AI Multi-Agent Enhanced Online Learning Communities: An Analysis of Collective Knowledge Construction

• In Event: From Digital Textbooks to YouTube: Innovations in Digital Learning Design

Abstract

This study investigates how AI multi-agent systems enhance collective knowledge construction in online learning environments through design-based research. Using the “Ideas-Agency-Community” framework, the research designed and implemented an online learning platform with AI Teacher, AI Peer, and AI Assistant agents supporting 411 students across multiple courses. Mixed-methods analysis including automated coding via GPT-4O examined differential contributions of AI versus human feedback in promoting knowledge construction. Results reveal AI agents excel in information provision (56.2% of responses) and structured guidance, while humans dominate opinion expression (96.7%), consensus building (100%), and emotional support (92.5%). Findings suggest AI multi-agent systems create complementary cognitive ecosystems that augment human collective intelligence.

Authors

• Xiaojie Niu, Shandong University

• Jingjing Zhang, Beijing Normal University