PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
August 19, 2025Mathematical Thinking and Learning15 citationsOpen Access

Middle school girls using generative AI to engage in mathematical problem-posing

View Full Paper
CWCandace WalkingtonMPMagdalena PandoLLLin Lin Lipsmeyer

Key Points

  • Girls used generative AI to create math story problems, with varied satisfaction regarding authenticity and expression.
  • Despite issues with generated problems, students remained optimistic about using ChatGPT for math tasks.
  • Participants expressed concerns about bias in outputs from DALL-E2, critiquing its effectiveness in visual generation.
  • The findings indicate potential for generative AI to enhance personalized student task generation in mathematics.

Abstract

Recent advances in generative artificial intelligence (GenAI) can support the creation of mathematical tasks. Although current tools have focused on having GenAI write tasks for learners, little research has yet explored how GenAI can further assist students in posing their own math problems. Here, we present a study of 36 middle school girls using GenAI to write mathematics story problems and generate visual illustrations. We examine how the girls interacted with GenAI during the process of problem-posing, what they saw as the strengths and weaknesses of using GenAI, and the quality of the mathematics story problems that the GenAI returned. We found that girls were optimistic about ChatGPT, despite it generating problems that had issues with authenticity and that some girls felt did not reflect their voice. The girls were more critical about DALL-E2, accentuating issues of bias. We share implications for using GenAI for personalized student task generation.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Walkington et al. (2025) studied this question.

synapsesocial.com/papers/68af4766ad7bf08b1ead47a0https://doi.org/10.1080/10986065.2025.2542724
Ask AI
Helpful
Bookmark
Share
View Full Paper