Artificial Intelligence (AI)-based systems, such as ChatGPT, have emerged as a potential resource to assist teachers in designing and planning mathematics lessons. In particular, recent studies analyzed the role of ChatGPT in mathematical problem-solving, pointing out its use in teacher training. However, despite the advantages of using ChatGPT in mathematics teaching, teachers must have the necessary didactic-mathematical knowledge to provide ChatGPT with prompts consistent with its needs and to evaluate the adequacy of the information returned by the AI. In this paper, we describe through a case study, how future teachers interact with ChatGPT to create a problem with a specific didactic-mathematical purpose: it must involve proportional reasoning and have a high cognitive demand. We analyze the actions of a future teacher and ChatGPT, the significance of the created problem and how the user evaluates or interprets it. We conclude the importance of adequate didactic-mathematical training for pre-service teachers to effectively use ChatGPT as a tool to create meaningful mathematical problems.

Keywords: Problem posing, proportional reasoning, teacher training, Artificial intelligence-based systems.