AI-Enhanced Learning for Future Educators: How Perceived Usefulness and Ease of Use Shape Self Efficacy and Academic Performance

Abstract

While the Technology Acceptance Model (TAM) has been widely applied to predict AI adoption, its extension to AI self-efficacy as a mediating psychological construct linking technology perceptions to academic performance remains underexplored, particularly among Indonesian pre-service teachers. This study addresses this gap by employing a quantitative cross-sectional survey involving 305 Indonesian pre-service teachers. Multiple regression, hierarchical multiple regression, and nonparametric group-comparison tests were used to analyze the data. Findings showed that perceived usefulness and perceived ease of use significantly predicted AI self-efficacy. However, AI self-efficacy did not significantly predict GPA after controlling for academic year, AI usage frequency, subscription status, and university background, suggesting that broad semester-based GPA is insufficiently sensitive for capturing AI competence gains. No significant differences in AI self-efficacy were found across gender,i university location, or academic year. These findings extend TAM by positioning AI self-efficacy as the belief construct mediating how pre-service teachers engage with AI as a professional learning tool, and highlight the need for teacher education programs to embed AI use within core pedagogical tasks to develop students' critical and evaluative AI capabilities.