Research Information System
International Journal of Learning, Teaching and Educational Research, vol.25, no.3, pp.463-493, 2026 (Scopus)
The rapid growth of artificial intelligence (AI) in mathematics education places new demands on teachers’ digital-pedagogical competence and their ability to use AI critically. Although research highlights both the potential and risks of AI tools, there is limited evidence on how structured instructional models help future teachers move beyond uncritical reliance on AI-generated explanations. This study examined a two-rotation AI-integrated blended learning model implemented with 94 pre-service mathematics teachers in Kazakhstan. A quasi-experimental pre–post design with a control group was utilised. The experimental group (n = 47) completed an eight-week intervention consisting of guided AI-supported construction (visualisation, modelling, micro-teaching) followed by a systematic analysis of AI outputs (error identification, human–AI comparison, modelling with real data). Quantitative analyses (t-tests, ANCOVA) and thematic analysis of reflections and artefacts revealed significant improvements in the experimental group. Digital pedagogy showed a large effect (d = 0.93), while professional skills and AI readiness demonstrated medium effects (d = 0.56). Qualitative findings indicate a shift from initial acceptance of AI outputs towards systematic verification and more informed pedagogical decision-making. Overall, the findings suggest that combining guided AI use with structured analytical reflection supports responsible and reflective AI integration in mathematics teacher education. The findings also suggest that teacher education programmes should systematically incorporate structured AI-integrated models that combine guided application with critical analysis to foster reflective, responsible, and pedagogically sound AI use.