AI Self-Efficacy, Motivation, and Engagement Among STEM Learners: Mediating and Moderating Pathways to Perceived Mathematics Performance in Ghanaian Higher Education

Abstract

Performance in mathematics (PM) has become a progressively influential determinant for learners to navigate in various STEM (science, technology, engineering, and mathematics) disciplines. Mathematics serves as the bedrock for science, art, technology, and engineering. For learners to build strong confidence and interest in these fields, the demand for highly acceptable mathematics perceived mathematics performance cannot be underestimated. This research explores students’ perceived mathematical performance through artificial intelligence (AI) self-efficacy and examines the interplay between motivation and engagement in AI usage and its impact on perceived mathematics performance. Sampling 528 students from a STEM-based program in Ghana, this study utilized “covariance-based structural equation modeling” (CB-SEM) to examine students’ AI self-efficacy, AI motivation, AI engagement, and perceived mathematics performance. The study’s results depicted that AI self-efficacy positively affects students’ motivation to use AI and significantly influences AI engagement, which subsequently impacts their perceived mathematics performance. Furthermore, students’ motivation for AI usage and their engagement with the tool mediate the link between AI self-efficacy and perceived performance. Both engagement and motivation in AI moderate the relationship between AI self-efficacy and performance. This investigation provides empirical evidence on how AI, if adapted ethically, might enhance students’ mathematics performance and improve the quality of mathematics concept retention through self-regulated learning or personalization. The results therefore offer insights into learner attitudes and behavioral intentions, while pointing to the need for future research connecting AI use with objective achievement data. For educational practice, the findings suggest that fostering AI self-efficacy through targeted training can create cascading benefits for motivation, engagement, and ultimately, perceived mathematics performance among STEM students.