From threat to tool: AI self-efficacy and user profiles in higher education

Abstract

Generative artificial intelligence (GenAI) has brought both challenges and opportunities for teachers in higher education. This study does not focus on restrictive methods; instead, it explores how educators can actively promote academic integrity. We contend that AI self-efficacy, characterized as a student’s confidence in their capacity to employ AI ethically and efficiently, constitutes a substantial determinant. This paper delineates significant findings derived from a survey administered by 177 university students. A moderation analysis indicates that the adverse correlation between AI usage frequency and academic integrity is markedly diminished among students exhibiting high AI self-efficacy. Moreover, a cluster analysis clearly delineates three distinct AI user profiles: Confident and Cautious Users, Pragmatic High Users, and Dependent Users. Demographic studies reveal a significant correlation between these profiles and the students’ academic disciplines. The results suggest that the focus of instruction should transition from a “one-size-fits-all” approach to tailored interventions designed to enhance student empowerment. This article outlines practical implications and customized strategies for each student profile.

https://doi.org/10.34105/j.kmel.2026.18.017