不同信任主体下基于经验迁移的信任建立:人与人工智能的比较OACHSSCD

As interactions between humans and artificial intelligence(AI)evolve from instrumental tool use toward novel forms of social relationships,the scope of trust agents has expanded beyond humans to include AI.This shift broadens traditional human-machine trust into a bidirectional construct of human-AI mutual trust,while also introducing trust from AI toward humans and trust among AI agents.However,existing research rarely integrates theoretical models from both interpersonal trust and human-machine trust.The underlying mechanisms of trust formation remain unclear,particularly due to limited attention to the role of prior knowledge-leading to contradictory conclusions in the literature.Integrating social and engineering psychological perspectives and grounded in a dynamic human-AI mutual trust framework,the present study proposes experience-based transfer as a core mechanism of trust formation.We examine three questions:(1)How do different trust agents-humans and AI-use learned experiences to shape initial and ongoing trust?(2)Can these experiences be transferred to new trust targets and novel contexts?(3)How are experience learning and transfer moderated by individual characteristics and interaction-process features?Using a new experimental paradigm incorporating AI agents as active participants,the study systematically investigates the fundamental processes of trust formation and updating.The findings contribute a dual-subject model of human-AI mutual trust and offer theoretical and empirical foundations for designing trustworthy AI and enhancing collaboration among multiple intelligent agents.