Academic stress and university students’ dependency on generative artificial intelligence: a multiple mediation model using PLS-SEM

BACKGROUND: As generative artificial intelligence (GAI) becomes increasingly embedded in academic work, concerns have emerged that students experiencing higher levels of stress may be more prone to developing reliance on these systems. This study examines the relationship between academic stress and GAI dependency and explores whether performance expectations, academic anxiety, and desire thinking are involved as mediating psychological correlates. METHODS: Guided by the Interaction of Person–Affect–Cognition–Execution (I-PACE) model, a survey was conducted among university students to assess academic stress, GAI dependency, and three proposed mediating factors. Partial least squares structural equation modeling (PLS-SEM) was employed to examine both direct and indirect relationships among these variables. Significance of path coefficients and effect sizes was evaluated using established criteria to determine the strength and relevance of each association. RESULTS: Academic stress was positively and significantly associated with GAI dependency. Performance expectations, academic anxiety, and desire thinking all functioned as meaningful mediators in this relationship. Among them, performance expectations accounted for the largest share of the indirect effect, while desire thinking and academic anxiety also made substantial contributions. These mediators explained a considerable proportion of the overall impact of academic stress on GAI dependency. CONCLUSIONS: The findings suggest that academic stress is associated with GAI dependency through multiple cognitive and emotional variables. By applying the I-PACE model to the context of GAI use, this study clarifies how stress-related factors link academic stress to students’ engagement with emerging technologies.