Skip to main content Skip to navigation

Student engagement in collaborative learning with AI agents in an LLM-empowered learning environment: A cluster analysis

Project Overview

The document explores the role of generative AI, specifically large language models (LLMs), in enhancing educational experiences, particularly in collaborative learning settings. Through the examination of student interactions with AI agents in a course powered by LLMs, the study categorizes learners into three types: active questioners, responsive navigators, and silent listeners. Each type demonstrates unique engagement patterns with AI, which underscores the necessity for adaptive educational systems that can cater to the varied learning styles of students. The findings suggest that leveraging LLMs can significantly boost student engagement and participation, thereby fostering personalized learning experiences. Ultimately, the research emphasizes the potential of generative AI to transform educational practices by supporting diverse learner needs and enhancing collaborative dynamics in learning environments.

Key Applications

MAIC platform with LLM-driven AI agents

Context: An introductory course titled Towards Artificial General Intelligence, targeted at university students in China

Implementation: Students engaged with multiple LLM agents across six modules, completing coursework and interacting through a messaging interface.

Outcomes: Identification of three distinct learner types; improved understanding of student engagement dynamics; potential for enhanced personalized learning experiences.

Challenges: Limited engagement from some students; challenges in capturing all forms of student interaction and engagement.

Implementation Barriers

Engagement Barrier

Some students exhibit minimal interaction with AI agents, raising questions about their learning engagement.

Proposed Solutions: Design AI interventions that prompt deeper interaction and engagement, such as open-ended questions or gamified elements.

Technical Barrier

Dependence on message length and content as primary measures of engagement may overlook other engagement aspects.

Proposed Solutions: Integrate additional data sources like clickstream analysis, eye-tracking, or neurophysiology data.

Project Team

Zhanxin Hao

Researcher

Jianxiao Jiang

Researcher

Jifan Yu

Researcher

Zhiyuan Liu

Researcher

Yu Zhang

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Zhanxin Hao, Jianxiao Jiang, Jifan Yu, Zhiyuan Liu, Yu Zhang

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies