Impact of Information Quality Perception of Generative AI on Critical Thinking: The Mediating Roles of Recommendation Behavior and Self-Efficacy

This study examines how users' perceptions of generative AI information qualityspecifically accuracy, reliability, and usefulness-affect their critical thinking, with a focus on the mediating roles of self-efficacy and recommendation behavior.A survey of 151 generative AI users was conducted, and the data were analyzed using structural equation modeling.The results revealed a significant direct negative effect of perceived information quality on critical thinking (B = -0.21,z = -1.98,p = .048),suggesting that higher trust in AI-generated content may reduce critical engagement and reflective analysis.However, self-efficacy significantly mediated this relationship (B = 0.17, z = 2.44, p = .015),indicating that users who are confident in their ability to use AI tools effectively are more likely to think critically about AI-generated information.Furthermore, a dual mediation analysis demonstrated that both self-efficacy and recommendation behavior jointly foster critical thinking (total indirect effect B = 0.30), even though the total effect of information quality perception on critical thinking was not statistically significant (B = 0.09, z = 1.25, p = .21).These findings suggest that while high perceived information quality may discourage critical thinking, users' self-efficacy and social engagement can offset this negative effect.The study provides valuable implications for AI literacy education and the design of AI systems that encourage thoughtful engagement.