Self-efficacy and instructor support as determinants of ChatGPT effectiveness in student learning

The rapid integration of generative artificial intelligence (AI) tools, such as ChatGPT, is reshaping learning experiences in higher education. However, the psychological and instructional conditions under which these tools enhance academic success remain insufficiently understood, particularly in low resource contexts. Using survey data from 286 university students in Somalia, this study investigates how ChatGPT usage influences students’ academic performance and learning outcomes, with particular attention to the mediating roles of academic self-efficacy and cognitive engagement (mental effort), and the moderating role of instructor support. Drawing on the Technology Acceptance Model, Social Cognitive Theory, and Cognitive Load Theory, eight hypotheses were tested using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results show that ChatGPT usage significantly enhances learning outcomes, self-efficacy, and cognitive engagement; however, its direct effect on academic performance is not statistically significant, suggesting that performance gains do not occur automatically from AI use. Mediation analysis reveals that self-efficacy significantly mediates the relationship between ChatGPT usage and academic performance, whereas cognitive engagement does not. Furthermore, instructor support significantly moderates the relationship between cognitive engagement and academic performance, strengthening the translation of mental effort into achievement. Overall, the findings highlight the critical role of psychological empowerment and instructional support in unlocking the educational benefits of generative AI, particularly in resource constrained higher education contexts.