Building an Instructional Design–Backed, GPT-Driven AI Tutor for Math Homework Support

Josh Fisher, Husni Almoubayyed, Stephen E. Fancsali, Steve Ritter, Logan De Ley, Zack Lee

Track: Innovations in AI for Education (Day 1)

Paper Length: short-paper (2 pages + references)

Keywords: large language model, chatbot, middle/high school mathematics, AI tutoring

TL;DR: Building AI education assistance embedded in local curricular contexts and with a strong instructional design provides for improved safety, accuracy, and reliability of these systems.

Abstract: We describe the latest iteration of a digital chatbot to support students using a widely-deployed print and digital work-text for mathematics, building on a rule-based, static chatbot implementation with 200k+ users to develop a version driven by large language models that strives to be safe, mathematically accurate, and instructionally clear.

Submission Number: 40