A student’s written solution can contain a promising idea, an unexpected interpretation, or a question worth exploring together. Recognizing these possibilities takes careful reading.
In a forthcoming chapter, Prof. Michal Yerushalmy, Dr. Shai Olsher, and Gil Sher (M.Ed. candidate) examine how ChatGPT analyzes twelve high-school students’ responses to a rich mathematical modeling problem. Students were asked to propose bus routes and explain what made a route the best choice.
The study compares AI-generated analyses with expert educators’ interpretations. It documents useful recognition of representations and reasoning, alongside difficulties interpreting sketches, connected explanations, and unconventional approaches. These findings highlight the importance of teachers’ professional judgment when using AI-generated assessments.
For Fundamatics, this research opens a conversation about using technology to attend more closely to students’ thinking and support inquiry in mathematics.
Read the chapter
Prof. Michal Yerushalmy · Dr. Shai Olsher · Gil Sher (M.Ed. candidate)
In press · Springer · English manuscript · 30 pages
Open the reader in a full window
Reference
Olsher, S., Yerushalmy, M., & Sher, G. (in press). Large language model as teachers’ assistants in inquiry-based learning: Analyzing student answers to rich mathematical problems. In Generative Artificial Intelligence in the Changing Mathematics Classroom. Springer.