Abstract
<jats:p>This article presents an approach to automated assessment of school mathematics assessment based on the integration of ChatGPT with the Ukrainian digital educational application Mriia. The proposed approach addresses the growing need for transparent, reproducible, and scalable pedagogical diagnostic tools capable of evaluating not only formal learning outcomes, but also the procedural logic of mathematical reasoning, solution structures, and the development of students’ mathematical competencies. The approach is implemented and empirically evaluated using the MRIIA+ prototype, a digital educational platform that enables the automated collection, analysis, and interpretation of educational data within the Mriia application. The institutionalisation of this application has created new opportunities for integrating artificial intelligence tools into formal educational processes. Within the MRIIA+ environment, the integration of ChatGPT and Mriia components supports a systematic three-way information exchange between students, teachers, and parents, as well as the intellectual analysis of educational content. Empirical analysis indicates that the effects of AI-assisted assessment are selective. For the error correction indicator, the automated approach demonstrated statistically significant advantages (Mann-Whitney U, p_BH < 0.001, r ≈ 0.4), whereas for feedback readability and learning engagement, traditional assessment methods remained more effective (Cohen’s d ≈ 2.0). A key feature of the proposed approach is its ability to automatically detect recurring student errors and analyse the structure of mathematical solutions, including handwritten work, while providing personalised feedback in a way that reduces teachers’ time expenditure without compromising pedagogical quality. The MRIIA+ prototype proved to be effective in online and blended learning contexts. At the same time, the results indicate that automated assessment performs optimally within a defined homeostatic plateau, representing a balanced interaction between AI-based analysis and the teacher’s pedagogical interpretation. This balance enhances assessment objectivity, supports regular learning activities, increases student motivation, and assists teachers in evidence-based decision-making in secondary school mathematics education.</jats:p>