Because AI outputs are no better than the human-generated data it’s trained on, the UN agency concludes judicial and social interventions will be necessary to address gen AI risks. Credit: Getty Images Generative AI’s outputs still reflect a considerable amount of gender and sexuality based bias, associating feminine names with traditional gender roles, generating negative content about gay subjects, and more besides, according to a new report from UNESCO’s International Research Centre on Artificial Intelligence. The report, published today, centered on several individual studies of bias, including tests for associations between gendered names and careers, frequently generated less positive responses to prompts related to LGBTQ+ individuals and women, and assigned stereotyped professions to members of different genders and ethnic groups. The researchers found three major categories of bias underlying generative AI technologies. The first is a data issue, in which an AI isn’t exposed to training data from underrepresented groups or doesn’t account for differences in sex or ethnicity, which can lead to inaccuracies. The second is algorithm selection, which can result in aggregation or learning bias. The classic example of this would be an AI identifying resumes from male job candidates as more desirable based on gender-based disparities already present in hiring practices. Finally, the study identified biases in deployment, where AI systems were applied to different contexts than the ones they had been developed for, resulting in “improper” associations between psychiatric terms and specific ethnic groups or genders. Each form of bias present within the large language models (LLMs) underpinning modern AI systems reflects the texts on which the LLMs are trained, the authors of the UNESCO report wrote in an introduction. Because these texts have been generated by humans, the LLMs, therefore, reflect human biases. “Consequently, LLMs can reinforce stereotypes and biases against women and girls, practices through biased AI recruitment tools, gender-biased decision-making in sectors like finance (where AI might influence credit scoring and loan approvals), or even medical or psychiatric misdiagnosis due to demographically biased models or norms,” they wrote. The researchers noted that their study was not without its limitations, discussing several potential challenges, including limitations on implicit association tests, data contamination, deployment bias, language limitation, and the lack of intersectional analysis. Nevertheless, the UNESCO study said, the evidence is that gender bias in particular is “pervasive” within AI systems, and it’s “critical” to include anti-discrimination measures at the ground level of the AI development cycle to address these issues. “Addressing these risks requires holistic approaches, including judicial and social interventions, alongside technological solutions that ensure AI’s equitable and responsible application,” the authors wrote. “Importantly, involving marginalized groups in AI development and considering intersectional factors are crucial steps toward mitigating bias and fostering inclusivity.” Related content brandpost Sponsored by HP How managed device services enable the inclusion of sustainability principles into IT Find out how to Improve resiliency, reduce costs and create better experiences, with the HP-sponsored IDC whitepaper on the benefits of managed device services. By Sherry Brecher May 24, 2024 3 mins Green IT brandpost Sponsored by HP Monitoring the surface of the sun with NASA’s Solar Dynamics Observatory NASA and its contractors are increasingly turning to AI as a reliable way to collect and analyze data. Find out how Z by HP has enabled them to transform the way they analyze data. By Sherry Brecher May 24, 2024 3 mins Artificial Intelligence brandpost Sponsored by HP Designing employee communication for the modern office Three keys to ensure collaboration, productivity, and connection. By Sherry Brecher May 24, 2024 2 mins Employee Experience brandpost Sponsored by HP How to implement zero trust device security Improve cybersecurity posture and build a foundation for corporate-wide trusted digital workspaces, with HP Anyware’s Zero Trust device security. By CIO May 24, 2024 3 mins Endpoint Protection PODCASTS VIDEOS RESOURCES EVENTS SUBSCRIBE TO OUR NEWSLETTER From our editors straight to your inbox Get started by entering your email address below. Please enter a valid email address Subscribe