AI-Based Adaptive Learning: A Literature Review on Personalization, Learning Outcomes, and Implementation Challenges
Keywords:
Artificial Intelligence; Adaptive Learning; Personalization; Learning Outcomes; Implementation ChallengesAbstract
Artificial intelligence (AI)-based adaptive learning can tailor instruction to learners’ knowledge, progress, and support needs, but its educational effectiveness and implementation remain uneven. This review examined how AI-enabled systems personalize learning, what outcomes have been reported, and which factors influence implementation. A structured literature synthesis focused on ten focal empirical and review sources published within the recent decade. Evidence was organized thematically around personalization mechanisms, learning outcomes, and implementation conditions. The findings show that systems personalize learning through learner modeling, adaptive task selection, individualized feedback, and adjustment of learning pathways. A meta-analysis of 45 independent studies reported a medium-to-large positive effect on cognitive outcomes compared with non-adaptive interventions (g = 0.70). However, results varied across contexts: a randomized trial involving 266 middle-school students found no significant average treatment effect, although learners in the lowest achievement decile improved by 0.358 standard deviations. Other studies reported improved science learning through adaptive task selection and increased mathematics achievement alongside reduced anxiety when systems considered cognitive and affective information. Implementation challenges included low learner participation, incomplete task engagement, technological constraints, data quality, privacy, equity, and the need for educator involvement. The review concludes that AI-based adaptive learning has meaningful potential, but benefits are not automatic or uniform. Effective implementation requires alignment with learning objectives, responsible use of learner data, sustained participation, and teacher oversight. Future research should report implementation fidelity and examine outcomes across learner groups and educational settings.