Prospective Teachers' Perceived Affordances and Constraints of Using ChatGPT in Mathematical Problem Posing

Abstract

This qualitative study explores prospective mathematics teachers' (PMTs) perceived affordances and constraints of using ChatGPT in mathematical problem posing. Data were collected from 23 PMTs through reflection papers, which documented their experiences using ChatGPT to pose real-life mathematical problems with varying cognitive demand levels. The study employed qualitative content analysis using both deductive and inductive approaches to examine PMTs' experiences with ChatGPT. The findings reveal that PMTs value ChatGPT for its immediate feedback, contextual diversity, and ability to provide personalized learning paths. Additionally, PMTs highlighted its potential as an assessment tool for evaluating the quality of posed problems. However, significant constraints were also identified, such as ChatGPT's limited ability to generate diverse and novel problems, difficulties in producing realistic and contextually rich problems, and mathematical inaccuracies. Additionally, PMTs noted ChatGPT's poor visualization and complex language use, particularly in higher cognitive demand tasks.