Trust in AI vs. human doctors: The roles of subjective understanding, perceived epistemic authority and social proof
Xiaotong Ding, Cai Xing
Abstract
Artificial intelligence (AI) is emerging as a valuable assistant in healthcare and can help alleviate the burden on overstretched medical systems. Yet, despite comparable or superior performance, AI doctors are often trusted less than their human counterparts. To identify the mechanisms underlying this trust disparity, we conducted four studies ( N = 1040). Study 1 utilized self-report measures to provide initial correlational evidence for both mediations. Studies 2a and 2b employed experimental designs to establish causal mediating effects. Study 3 explored a potential boundary condition for the mediation. The results revealed that, subjective understanding and perceived epistemic authority mediated the effect of doctor type (AI vs. human) on trust. Specifically, compared to human doctors, individuals experienced greater subjective difficulty in understanding the diagnostic process of AI, and they perceived AI doctors as possessing lower epistemic authority. These ultimately led to the lower levels of trust in AI doctors. Moreover, social proof moderated the mediating effect of subjective understanding but not that of perceived epistemic authority. These findings support and enrich the three-layered model of trust in automation, and offer practical guidance for the development of trustworthy AI doctors. • Individuals show lower trust in AI doctors compared to human doctors. • Subjective understanding mediates the effect of doctor type (AI vs. human) on trust. • Perceived epistemic authority mediates the effect of doctor type (AI vs. human) on trust. • Social proof moderates the mediating effect of subjective understanding. • The research adopts measurement-of-mediation and manipulation-of-mediator designs.