Improving Collaborative Learning Performance Based on LLM Virtual Assistant
Rongxuan Wei, Kangkang Li, Jiaming Lan
Abstract
The breakthroughs achieved by large language models have further sparked public enthusiasm for transforming “old” education with “new” technology. This study explores how to promote the development of collaborative learning towards intelligence and personalization. Based on theories of human-computer collaborative learning, socialist constructivism, and social interaction, this research proposes an intelligent collaborative learning model based on Large Language Models (LLM). Utilizing ChatGPT and the WeChat system, a virtual assistant based on LLM was developed to create a collaborative learning environment, providing functions such as problem feedback, resource gathering, and inspiration. The study employed a quasi-experimental research method, focusing on the “Educational Game Design” section of the “Educational Artificial Intelligence” course to test the efficacy of the proposed learning model. The results indicate that this model and its system significantly enhance the learners' group cohesion and collaborative inclination, aiding in the better construction of their knowledge systems. Learners in this collaborative learning model experience higher emotional satisfaction and increased willingness to learn, thereby improving their academic performance.