AI integration in mathematics education: A systematic review of implementation challenges and pedagogical implications (2017-2025)

The integration of artificial intelligence (AI) in mathematics education has expanded rapidly, offering both transformative opportunities and critical challenges. The growing volume and diversity of research in this field have also made it increasingly difficult to synthesize findings. This paper examines 54 studies published between 2017 and 2025 to identify the key challenges and pedagogical approaches associated with AI use in mathematics education, using the Scopus database as the primary source and following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The findings identify major challenges, including the conflict between AI-generated answers and deep learning, overreliance on AI that reduces cognitive effort, concerns about assessment validity and academic integrity, and limitations in technological–pedagogical integration and teacher readiness. Additional issues include constraints in supporting deep mathematical reasoning, equity and access disparities, ethical risks, and the reduction of mathematical discourse and collaboration. Despite these challenges, emerging pedagogical approaches demonstrate how AI can be used to enhance learning. These include AI-supported conceptual understanding, inquiry-driven and problem-centered pedagogy, adaptive and personalized learning pathways, teacher-mediated AI integration, and the expansion of mathematical practices through computational and multimodal engagement. Building on these findings, the study proposes a conceptual framework structured around three interrelated dimensions: cognitive–epistemic, instructional–interactional, and systemic–ethical, capturing how AI influences knowledge construction, instructional processes, and broader educational contexts. Furthermore, a structural model is proposed to illustrate the relationships between challenges and pedagogical approaches.