Exploring Pre-Service Mathematics Teachers' Diagnostic Thinking During AI-Supported Task Design: A Configurational Mixed-Methods Pilot Study.

Developing pre-service mathematics teachers' ability to understand students' mathematical thinking and anticipate students' difficulties remains a challenge in teacher education. This exploratory mixed-methods study involved 21 pre-service mathematics teachers in a six-week AI-supported mathematical task design module. The study combined pre-post assessment, fuzzy-set Qualitative Comparative Analysis (fsQCA), and qualitative analysis of participants' AI interactions and reflections. The key findings were as follows: (1) Participants showed significantly higher post-test scores in overall diagnostic thinking and in perception, interpretation, and judgement than at pre-test. (2) The configurational analysis identified a comparatively stable pattern associated with high post-intervention diagnostic thinking, combining lower initial diagnostic thinking with high AI acceptance and high prompting knowledge density. (3) Qualitative evidence further showed that AI served as an interactive resource for exploring mathematical and pedagogical considerations, evaluating generated suggestions, and revising task designs. Exploratory comparisons suggested that AI use could range from broader knowledge-oriented support to more selective verification and refinement. The findings highlight the heterogeneous nature of AI-supported professional learning and underscore the importance of critical evaluation and professional agency in teacher-AI collaboration.