Do Users Really Value Human Warmth and AI Skill? Experiments Reveal a Persistent Preference for Human Nutritionists

Conventional wisdom—and much of the trust literature—assumes that consumers prize benevolence in human experts but competence in AI. We put this stereotype to a stringent test through two 2 × 2 between‑subjects experiments (Study 1: N = 328; Study 2: N = 329) that manipulated benevolence (high vs. low) and competence (high vs. low) in health advertisements featuring either a human or an AI nutritionist. PROCESS Models 4 and 7 evaluated a dual‑mediation pathway in which perceived health benefits and trust linked experimental cues to adoption (purchase) intention. Results overturned the stereotype. Across all cue combinations, ads featuring human nutritionists outperformed their AI counterparts on perceived benefits, trust, and adoption intention. The human advantage widened under low‑benevolence or low‑competence conditions and narrowed—yet persisted—when both cues were high. Mediation analyses confirmed that source effects operated sequentially through perceived benefits and trust. These findings expose the limits of competence signaling for overcoming algorithmic aversion in healthcare contexts. To approach parity with human advisers, AI nutritionists must demonstrate not only technical skill but also convincing cues of benevolence, suggesting new design imperatives for AI‑enabled health communication.