Psychological Ownership as a Boundary Condition in AI Recommendation: Credibility Formation and Behavioral Intentions in Electronic Commerce

Abstract

AI recommendation agents increasingly mediate consumer decision-making in electronic commerce, yet algorithm-based agents often suffer credibility deficits relative to human sources. This research examines how recommendation agent type (human vs. AI) influences behavioral intentions through perceived credibility and how psychological ownership moderates this process. Across two controlled online experiments, consumers evaluated recommendations delivered by human or AI agents. Study 1 shows that baseline AI agents are perceived as less credible than human agents, while AI agents receiving minimal user involvement (naming) exhibit partially improved credibility, and that credibility mediates the effects of agent type on intention to use the system, recommendation acceptance, and purchase intention. Study 2 introduces a stronger psychological ownership manipulation through AI agent customization. Results indicate that customization strengthens psychological ownership, which reduces the credibility gap between AI and human agents and, when ownership is high, even allows AI agents to be evaluated as more credible. Conditional process analyses confirm that psychological ownership moderates both the effect of agent type on credibility and the indirect effects on behavioral intentions. Overall, the findings demonstrate that credibility toward AI recommendation agents is dynamically shaped by user–agent relational experiences. By integrating algorithm aversion, source credibility, and psychological ownership perspectives, this research advances understanding of consumer–AI interaction and provides design insights for AI-enabled recommendation systems in electronic commerce.