Student–GAI Interaction and Learning Outcomes: A Serial Mediation Model of Self-Efficacy and Cognitive Engagement

Abstract

This study investigates the impact of student interaction with Generative Artificial Intelligence (GAI) tools on academic achievement within the higher education sector of Pakistan. Utilizing a two-wave time-lagged research design, data were collected from 427 undergraduate students across public and private universities in the Islamabad Capital Territory. Structural equation and bootstrapped mediation analyses reveal that student–GAI interaction exerts a positive effect on academic performance (β = 0.154, p < 0.001). This relationship is significantly explained by a sequential mediation pathway: interactions with GAI cultivate academic self-efficacy, which subsequently enhances cognitive engagement, ultimately driving higher learning success. The final structural model accounts for 31.5% of the total variance in student cumulative Grade Point Average (CGPA). These insights indicate that the primary educational utility of GAI stems from its psychological and behavioral scaffolding rather than basic task automation. The paper concludes with actionable recommendations for institutional policymakers and educators to design pedagogical frameworks that foster critical AI literacy while bridging digital equity access gaps.