AI as a Thinking Partner in Sustainability-Oriented Mathematical Modelling: An Exploratory Study of Pre-Service Teachers’ Comparative Reflection

Generative artificial intelligence (AI) is mostly used in education to generate solutions, provide feedback or support instruction; approaches that use AI outputs as objects of comparison and reflection remain less explored. This small-scale exploratory qualitative case study examined AI-supported comparative reflection in sustainability-oriented mathematical modelling. Twelve pre-service mathematics teachers individually developed models for a task on water use in a school garden and then compared them with a model produced by ChatGPT (GPT-4o), reflecting on the comparison in writing. Modelling products, written comparisons and reflection forms were analysed using inductive content analysis. The comparison prompted participants to question assumptions, identify limitations in both models and consider alternative variables and representations; most described how they would revise their models. Their accounts contained indications of critical, creative and metacognitive reasoning. However, several participants treated the AI output as an answer key, some of their critiques echoed weaknesses that the AI had listed about itself, and none noticed a computational inconsistency deliberately retained in the output. As the design included no comparison condition or pre–post measure, the findings describe patterns of reasoning displayed during the activity, not improvements in thinking skills or modelling competence. Implications for teacher education are discussed.