Using Learning Analytics to Unveil Human–AI Collaborative Patterns Between High and Low Performance Students in Instructional Design Activities

Human-AI collaborative patterns represent the recurring sequences of learning processes that emerge during human-AI interactions in educational contexts. Most research investigated human-AI collaboration as outcome-focused activities. From a learning analytics perspective, this research developed a dual-layered coding framework to identify students’ behaviors and corresponding cognitive content with the purpose of examining the collaborative patterns during instructional design activities. Students were first classified into the high and low-performance groups based on instructional design performances, and then the instructional design competency, behavioral-cognitive patterns and perceptions about GenAI were compared between two groups. Results showed that two groups developed different collaborative patterns despite same AI access. High-performance students demonstrated integrated behavioral-cognitive patterns, using copying as a strategy to transition from content exploration to innovative elaboration. Low-performance students exhibited fragmented behavioral-cognitive patterns. This research challenged assumptions about AI’s equal effects in learning and suggested differentiated scaffolding strategies to support equitable collaboration.