Near-peer and peer-to-peer teaching are powerful, scalable sources of instruction, but translating the necessary teacher training to the same scale is difficult. Inadequate preparation can negatively impact both students and teachers– what if there were a way to expand the possibilities of teacher learning experiences by eliminating the potential for harm that teachers-in-training could inflict on real students? We present GPTeach, an interactive chat-based teacher training tool that allows teachers to practice with simulated students. We performed two studies: one A/B test with 10 participants where half received a baseline and half received our tool, and one think-aloud study where all 14 participants received our training tool. In both studies participants were tasked with taking the role of a TA in office hours and going through six teaching sessions with two GPT-simulated students in each. We found that our tool provides the opportunity for teachers to get valuable teaching practice without the pressures of affecting real students, allowing them to iterate their responses both during each session and across sessions. Additionally, participants enjoyed flexibility in tailoring their responses according to student personas and needs, as well as specified learning goals. In this paper we also provide a set of observations to inform future work in this area. We conclude with a discussion of actionable design ideas for such systems, as well as other ways to use this tool for evaluating teachers and students.
- Synergizing Human-AI Agency: A Guide of 23 Heuristics for Service Co-Creation with LLM-Based Agents - December 1, 2023
- AI-Augmented Surveys: Leveraging Large Language Models and Surveys for Opinion Prediction [imputation] - November 29, 2023
- Enhancing Human Persuasion With Large Language Models - November 29, 2023