Kyungri Kim

8 years as a practicing elementary school teacher — conducting live empirical research inside real classrooms, with direct access to authentic learner behavior and AI interaction data.

Collects and analyzes tens of thousands of learner–AI interaction logs using deep learning and advanced quantitative methods to uncover how AI-mediated learning actually works — and when it doesn't.

Designs taxonomy-driven interaction data structures and generative AI-mediated instructional systems grounded in real educational context — from chatbot architecture to classroom deployment.

Defines generative AI as a Cognitive Collaborator — not a tool, but a structured thinking partner — and investigates how its role in learning can be precisely designed and empirically validated.

Researches the sweet spot of learner agency: the optimal point where learners maintain cognitive ownership while AI scaffolding maximizes learning outcomes — grounded in Strategic Cognitive Offloading theory.