A systematic review of artificial intelligence techniques for collaborative learning over the past two decades

Abstract

This systematic review focuses on publications related to studies of the use of artificial intelligence (AI) for collaborative learning. The use of AI for collaborative learning is a recent phenomenon and a systematic review of such studies is lacking. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol, 41 journal articles were shortlisted. These articles were analysed in terms of the contexts of study (year, group size, platforms of interaction, and types of learners). Using a thematic approach, two broad foci and five sub-categories of use of AI for collaborative learning were identified: (1) learning outcomes – (a) collective performances, and (b) content of learning; (2) social interactions and processes – (c) sentiments and emotions, (d) discourse patterns and talk moves, and (e) learner characteristic and behaviours. The AI techniques were coded according to three broad purposes (discovering, learning and reasoning) and nine techniques: clustering, ensemble, regression algorithms, deep learning, decision trees, natural language processing, instance-based, fuzzy logic, and agents. A Sankey diagram was used to depict the relationships among the nine AI techniques to the five sub-categories of how AI was used to support collaborative learning. The gaps in the selected articles and limitations of the current review were discussed to suggest future areas of study.