Understanding learning engagement in Indian higher education through epistemic awareness, perceived usefulness, and trust in generative AI

Abstract

Generative artificial intelligence (AI) is rapidly reshaping higher education by enabling the scalable generation of explanations, drafts, and summaries. However, its educational value remains dependent upon the student's ability to critically evaluate AI-generated content, calibrate trust, and regulate its use. Integrating andragogical learning theory, trust in automation, and technology acceptance frameworks, this study investigates the associations among epistemic awareness, perceived usefulness, trust in generative AI, AI use, and learning engagement. Furthermore, the study evaluates the proposed over-trust hypothesis by examining whether extremely high trust in generative AI is associated with reduced learning engagement. Utilising a cross-sectional survey of 207 undergraduate and postgraduate students in Indian higher education, a context notably underrepresented in existing literature, the study employed structural equation modelling (SEM) to assess the proposed measurement and structural models. Results indicated that perceived usefulness was positively associated with trust and behavioral integration in the proposed model. Conversely, Epistemic Awareness did not significantly predict trust, indicating that the hypothesised positive association was not supported in this sample. The hypothesised nonlinear over-trust effect was not supported, as the quadratic association between trust and learning engagement was not significant. These findings contribute to the growing literature on human-AI learning by suggesting that AI engagement may be understood as a regulated process rather than solely a technology adoption process.