2026-07-16 ペンシルベニア州立大学（Penn State）
＜関連情報＞
- https://www.psu.edu/news/research/story/ai-disagreement-may-shake-patient-trust-doctors
- https://www.sciencedirect.com/science/article/abs/pii/S1071581926000996
AIが異論を唱えるとき：セカンドオピニオンが患者の医師への信頼に与える影響 When AI Disagrees: The Effect of Second Opinion on Patients’ Trust in Doctors
Cheng Chen, Yuan Sun, Mengqi Liao, S. Shyam Sundar
International Journal of Human-Computer Studies Available online: 21 April 2026
DOI:https://doi.org/10.1016/j.ijhcs.2026.103824
Highlights
…AI agreement increased perceived recommendation credibility
…AI disagreement increases perceived doctor laziness
…AI disagreement increases perceived medical uncertainty
…Perceived human-likeness of the doctor matters
Abstract
As we increasingly turn to AI tools for second opinions, how might that affect our trust in doctors? We study this in the context of mental health consultation and medical advice, by examining how agreement or disagreement by an AI system influences patients’s trust in their doctors. We conducted an experiment (N=135) in which participants interacted with a Large Language Model (LLM)-simulated doctor during a stress management therapy session. At the end of the consultation, the doctor offered to consult an AI assistant for a second opinion. Results showed that AI agreement enhanced perceived recommendation credibility, whereas AI disagreement increased perceptions of medical uncertainty and doctor laziness. Since patients may attribute varying levels of anthropomorphism to an LLM-simulated doctor, when the doctor was perceived as more human-like, AI disagreement increased perceived medical uncertainty and doctor laziness. In contrast, AI agreement enhanced perceived recommendation credibility. These changes in perceptions further influenced patients’ cognitive, affective, and behavioral trust in the doctor. We discuss these findings to advance theoretical understanding of agency negotiation within doctor-patient-AI interactions, offer practical suggestions for how doctors can better communicate AI disagreement and agreement to patients, and provide methodological insights for leveraging LLMs for experimental design and message manipulation. Limitations of using LLMs to impersonate human professionals, as well as the limited generalizability of the findings derived from LLM-generated role-play, are also discussed.
