AI as a New Problem Generator in the Context of Mathematical Modelling
Having been studied for centuries, mathematical modelling is an important perspective in mathematics education; however, teachers often face difficulties in accessing appropriate and high-quality modelling activities for classroom use. This study investigates the capacity of ChatGPT and its customised version, the Mathematical Modelling Problem Generator (MMPG), to produce mathematically and pedagogically meaningful modelling problems. Four AI-generated model-eliciting activities were comparatively analysed: three produced in independent ChatGPT-4 sessions and one produced by MMPG, all focusing on ratios and proportions. The problem scenarios were evaluated using five characteristics from established frameworks, focusing on context (non-routine and real-life relevance) and complexity (multiple steps, uncertain solution paths, and multiple solutions). The results show that ChatGPT failed to produce MEAs that met key modelling characteristics, whereas MMPG successfully generated non-routine, multi-step MEAs with unclear solution paths and multiple valid solutions. Therefore, MMPG can be considered an effective tool for supporting pedagogical practice by improving teachers’ access to and design of content-aligned, high-quality modelling activities.
Authors
- Elif Saygı (ORCID: https://orcid.org/0000-0001-8811-4747)
- Gizem Ünsal (ORCID: https://orcid.org/0000-0002-4379-423X)
- Şeyma Şengil-Akar
Institutions
- Tokat Gaziosmanpaşa Üniversitesi (TR)
- Hacettepe University (TR)
- Kastamonu University (TR)
Publication Details
- Journal
- International Journal of Education in Mathematics Science and Technology
- Published
- 2026-11-22
- DOI
- https://doi.org/10.46328/ijemst.7674
- Primary Topic
- Mathematics Education and Pedagogy
- Type
- article
- Field-Weighted Citation Impact
- 0.00