Does AI communication strategy matter for health persuasion? Comparing three archetypes in visual health education

Introduction Artificial intelligence (AI) systems are increasingly explored as tools for visual health education, yet whether their communication style influences persuasive outcomes remains unclear. This study compared three AI communication strategy archetypes to test whether emotional or cognitive pathways more strongly drive behavioral intention. Methods In a pre-post quasi-experimental design, 367 Chinese university students (84.5% with myopia) were assigned to interact with one of three AI health assistants embodying distinct communication strategy archetypes: an Empathetic Partner (warm, humorous, gain-framed), an Authoritative Expert (professional, serious, loss-framed), or an Objective Informant (neutral, data-driven, balanced-framed). Outcomes were assessed using the HEAT (Health Engagement with AI Technology) framework across three dimensions: Information Processing, Relational Engagement, and Action Empowerment. Results All three archetypes produced substantial pre-to-post knowledge gains (all d z 1.0). However, they diverged on relational and motivational outcomes. The Empathetic Partner elicited significantly stronger perceived relationship quality ( M = 4.47 vs. 4.09 and 4.15, p 0.001, d = 0.68), higher confidence in acquired knowledge ( p = 0.001), and greater behavioral intention ( p = 0.011) than the other two archetypes. Hierarchical regression controlling for baseline scores revealed that AI relationship quality was the strongest independent predictor of behavioral intention (β = 0.346, p 0.001), whereas knowledge acquisition contributed no significant additional variance (β = −0.024, p = 0.557). Discussion When identical health content is delivered through different AI communication strategies, the quality of the perceived relationship, rather than the quantity of information transferred, most strongly determines persuasive effectiveness among university students. These findings suggest that AI health assistants targeting young adult populations should prioritize empathetic, relationally warm communication.