Background: Artificial intelligence (AI) has emerged as a transformative tool in education, offering personalized feedback, adaptive learning, and data-driven instructional strategies. While individual studies have explored its role in mathematics, findings remain fragmented. This study employs a meta-analysis to systematically evaluate the overall effect of AI applications on students’ mathematics performance. Methods: A comprehensive search was conducted across five databases (Google Scholar, Scopus, ERIC, Web of Science, and SCISPACE) for peer-reviewed quantitative studies published between 2017 and 2025. A total of 1,042 articles were identified, and after applying strict inclusion criteria guided by the PRISMA framework, 7 studies were retained. Data extraction followed a standardized protocol, and methodological quality was appraised using the Joanna Briggs Institute Critical Appraisal Checklist. Statistical analyses were performed with JASP, including effect size estimation, heterogeneity testing, and publication bias diagnostics. Results: The pooled effect size was g = 0.603 (p < 0.001, 95% CI = 0.393–0.813), indicating a moderate-to-large positive impact of AI on mathematics performance. Heterogeneity was substantial (Q(6) = 52.41, I² = 90.08%), suggesting variability across contexts and intervention types. However, fail-safe N analysis (N = 520.124) confirmed robustness against publication bias. Conclusion: Findings demonstrate that AI significantly enhances mathematics learning, supporting constructivist and socio-cultural theories of education where AI scaffolds learner self-efficacy and engagement. Policymakers and educators are encouraged to strategically integrate AI into curricula and teacher training to maximize its transformative potential in mathematics education.
Maanu et al. (Sat,) studied this question.