This quasi-experimental study asks whether a cooperative flipped classroom supported by an AI chatbot improves university students' collaboration and self-regulated learning, along with their academic performance, and whether students with different levels of prior knowledge respond in different ways. Drawing on mastery learning theory, it compared 154 junior students over 11 weeks with the same instructor. One group used a chatbot during pre-class group work, and a control group followed conventional group-based flipped learning. Interviews with 10 students from the chatbot group were examined with Epistemic Network Analysis.
- The chatbot group scored significantly higher than the control group on posttest measures of collaboration and self-regulated learning.
- The chatbot group also earned significantly higher posttest academic performance scores.
- Within the chatbot group, students who interacted with the chatbot more did significantly better academically than those who interacted less.
- Students with lower prior knowledge showed denser collaboration networks, needing more cognitive effort and more frequent switching between dimensions to coordinate group work.
- Students with higher prior knowledge showed stronger links between self-regulated learning dimensions such as seeking and reflecting, which the authors read as more integrated self-regulation.
The authors conclude that the results support mastery learning theory and that a chatbot can act as a mastery-oriented scaffold, helping groups divide tasks and clarify goals during pre-class discussion so students arrive better prepared for class. They recommend differentiating support by prior knowledge, with clearer goals and stronger cognitive scaffolding for learners who start with less, and tasks centred on integrating and reflecting on knowledge for higher achievers. They also suggest the approach suits large or resource-constrained teaching settings.
