Slice of AI & Pie

Please join us in the Scholar's Lab for this month's Slice of AI & Pie with Liang Zhang, Postdoctoral Research Associate at the University of Georgia.

Join the meeting by Zoom

Working with AI Partners: How Multi-Agent Systems Can Support Teaching and Learning

The rapid rise of generative AI has brought both excitement and concern to higher education, particularly for faculty and students adapting to these new tools in their daily work. Beyond simple chatbots, emerging research highlights the potential of multi-agent AI systems—teams of AI “partners” that collaborate to support essential educational practices such as tutoring, feedback, and assessment. This keynote explores how multi-agent AI agents, powered by large language models (LLMs), can enhance teaching excellence while addressing the needs of both students and instructors. Drawing on case studies, I will demonstrate how AI agents can (1) co-design multimodal lesson materials in physics, combining text explanations with automatically generated images to clarify misconceptions; (2) engage in peer-style dialogue to model collaborative problem solving in mathematics, where improving accuracy is fundamental for reliable assessment; and (3) participate in trialogue systems that enable conversation-based evaluations of student knowledge and skills. Findings from a systematic review of LLM-based AI agents in education will also be shared, highlighting both promising opportunities and ongoing challenges related to evaluation, equity, and ethics. For educators, the message is clear: AI  agents should not replace human teachers or diminish human faculty expertise. Instead, when thoughtfully integrated, they can serve as trusted partners that save time, enrich pedagogy, support student learning, and reimagine what teaching excellence can look like in the age of AI.