Effect of Generative AI on Undergraduates’ Learning Outcomes: A Systematic Review and Meta-Analysis

• In Event: AI-Enhanced Student Learning Outcomes

Abstract

This meta-analysis synthesizes 57 studies and 97 effect size estimates, confirming that generative AI (Gen-AI) significantly enhances undergraduates’ learning outcomes (g+ = 0.804). The strongest effects are seen in language skills (g+ = 2.331), followed by academic achievement, affective-motivational status, and higher-order thinking, while effects on metacognition are minimal. Guided by the AT-MCSCL framework, the study identifies multiple moderating factors, including learner characteristics (freshmen and language learners), study design (sample size, intervention duration, publication type), and contextual features (learning setting, region, ICT level, socioeconomic status, power distance). These findings provide comprehensive evidence on the overall effectiveness and contextual variability of Gen-AI in higher education.

Author

• Shuzhen Chen, Chinese University of Hong Kong