Teachers adjusted problem topics as the most common refinement move¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Of the 179 teacher moves used to refine 46 personalized problems, 48 involved adjusting the problem's topic, 16 involved adjusting to fit local geographical context, and 17 involved changing names of people. → Walkington 2025

Evidence¶

Walkington 2025¶

Walkington, C., Beauchamp, T., Ikram, F., Koçyiğit Gürbüz, M., Xia, F., Lee, M., Lan, A. (2025). Mathematics Teachers’ Interactions with a Multi-Agent System for Personalized Problem Generation. https://arxiv.org/abs/2604.12066. doi:10.1007/978-3-032-29770-9_5

q2 · i? · design · r2

Thematic analysis of teacher prompting moves in the PPB logs found that topic adjustment was the most common code under the authenticity theme. Teachers also made 16 prompts asking the LLM to adjust to fit their local geographical context and 17 moves asking to change names of people, often to the names of students in their class.

"In 48 of the 179 moves, the code indicated that the teacher adjusted the problem’s topic; e.g., “Add that the student loves Dr. Pepper but his parents won't buy for him.”"

Discussion¶

Learner Variables¶

• Domain Context — outcome: instruction changes it

Related Claims¶

• Mathematical hallucinations were rare but teachers used professional knowledge to adjust mathematical properties — related
• Realism issues were detected by agents during generation but were rare in final versions — related
• The three refinement strategies make distinct trade-offs when personalizing problems to topics that resist natural alignment with the original math — related
• Authenticity and realism are the most frequent failure modes in initial LLM-personalized math problems, and a single refinement iteration substantially reduces failures — related