When ChatGPT Speaks About Health: Examining Perceptions of Warmth and Competence Toward AI as a Health Information Source

Recent research indicates that a majority of individuals (78.4%) are willing to use generative AI tools such as ChatGPT for health-related purposes. However, how individuals perceive and evaluate AI as a source of health information remains less understood. Drawing on literature from AI aversion and the stereotype content model, this study compares perceptions of warmth and competence between ChatGPT and human doctors. Results indicate that when delivering the same health information on moderate drinking, ChatGPT is perceived as less warm and competent than human doctors. This perception further leads to decreased intentions to share the information. Additionally, we examine the moderating effect of prior AI exposure, finding that discrepancies in perceived warmth and competence disappear for those with extensive AI exposure. Individuals with substantial AI experience perceive similar levels of warmth and competence and are equally likely to share health information from ChatGPT as from a human doctor.