Asian Research Journal of Mathematics

This narrative review synthesises 20 studies on artificial intelligence (AI)-supported calculus instruction, focusing on the relationship between teacher preparation and student outcomes within U.S. STEM pipelines. The evidence suggests that AI tools, including intelligent tutoring systems, generative chatbots, and adaptive platforms, may support student engagement and, under guided instructional conditions, conceptual understanding and problem-solving. However, the findings indicate that these benefits are conditional on teacher guidance, pedagogical alignment, and active student use. The literature remains fragmented: studies of AI tools rarely examine teacher preparation, while studies of teacher readiness seldom measure student calculus outcomes. This review integrates these strands and proposes a cross-level framework linking AI design, teacher preparation, instructional implementation, and student achievement. It also identifies major gaps, including the predominance of small-scale studies, limited longitudinal evidence, limited attention to equity, and the absence of direct research on U.S. STEM pipeline outcomes. The proposed framework offers practical guidance for teacher preparation programmes and institutional policies. It suggests that professional development should focus on helping teachers interpret AI-generated feedback, design effective prompts, and align AI tools with conceptual learning goals. These findings imply that AI-supported calculus instruction should be treated as a coordinated instructional system rather than a technology add-on if it is to contribute meaningfully to the U.S. STEM pipeline.