Factors influencing learners learning interest when using AI chatbots-assisted math leaning in higher education

Introduction This study investigates the determinants of undergraduate learners’ interest in AI-supported mathematics education. Drawing upon the Unified Theory of Acceptance and Use of Technology (UTAUT) and Social Cognitive Theory, a structural model was proposed and empirically tested, incorporating mathematics anxiety, self-efficacy, performance expectancy, and effort expectancy. Methods Data were collected from 247 Chinese undergraduates and analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). Results The findings indicate that performance expectancy, effort expectancy, and self-efficacy are significant predictors of learning interest, with effort expectancy exerting the strongest influence. Self-efficacy also indirectly affects learning interest through its influence on perceptions of the system’s usefulness and ease of use. Contrary to prior research, mathematics anxiety did not significantly predict either self-efficacy or learning interest, suggesting that AI-facilitated environments may buffer negative emotional effects. Academic major moderated the relationship between mathematics anxiety and learning interest, reflecting disciplinary differences in motivational dynamics. Discussion This research contributes to theory by integrating motivational constructs into technology acceptance models and extending AI applications to cognitively demanding domains. Practical implications include prioritizing user-centered design and targeted self-efficacy interventions to enhance learner engagement.