Asian Journal of Education and Social Studies

Artificial intelligence (AI) is increasingly used to support mathematics teaching, yet research examining its instructional effects and implementation conditions in primary and secondary schools remains fragmented. This study reviewed published research on AI-assisted mathematics teaching between 2022 and 2026. Relevant records were retrieved from the Web of Science database using topic searches for AI and mathematics education. Of 262 initially identified records, 44 studies met the stated inclusion criteria and were included in a thematic synthesis. Three principal research areas emerged: instructional effects, factors influencing effectiveness, and strategies for improvement. The reviewed studies generally reported favourable outcomes for students’ mathematics achievement, conceptual understanding, interest, motivation and classroom engagement, although some interventions produced limited or inconsistent benefits. AI tools also supported teachers’ lesson planning, resource preparation, assessment, learning diagnosis and instructional decision-making. Reported outcomes depended on teacher-related factors, including AI literacy, attitudes and pedagogical competencies; student-related factors, including self-regulation, digital literacy and interaction patterns; technological factors, particularly accuracy, usability and personalisation; and contextual factors, including infrastructure, family support and educational resources. Suggested improvements included professional development, teacher-led implementation, clearer ethical and data-governance arrangements, and the development of more reliable mathematics-specific tools. The findings indicate that AI can provide useful instructional support, but its effectiveness depends on appropriate educational integration and local conditions. Longitudinal classroom research is needed to clarify sustained outcomes and the interaction of these factors.