Between epistemic empowerment and moral anxiety: Chinese patients’ ambivalence toward AI-assisted diagnosis

The rapid advancement of generative artificial intelligence (AI) has led an increasing number of Chinese patients to incorporate AI tools into their clinical encounters. While prior research has explored individual attitudes toward AI-mediated diagnosis, less is known about the broader socio-psychological dynamics that shape patient engagement with AI. Drawing on qualitative interviews with 60 Chinese stakeholders, this study examines patients' socio-psychological responses to AI-assisted diagnosis. Our findings reveal a persistent tension. On the one hand, patients experience epistemic empowerment. AI renders complex biomedical knowledge accessible, enables systematic verification of physicians’ recommendations, and supports more substantive participation in clinical decision-making. On the other hand, patients report moral unease grounded in Confucian norms. Many fear that invoking AI may appear self-serving, disrupt hierarchical interactional structures, or implicitly challenge professional authority. To navigate this dissonance, patients employ covert strategies, including authorship obscuration, third party endorsement, and discursive translation to preserve interpersonal harmony. While these strategies reduce immediate interactional friction, they can inadvertently undermine long-term trust by fostering implicit comparisons between clinicians and AI. As AI systems become increasingly accurate, comprehensive, and contextually adaptive, some patients may gradually transfer epistemic trust from clinicians to algorithmic judgments. In response, we advocate a clinician-stewarded, evidence-based integration of AI insights, supported by institutional protocols and governance structures that balance professional authority, workload recognition, and patient empowerment.