Artificial Intelligence in Mathematics Education: A Systematic Review of Educational Evidence and Responsible Integration

Artificial intelligence has become an important component of contemporary education, creating new opportunities for teaching, learning, assessment, and educational decision-making while raising significant pedagogical and ethical challenges. This study systematically reviews contemporary empirical evidence on the application of artificial intelligence in mathematics education to identify the conditions associated with effective implementation, synthesise reported educational benefits and limitations, and determine the pedagogical principles supporting its responsible integration into mathematics teaching and learning. The review followed internationally recognised systematic review standards, with methodological quality assessed using an established appraisal framework and findings synthesised through narrative evidence synthesis. The study demonstrates that the educational value of artificial intelligence depends primarily on pedagogical implementation rather than technological capability alone. The findings indicate that artificial intelligence is most effective when it supports mathematical reasoning, conceptual understanding, formative assessment, differentiated instruction, and personalised learning without replacing learners' active cognitive engagement. Successful implementation depends on pedagogically sound instructional design, teacher guidance, artificial intelligence literacy, ethical awareness, adaptive technological support, and clearly defined educational objectives. The findings provide evidence-based guidance for teachers, teacher educators, curriculum developers, educational institutions, and researchers while identifying priorities for future empirical research. Overall, artificial intelligence should be understood as a pedagogical resource whose educational effectiveness depends primarily on the quality of instructional implementation and educational context.