This study proposes the Machines-In-Society (MIS) model to explain how people trust AI in human social roles. Trust arises from perceived social agency, shaped by human-like cues and intellectual competence. Study 1 used path analysis to confirm the model’s fit in general contexts, while Study 2 experimentally validated it with AI assistants. Findings show that AI perceived as socially and intellectually capable gains more trust in performing human roles. The research identifies core psychological mechanisms behind trust in socially functioning AI and provides practical insights for designing trustworthy AI systems within societal settings.
Hong et al. (Fri,) studied this question.