Between Efficacy, Goals and AI: A Mixed-Methods Study of Students’ Generative AI Reliance in an EMI University of Kazakhstan

Abstract

Despite the rapid integration of artificial intelligence into Kazakhstan’s higher education sector, the motivational mechanisms underlying students’ dependence on these tools remain underexplored, particularly the roles of academic self-efficacy and goal orientations. This study examines how these factors shape undergraduate students’ reliance on generative AI at an English-medium university in Kazakhstan. Specifically, it explores how different combinations of mastery and performance orientations, together with varying levels of self-efficacy, predict patterns of AI dependence and how students themselves interpret this reliance in their academic lives. A mixed-methods design was employed. Survey data were collected from 55 undergraduate students in the social sciences and humanities, and four semi-structured interviews were conducted to explore the meanings, motivations, and lived experiences underlying AI use. Guided by Goal Orientation Theory (Elliot & McGregor, 2001) and the Academic Self-Efficacy theory (Bandura, 1997), the findings show that academic self-efficacy is the strongest predictor of AI dependence. Students with lower self-efficacy reported significantly higher reliance on AI regardless of their goal orientation. Although the interaction effect between goal orientation and self-efficacy was not statistically significant, descriptive patterns aligned with the predicted hypotheses: mastery-oriented students with high self-efficacy showed the lowest dependence, whereas performance-oriented students with low self-efficacy showed the highest. Interviews further revealed that students use AI not only for efficiency, but also as emotional and linguistic support during moments of academic pressure and self-doubt. Overall, the findings suggest that AI dependence is shaped less by the technology itself and more by the motivational and psychological conditions within which learning occurs, and contextual factors like insufficient institutional support, highlighting the need for institutional initiatives to strengthen academic self-efficacy, foster critical and reflective AI use to prevent erosion of mastery experiences and skills, for instructors to design process-oriented assessments and for policymakers to address the long-term risks of AI reliance.