AI in the Classroom: Didactical Misalignments in Geometry Between Czech and Anglo-Saxon Contexts

Abstract

As artificial intelligence (AI) tools—such as chatbots and large language models—become increasingly accessible in educational settings, both teachers and students are relying on them more during the learning process. These tools provide various pedagogical benefits. However, their integration also introduces didactical risks, particularly when their outputs reflect implicit assumptions and educational paradigms that diverge from those in specific national curricula. This paper explores such risks in the context of lower secondary mathematics education (ages 11–15), focusing on geometry instruction in Czechia. The study builds on the differing conceptualizations of square and rectangle in Czech and Anglo-Saxon didactics. In the Anglo-Saxon tradition, a square is typically regarded as a special type of rectangle, emphasizing hierarchical classification. In contrast, Czech didactics treats these shapes as categorically distinct. This difference reflects broader didactical orientations: Czech mathematics education often emphasizes analytical decomposition and local precision, while Anglo-Saxon approaches favor structural generalization and class inclusion. These contrasting tendencies are mirrored in curricular goals, instructional strategies, and classroom expectations. This divergence becomes especially problematic when AI models—trained largely on English-language data—produce responses that implicitly reflect Anglo-Saxon conventions, which may conflict with the Czech didactical contract. The study uses a comparative, non-experimental methodology to analyze responses from multiple AI systems, including ChatGPT, Gemini, Claude, Copilot, and Mistral. Prompts were administered in both Czech and English to assess the consistency and contextual adaptability of the models. Findings suggest that AI tools may inadvertently reinforce foreign conceptual frameworks, creating tension in cross-cultural educational contexts. The paper highlights the importance of contextual sensitivity, critical digital literacy, and pedagogical oversight in the integration of AI into mathematics instruction. By revealing how culturally embedded definitions in geometry—shaped by language and curriculum—can clash with AI-generated content, this paper offers a relevant perspective for educators facing misalignment between linguistic training data and local didactical norms.