A Comparison of Laypeople’s Use of Large Language Models vs. Search Engines for Health Queries: Survey Study (Preprint)

BACKGROUND Search engines have transformed the way laypeople access health information. Large language models (LLMs) are poised to enable another shift in health information seeking. OBJECTIVE Characterize how laypeople’s early use of large language models (LLMs) for health information compares to the use of search engines. METHODS We conducted a screening survey about the use of LLMs and search engines, and a follow-up survey probed the use of LLMs compared to search engines for health queries with 281 U.S. participants recruited on Prolific. RESULTS LLMs were perceived as less useful than search engines for answering health-related questions but elicited less negative feelings in response to results, seemed more human, and were perceived as less biased. CONCLUSIONS Search engines remain the main source of information for laypeople seeking health information, yet positive perceptions of LLMs suggest their use may grow. With greater public reliance on LLMs, it is increasingly important for clinicians and health organizations to partner with LLM providers to improve the quality of health-related output.