Critical AI literacy as a buffer against AI overtrust in students

Abstract Background The growing use of AI tools such as ChatGPT in academic settings has raised concerns about AI overtrust, yet its psychological correlates remain poorly understood. This study examined how academic self-efficacy, intrinsic motivation, and extrinsic motivation are associated with AI overtrust, and tested the sequential mediating roles of performance expectancy and AI literacy in these relationships. Methods A cross-sectional survey was conducted with 300 university students in South Korea who had prior experience using AI tools for academic purposes. The questionnaire measured academic self-efficacy, intrinsic and extrinsic motivation, AI literacy, and AI overtrust. Serial mediation analysis was conducted using the PROCESS macro [1] to test direct and indirect effects, including pathways sequentially mediated by performance expectancy and AI literacy. Indirect effects were evaluated using 5,000 bootstrap samples with bias-corrected 95% confidence intervals. Results Academic self-efficacy and intrinsic motivation were both positively associated with AI overtrust: learners who reported higher confidence and greater autonomy also reported higher overtrust in AI-generated information. Extrinsic motivation was numerically more strongly associated with AI overtrust than intrinsic motivation. AI literacy was not significantly correlated with AI overtrust at the bivariate level, but showed a significant negative indirect association within the mediation model. Sequential mediation analysis further indicated that, across all three psychological traits, higher performance expectancy combined with higher AI literacy was associated with lower AI overtrust. Conclusions These findings suggest that learners with higher academic self-efficacy and stronger intrinsic motivation also tend to report higher AI overtrust. Although greater academic self-efficacy may reflect more autonomous use of AI tools, it was nonetheless associated with higher overtrust in the present sample. AI literacy education tailored to learners’ psychological profiles may help support critical thinking, ethical awareness, and responsible AI use.