teams behind AI systems, auditing algorithms for bias, and integrating feedback from women and girls in the digital space. By doing so, the technology can evolve to better represent the realities of all individuals.
Wickramanayake stressed that the solution requires more than technical adjustments. It demands a cultural shift in how AI is conceptualized and implemented. "We need to challenge the assumptions that underpin AI systems," she said. "Only then can we create technologies that empower women and advance equality in the digital age."
As AI continues to transform labor markets and daily life, the call to action grows stronger. Without targeted measures, communities already facing systemic exclusion may see their struggles exacerbated. The integration of AI into society offers immense potential, but its success hinges on whether it is built with the voices of all its users in mind.