Assessing Creativity in Mathematical Modeling Solutions Produced by ChatGPT and Preservice Teachers

Creativity, a core 21st-century competency, is especially important in mathematical modeling because modeling requires mathematically sound, original, useful, and plausible solutions to real-world problems. The emergence of large language models raises questions about how established creativity indicators apply to artificial intelligence (AI)-generated modeling products. In this study, we analyzed written solutions produced by 140 preservice teachers (PSTs) and one archived zero-shot ChatGPT response per task using a modeling-specific framework based on fluency, originality, and usefulness. We examined three complex modeling tasks: cover-up, sand carrier, and eclipse. The PST corpus served as the task-specific reference distribution, while ChatGPT outputs were positioned descriptively within these distributions rather than treated as estimates of general or stable AI performance. Findings showed strong task dependence and a substantial floor effect on the eclipse task, with many PSTs providing incomplete or uninformative responses. Under the specified settings, three ChatGPT solutions ranked highly within the PST score distributions, primarily because their approaches were judged useful and coherent. Some PST solutions exhibited broader strategic variation and several original approaches, although these were often underdeveloped or undermined by validation errors. Fluency also varied across tasks, with the eclipse task eliciting the fewest complete approaches. These condition-specific findings suggest that creativity in modeling is best understood as a multidimensional, task-sensitive profile rather than a single performance measure. They also highlight complementary creativity patterns in PST- and ChatGPT-produced solutions and their potential for AI-supported modeling pedagogy in teacher education, especially for orchestrating and critically evaluating AI-generated solutions.

Authors

Institutions

Publication Details

Journal
Asian Journal for Mathematics Education
Published
2026-11-20
DOI
https://doi.org/10.1177/27527263261502466
Primary Topic
Mathematics Education and Pedagogy
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Assessing Creativity in Mathematical Modeling Solutions Produced by ChatGPT and Preservice Teachers

Mustafa Çevikbaş, Gabriele Kaiser, Xiaoli Lu
8 links
Asian Journal for Mathematics Education
Mathematics Education and Pedagogy
article

Assessing Creativity in Mathematical Modeling Solutions Produced by ChatGPT and Preservice Teachers

Mustafa Çevikbaş, Gabriele Kaiser, Xiaoli Lu
article en

Abstract

Creativity, a core 21st-century competency, is especially important in mathematical modeling because modeling requires mathematically sound, original, useful, and plausible solutions to real-world problems. The emergence of large language models raises questions about how established creativity indicators apply to artificial intelligence (AI)-generated modeling products. In this study, we analyzed written solutions produced by 140 preservice teachers (PSTs) and one archived zero-shot ChatGPT response per task using a modeling-specific framework based on fluency, originality, and usefulness. We examined three complex modeling tasks: cover-up, sand carrier, and eclipse. The PST corpus served as the task-specific reference distribution, while ChatGPT outputs were positioned descriptively within these distributions rather than treated as estimates of general or stable AI performance. Findings showed strong task dependence and a substantial floor effect on the eclipse task, with many PSTs providing incomplete or uninformative responses. Under the specified settings, three ChatGPT solutions ranked highly within the PST score distributions, primarily because their approaches were judged useful and coherent. Some PST solutions exhibited broader strategic variation and several original approaches, although these were often underdeveloped or undermined by validation errors. Fluency also varied across tasks, with the eclipse task eliciting the fewest complete approaches. These condition-specific findings suggest that creativity in modeling is best understood as a multidimensional, task-sensitive profile rather than a single performance measure. They also highlight complementary creativity patterns in PST- and ChatGPT-produced solutions and their potential for AI-supported modeling pedagogy in teacher education, especially for orchestrating and critically evaluating AI-generated solutions.

Asian Journal for Mathematics Education
Vrije Universiteit Brussel (BE), Universität Hamburg (DE), East China Normal University (CN)
Quality education
Mathematics Education and Pedagogy
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon