An interdisciplinary team of University of Washington researchers, including many affiliated with the Center for an Informed Public, was one of 14 to be awarded a Tier 1 pilot grant this fall from the UW Population Health Initiative for a project that aims to help artificial intelligence developers, healthcare providers, public health organizations and policymakers improve the safe application of generative AI for people’s health information seeking and decision making.
Core to the project, “Understanding the Role of AI-Integrated Information Seeking Tools in Users’ Evaluation of Health (Mis)information,” are implications for health equity since populations with limited access to professional medical resources may rely more on online health information, making the accuracy and influence of these tools vital for improving population health and reducing disparities.
According to the project abstract:
<blockquote>Many people rely on online search engines for timely and low-cost health information and advice. In 2023, almost 13 million U.S. adults reported using generative AI as their primary tool for search, a number that is projected to increase with search engines like Google and Bing adopting these technologies. While these tools offer potential benefits, their risks, particularly regarding their influence on users’ perceptions and evaluation of health misinformation online, remain underexplored. Previous studies have shown that generative AI models that power these tools are prone to inaccurate and biased results, effects that are known to disproportionately impact marginalized and underrepresented groups, raising concerns of further exacerbating health inequity.</blockquote>
The project investigators are led by CIP postdoctoral scholar Yiwei Xu, who is based at the Information School; Xinyi Zhou, a Paul G. Allen School of Computer Science & Engineering postdoctoral scholar; Saloni Dash, a CIP graduate research assistant and doctoral student based at the Information School; CIP co-founder and director Emma S. Spiro, an Information School associate professor; CIP faculty member Amy X. Zhang, a Paul G. Allen School of Computer Science & Engineering assistant professor; and Wang Liao, a Department of Communication assistant professor.
In a separate project, “Evaluating large language models in conveying determinants of mental health,” funded through a recently awarded UW Medicine Garvey Institute for Brain Health Solutions grant, Xu, working as a co-investigator, will be evaluating the bias and accuracy of large language models when conveying the causes of mental health disorders, including anxiety. This is a collaboration with Xinyi Zhou (Lead) and Tim Althoff from the Paul G. Allen School of Computer Science & Engineering, and Sarah Gollust from University of Minnesota School of Public Health.
Photo above: The Hans Rosling Center for Population Health on the campus of the University of Washington in Seattle.