Authority Over Sympathy: How AI Emotional Tone and Thinking Dispositions Shape Hallucination Detection in Health Communication

Drawing on the Misinformation Recognition and Response Model (MRRM), this study investigates how AI emotional tone and thinking dispositions influence hallucination detection in health communication. A Wizard-of-Oz experiment (N = 476) examined the role of authoritative and sympathetic tones as independently manipulated tone in AI-generated health responses containing embedded hallucinations. The results showed that authoritative tone significantly reduced participants’ hallucination detection sensitivity. While sympathetic tone did not show a significant effect. The analysis also indicated that thinking dispositions did not significantly moderate the influence of tone on hallucination detection sensitivity.