Generative Artificial Intelligence (GenAI) provides personalised explanations, immediate feedback, and interactive
problem-solving support; however, evidence of its effectiveness in strengthening mathematical reasoning in
developing educational contexts remains limited. This study developed an AI-Assisted Mathematics Learning Model
and illustrated its evaluation through a quasi-experimental mixed-methods design involving 120 secondary school
learners and 10 mathematics teachers. Learners were divided equally between a GenAI-assisted experimental group
and a conventional-instruction control group. Data were obtained through mathematical reasoning tests,
questionnaires, classroom observations, and semi-structured interviews. In the analysis, the experimental group
achieved a significantly higher adjusted post-test mean than the control group, F(1,117) = 180.96, p < .001, partial η2
= .607. Synthetic qualitative evidence illustrated potential benefits related to personalised explanations, immediate
feedback, strategy comparison, and reflective problem-solving, while highlighting concerns about inaccurate
responses, learner dependence, technological access, and teacher mediation. The proposed model provides a
structured framework for investigating the responsible integration of GenAI into Namibian secondary school
mathematics education.

Keywords: Generative Artificial Intelligence; mathematics education; mathematical reasoning; secondary school
learners.