ChatGPT for Mathematics: Supporting Students’ Computational Thinking in a Classroom Activity

The growing use of generative artificial intelligence in education raises questions about how students engage with mathematical tasks when supported by tools such as ChatGPT. This study explores how different configurations of ChatGPT shaped students’ computational thinking during a Python Turtle classroom activity. Two 13th grade school classes worked in small groups on the same task: one used standard ChatGPT, the other a custom-configured version designed to foreground mathematical explanation, incremental development, and debugging. Data included students’ prompts, discussions, worksheets, and final artefacts. The comparison revealed different interactional patterns. Students using standard ChatGPT more often relied on short prompts and direct corrections, whereas those using the custom version more often refined prompts, discussed constraints, and revised their work through debugging. These findings suggest that the educational value of ChatGPT depends not only on access to the tool, but also on its pedagogical configuration.