Help me, Doctor AI? A cross-national experiment on the effects of disease threat and stigma on AI health information-seeking intentions

Abstract

Generative AI chatbots are emerging as novel sources for health information. Adopting a cross-national perspective, this study examines how disease-related factors—namely, disease threat and stigma—influence both individuals' intentions to seek health information via generative AI and their preferences for AI compared to traditional interpersonal sources like doctors and peers. In a preregistered 2x2 online experiment, participants from Austria, Denmark, France, and Serbia (N_total = 1951) encountered written scenarios about their health that manipulated disease threat (low vs. high) and stigma (low vs. high). The sample was stratified to ensure representativeness for age, gender, and educational level across the countries studied. Results showed no main effect of disease threat on AI information-seeking intentions, but stigma significantly influenced preferences, particularly in mild health conditions. Participants were more likely to consult AI over peers for stigmatized conditions, highlighting the role of AI's anonymous interface in reducing social judgment. Country differences further revealed that national contexts also shape AI adoption: while participants in Denmark and France showed a stronger preference for AI over peers, those in Serbia and Austria preferred peers over AI. Additionally, AI trust and literacy emerged as the strongest predictors of both AI usage intentions and preferences. These findings indicate that gen AI tools can play a complementary role in the health information ecosystem, particularly for stigmatized conditions and in contexts where traditional sources are perceived as less accessible or judgment-free.