Artificial intelligence (AI) has become an integral part of our daily lives, influencing everything from social media algorithms to personalized recommendations. However, a new report led by researchers from UCL has shed light on a troubling trend – the prevalence of gender bias in popular AI tools. The study, commissioned by UNESCO, focused on examining stereotyping in Large Language Models (LLMs), including prominent platforms like Open AI’s GPT series and META’s Llama 2.
The findings of the report revealed a concerning pattern of discrimination against women and individuals from diverse cultural and sexual backgrounds. The generated content by the Large Language Models demonstrated strong stereotypical associations, linking female names with words like “family,” “children,” and “husband,” perpetuating traditional gender roles. On the other hand, male names were more frequently linked to terms such as “career,” “executives,” and “business,” reinforcing gender-based stereotypes.
The study also highlighted gender-based stereotyping in the text generated by AI tools, showcasing negative portrayals of individuals based on their cultural or sexual identity. For instance, women were often depicted in domestic roles such as “domestic servant,” “cook,” and “prostitute,” while men were more commonly assigned high-status professions like “engineer” or “doctor.”
Dr. Maria Perez Ortiz, a researcher from UCL Computer Science and a member of the UNESCO Chair in AI at UCL team, emphasized the urgent need for an ethical overhaul in AI development. She stressed the importance of creating AI systems that reflect the diverse human population and promote gender equality. As a woman in the tech industry, Dr. Ortiz called for a shift towards AI technologies that uplift, rather than undermine, gender equity.
The UNESCO Chair in AI at UCL team, in partnership with UNESCO, aims to raise awareness about gender bias in AI tools and work towards developing solutions. By organizing workshops and events involving AI scientists, developers, tech organizations, and policymakers, the team seeks to address the deep-seated gender biases embedded in large language models.
Professor John Shawe-Taylor, the lead author of the report, highlighted the need for a global effort to combat AI-induced gender biases. As the UNESCO Chair in AI at UCL, he stressed the importance of creating AI technologies that uphold human rights and promote gender equity. The report not only exposes existing inequalities but also sets the stage for international collaboration in shaping a more inclusive and ethical AI landscape.
The presentation of the report at key events, such as the UNESCO Digital Transformation Dialogue Meeting and the United Nations session on gender equality, signifies a growing recognition of the need to address gender bias in artificial intelligence. By challenging stereotypes and advocating for ethical development practices, researchers and policymakers can pave the way for a more equitable and inclusive AI future. Just as past disparities in STEM fields do not define the capabilities of women, it is imperative to create AI technologies that reflect and respect the diversity of the human experience.
Leave a Reply