Phase-Specific Effects of AI in Self-Regulated Learning: A Meta-Analysis of a Decade of Interventions

• In Event: Virtual Poster Session
  In Virtual iPresentation Only: SIG-Studying and Self-Regulated Learning Virtual Poster Session

Abstract

This meta-analysis addresses inconsistent findings on AI's role in self-regulated learning (SRL) by systematically examining 35 empirical studies (2013-2025) through Zimmerman's cyclical model. Results reveal a moderate overall effect of AI interventions (g=0.507), with significantly stronger impacts during the task performance phase (g=0.574). A key finding is the superior efficacy of generative AI (g=0.799) over rule-based and data-driven paradigms across all SRL phases. Critically, this synthesis highlights a "performance-competence divide," where AI excels at scaffolding immediate tasks but its ability to cultivate transferable competence remains unproven. This research offers a nuanced, evidence-based map for designing future AI systems that foster genuine learner autonomy.

Authors

• Jun Xu, East China Normal University

• Yuying Luo, Shanghai University

• Chengliang Wang, Australian Catholic University

• Hao Zheng, East China Normal University

• Yonghe Wu, East China Normal University