#314: Reconceptualizing Activity Theory for Human-AI Teaming in Computer-Supported Collaborative Learning

The rise of artificial intelligence (AI) and large language models (LLMs) are reshaping the role of technology in Computer-Supported Collaborative Learning (CSCL). Once viewed as mediating tools, AI systems can now act as active teammates that co-construct meaning, share responsibility, and influence team cognition. This paper addresses the lack of theoretical grounding in existing human-AI teaming research by reconceptualizing Engeström’s Activity Theory. Specifically, we extend the six components of the activity system (subject, tool, rules, community, division of labor, and object) to account for AI’s dual role as both tool and quasi-subject within collaborative learning activities. Through this theoretical mapping, we highlight emerging questions for future research, including but not limited to issues of learner agency, trust, equity, and epistemic alignment. We conclude with a call for the CSCL community to revisit and expand its theoretical frameworks to better capture the evolving dynamics of human-AI teaming.