Integrating Self-Regulated Learning in Generative AI Chatbots: A Quasi-Experimental Study in Programming Education

• In Event: AER Roundtable Session Friday 1:45 pm - Gold 4
  In Roundtable Session: Designing for Self-Regulation: AI, Scaffolding, and Learner Agency

Abstract

This study integrates Self-Regulated Learning theory into generative AI chatbots for programming education. Using a quasi-experimental design, 38 university students were assigned to basic-guided (control) or SRL-enhanced (experimental) chatbot groups for 4-week Python learning. While the basic-guided group showed better immediate learning outcomes, the SRL-enhanced group significantly improved self-regulated learning abilities (p = .002). Notably, help-seeking behavior diverged: increasing in the SRL-enhanced group but declining in the basic-guided group. The SRL-enhanced design also promoted deeper engagement with more conversation turns. This research reveals the trade-off between short-term performance and long-term capability development in AI-assisted learning, providing an empirical framework for translating SRL theory into AI educational system design.

Authors

• Wei-Chih Hsu, National Yang Ming Chiao Tung University

• Jerry Chih-Yuan Sun, National Yang Ming Chiao Tung University