Artificial intelligence agents in computer-supported collaborative learning: A systematic literature review

Artificial intelligence (AI) agents are rapidly reshaping the landscape of computer-supported collaborative learning (CSCL), presenting both new opportunities and challenges for educators and learners. Despite their increasing prevalence, there remains a lack of comprehensive, theory-informed synthesis regarding the roles and impacts of AI agents within CSCL. This systematic literature review analyses 46 empirical studies published between 2014 and 2025, with the aim of clarifying the methodological and contextual characteristics of the field, the pedagogical and technological features of AI-mediated CSCL environments, the instructional and mediational functions of AI agents, and the associated learning outcomes across cognitive, behavioral, social, and emotional domains. Guided by the community of inquiry model and learning engagement theory, the review identifies a predominant focus on post-secondary settings, with mixed methods designs being most common. AI agents most frequently facilitate small group collaboration and problem-solving, typically through text-based online platforms. Their functions encompass cognitive scaffolding, social facilitation, and instructional orchestration, with recent developments enabling more adaptive and participatory roles. While cognitive gains are consistently reported, the effects on behavioral, social, and emotional outcomes appear context-dependent and highlight the need for nuanced agent design. The alignment between agent functions and learning outcomes is strongest within the same domain, yet important cross-domain influences are also evident. This review concludes by outlining implications for policy, theory, and practice, underscoring the necessity for equitable access, expanded conceptual frameworks, and context-sensitive deployment of AI agents to ensure meaningful and responsible integration into future CSCL environments.