The chain mediating effect of academic anxiety and performance expectations between academic self-efficacy and generative AI reliance

Abstract

While Generative Artificial Intelligence (GenAI) tools have been used by an increasing number of students, many teachers and researchers recognized that excessive use of these tools may have adverse impacts on students’ academic development. However, the underlying mechanisms driving GenAI reliance and its consequences remain underexplored. This study aimed to address these gaps by investigating the relationships among academic self-efficacy, academic anxiety, performance expectations, and GenAI reliance, as well as examining the consequences of GenAI reliance and coping strategies to mitigate these issues. Data were collected from 337 university students in a Chinese university and analyzed using Pearson correlation, chain mediation effect, and thematic analysis. The results indicated no significant direct relationship between academic self-efficacy and GenAI reliance. However, academic anxiety and performance expectations were found to mediate this relationship. The most salient consequences of GenAI reliance were decreased independent thinking ability, reduced innovative thinking ability, restricted self-reflection, and limited critical thinking ability. Additionally, students’ self-initiated and expected teacher-initiated strategies to alleviate these consequences were recognized. This study contributes to the limited understanding of GenAI reliance and provides valuable insights for educators and academic institutions in directing students toward the appropriate and critical utilization of GenAI tools.