ChatGPT and Academic Writing Self-Efficacy: Unveiling Correlations and Technological Dependency among Postgraduate Students

Abstract

This study investigates the impact of ChatGPT usage on academic writing self-efficacy among 148 postgraduate students from Dhar EL Mahraz in Fez and Moulay Smail in Meknes, Morocco. It aims to illuminate the dynamics between ChatGPT usage and self-efficacy in academic writing. The study employed statistical analyses such as independent-sample t-tests and ordinal logistic regressions within a cross-sectional framework to explore the dynamics between ChatGPT use, reliance, and self-efficacy in scholarly writing. Results reveal that ChatGPT users reported higher self-efficacy scores (M=3.96, SD=0.61) compared to non-users (M=3.44, SD=1.38), suggesting the potential of ChatGPT in enhancing writing self-efficacy by providing continuous feedback and fostering a sense of accomplishment. Further analyses indicate that writing self-efficacy significantly predicts the duration of ChatGPT usage (Estimate = .692, p = .011), suggesting a positive correlation between self-efficacy and prolonged ChatGPT engagement. Furthermore, the frequency of ChatGPT use was found to predict reliance on the tool (β = .619, p < .001), underscoring a habituation effect. These findings contribute to the growing discourse on the role of AI in academic writing, emphasizing the benefits of ChatGPT in fostering writing self-efficacy while cautioning against over-reliance. Recommendations include longitudinal studies to assess the evolution of writing self-efficacy with continued ChatGPT use and broader research encompassing various AI tools and academic levels. This study sheds light on the advantages of AI in education and draws attention to the need for a balanced integration of technology in academic practices.