Dependency on Generative AI as predictor of Self-Efficacy in Writing and Defending Academic Texts Among University Students

Abstract

The current study investigated the relationship between the dependency on Generative Artificial Intelligence (GenAI) and self-efficacy in writing and defending academic texts among university students. The sample for this cross-sectional research design was university students across various degree levels. To measure the study variables Self-Efficacy for Writing and Defending Academic Texts Scale and the Generative AI Dependency Scale were used. The results revealed a significant negative relationship between overall GenAI dependency and total academic text self-efficacy. Specifically, the withdrawal dimension of AI dependency exhibited inverse associations with both writing self-efficacy and oral defense self-efficacy. Comparative analyses revealed that male students reported higher writing self-efficacy than female students. Moreover, significant differences were observed across academic qualification levels, with postgraduate students demonstrating higher self-efficacy in academic writing compared to undergraduate students. The results indicate that although cognitive involvement with AI has no direct negative impact on writing confidence, behavioral reliance and withdrawal symptoms share a strong negative relationship with students' confidence in drafting and defending scholarly work and orally justifying scholarly work. Consequently, higher education institutions must balance technological assistance with interventions that safeguard students' independent academic competence and defense readiness.