Prototyping A Multi-Agent System to Enhance AI-Human Collaboration in IEP Development

• In Event: Transforming Assessment and Learning: Harnessing AI for Educational Excellence

Abstract

Individualized Education Programs (IEPs) are legally binding documents that outline annual educational goals, specially designed instruction, and other support necessary to ensure access to the general education curriculum for students with disabilities. Developing high-quality IEPs is complex and involves collaborative efforts from multiple stakeholders. In this study, we reported on the development and evaluation of a prototype of a multi-agent system powered by large language models (LLMs), called CoIEP, which can be used to streamline the complex process of creating high-quality core IEP components. Additionally, the system can serve as a professional learning tool, especially for pre-service and novice educators. We discussed how the system is designed to address common pitfalls of using LLMs (e.g., hallucinations) for IEP development.

Authors

• Ling Zhang, University of Wyoming

• Haidee A. Jackson, University of Texas - Permian Basin

• Sohyun Yang, Fort Hays State University

• Richard Carter, Indiana University

• Jihong Zhang, University of Arkansas at Fayetteville

• Jennifer A. Diliberto, University of North Carolina - Chapel Hill