A topic modeling analysis of health decision delegation to generative artificial intelligence

Generative artificial intelligence (GenAI) platforms are rapidly becoming trusted companions in users’ medical decision-making. Although users increasingly rely on these tools, little is known about the types of health decisions they delegate to GenAI and the motivational processes affecting such delegation. This study examines the health-related tasks that users describe outsourcing to GenAI and, drawing on regulatory focus theory, investigates how these delegated tasks map onto prevention- and promotion-focused self-regulatory motives. We analyzed 15,592 naturally occurring public conversations referencing GenAI and health from over 10 social media platforms. Comments were preprocessed and analyzed with Latent Dirichlet Allocation (LDA) to identify dominant themes, and temporal orientation (past, present, future) was assessed using a zero-shot classification model. The results revealed two distinct thematic clusters. The first (71.06%) relates to prevention-focused delegation, including symptom checking, diagnostic clarification, and risk-related reassurance. The second (28.93%) reflected promotion-focused delegation, such as personalized diet plans, exercise routines, and lifestyle optimization. Prevention-related comments were more likely to reference past-oriented experiences, whereas promotion-related comments showed present- and future-oriented behaviors, indicating a stronger future orientation. Our findings highlight the need for design, governance, and patient-safety frameworks that account for motivational heterogeneity in GenAI use, particularly as individuals increasingly rely on GenAI tools for health-related guidance.