ChatGPT and mathematical modeling of motion for teaching linear functions in secondary education

This exploratory study analyzes the pedagogical mediation conditions under which ChatGPT's intervention in mathematical modeling amplifies coordination among semiotic registers and favors functional thinking. Twelve second-year secondary school students participated in four teams that modeled the rectilinear motion of a person captured on video and constructed piecewise linear functions. From the articulation between the theory of semiotic representations and Realistic Mathematics Education, the study documented how the teams coordinated the four registers when interacting with ChatGPT. The analysis identifies four conditions that made that mediation productive: the selection of a contextualized situation, the iterative refinement of the prompt, the open stance of the teacher, and the availability of multiple representations as an object of contrast. Given its limited scope, the results offer a first approach to how generative artificial intelligence redistributes cognitive work toward conceptual analysis, and require replication in other groups.