Artificial intelligence in mathematics education: a two-dimensional conceptual framework and meta-review

Source abstract

Abstract Artificial Intelligence (AI) is transforming mathematics education just as it is reshaping many other domains of modern life, creating new opportunities that fundamentally change the way students and teachers work. AI has become an increasingly prominent topic in mathematics education, generating a growing body of systematic reviews. However, the review-level literature remains fragmented across different AI paradigms, system types, and methodological approaches. This meta-review synthesizes existing systematic reviews on AI in mathematics education and provides a conceptually oriented interpretation of how AI is shaping mathematics education. We specifically identify and classify the types of AI present in mathematics education as well as the educational functions it fulfills. To organize the synthesis, we propose a conceptual framework that distinguishes between a technological dimension of AI, referring to major AI paradigms and system types, and a functional dimension, referring to the educational roles AI can serve in mathematics education. Following predefined eligibility criteria, 13 systematic review studies were included. The findings show that review-level research on AI in mathematics education is recent and rapidly expanding. Our review maps several AI technologies and systems, including symbolic AI, machine learning, and generative AI. In functional terms, AI was most often conceptualized as adaptive infrastructure, learning partner, teaching tool, and research tool, with a smaller but emerging role of AI as learning content itself. This meta-review provides a structured synthesis of how AI is represented in mathematics education review literature and offers a conceptual basis for future research and responsible AI integration in the field.