AI-Driven Support for Teaching Mathematical Concepts in Engineering Subjects to Academically At-Risk Students

Abstract

Mathematical competence is essential for engineering subjects such as circuit analysis, signal processing, fluid mechanics, heat transfer, control systems, and structural analysis. However, some students require additional instructional time, repeated practice, and multiple representations before they can confidently connect mathematical procedures with engineering meaning. This paper presents an evidence-based narrative review of artificial intelligence (AI) support for teaching mathematical components of engineering subjects to such learners. The paper contributes a focused synthesis of AI-supported mathematics learning in engineering contexts and proposes a human-in-the-loop framework with neuro-symbolic verification and scaffold fading. The manuscript does not report empirical classroom results; instead, it integrates literature on intelligent tutoring systems, Bayesian and deep knowledge tracing, cognitive load theory, universal design for learning, learning analytics, and generative AI. The review indicates that AI may support prerequisite diagnosis, adaptive sequencing, worked-example practice, formative feedback, and teacher alerts when used as a supervised support layer rather than as a replacement for teachers. Key safeguards include mathematical verification, learner privacy, avoidance of permanent deficit labels, culturally inclusive examples, and gradual reduction of AI support to promote independent problem solving. The proposed framework and implementation plan are intended for the future empirical validation in engineering education.