The association between groups' interactions with the Visual-GenAI learning analytics feedback and student engagement in CSCL

Promoting student engagement has long been a vital subject in the research of Computer-Supported Collaborative Learning (CSCL). Previous research has indicated the potential of AI-based visual learning analytics feedback and generative AI (GenAI) feedback in this context. However, there is currently a lack of definitive research on the combined impact of these two types of intelligent feedback in CSCL. Additionally, limited attention has been paid to how groups utilize these tools in CSCL practice and the differences that may exist. In this study, we developed an Visual-GenAI learning analytics feedback tool that integrates AI-based visual learning analytics feedback and GenAI-based feedback. We then evaluated the differences in groups' interactions with this Visual-GenAI learning analytics feedback and its association with student engagement and academic performance. The study employed a mixed-methods approach, combining quantitative analysis of feedback interaction log data, content analysis of group discussion data, and qualitative analysis of students' perceptions of different feedback tools through surveys. Our results show that groups exhibit four distinct levels of feedback interaction behavior patterns with the Visual-GenAI learning analytics feedback. These four patterns exhibit significant differences in behavioral engagement, emotional engagement, cognitive engagement, and academic performance. This study's significance lies in its potential contribution to future research on examining group behavior and optimizing learning using AI-based visual learning analytics feedback and GenAI-based feedback.