Modeling Teacher-Student Co-regulation as a Relational Process in AI-Supported Learning

In formal education, regulation is inherently relational: teacher and student continuously adapt to one another through dynamic interaction, responding to and shaping each other’s actions over time. Despite its theoretical centrality in the learning sciences, teacher-student co-regulation has not yet been formalized as an explicit computational construct within AI-supported educational systems. AI in Education is evolving along two largely separate trajectories: student-facing systems that function as increasingly autonomous personalized tutors, and teacher-facing systems that position educators as recipients of analytics and performance indicators. This architectural divide is not merely technical. It implicitly reshapes the pedagogical relationship: learning becomes individually optimized, while teaching risks becoming observational rather than co-regulatory. This doctoral project investigates how teacher-student co-regulation can be formally represented and empirically studied as a relational process in AI-supported learning environments. It conceptualizes co-regulation as a dynamic coupling between student learning states and teacher instructional actions and examines how patterns of alignment and misalignment emerge over time using longitudinal digital trace data. A prototype system will be developed as a research instrument to operationalize this relational model and enable in-situ investigation of co-regulatory processes. By extending learner modeling from the individual to the relational level, the project introduces co-regulation as a unit of computational analysis in AI in Education and studies how its explicit representation shapes teacher-student coordination over time.