AI-Driven Mathematics Education: Exploring Integrated Pathways from a STEM Education Perspective

Abstract

Artificial intelligence (AI) is transforming mathematics education by functioning as a cognitive mediator in STEM learning. This study proposes a three-layer model—diagnostic, regulatory, and generative—to integrate AI into teaching. The diagnostic layer profiles learners through knowledge tracing and error analysis; the regulatory layer offers real-time adaptive support; and the generative layer fosters conceptual transfer and higher-order reasoning through simulations and AI-generated tasks. A holistic framework incorporating cognitive, emotional, and social dimensions supports both self-regulated and co-regulated learning. Using a mixed-methods design, the study examines AI’s impact through experiments, learning trajectory analysis, and classroom observations, supported by a multi-level evaluation system.