Toward hybrid teaching intelligence: investigating the potential of teacher–AI collaboration using large language models

This study examines Norwegian teachers’ dispositions regarding the use of large language models (LLMs) to support their teaching. To achieve this, we conducted a qualitative exploratory study that employed semi-structured interviews. Specifically, we interviewed 13 Norwegian school teachers with previous experience in artificial intelligence (AI)-enabled learning systems, and asked them about their understanding and concerns regarding the use of LLMs in education. The coding of the interview corpus suggests that teachers have diverse opinions on the use of LLMs for teaching, and, in some areas, lack knowledge and require further training and support. At the same time, they perceive LLMs as an opportunity for specific teaching tasks, such as marking and personalised learning. Maintaining agency is perceived as an extremely important quality for successful and symmetric teacher–LLM collaboration. Moreover, we identify challenges relating to a lack of reliable evidence on the effectiveness of LLMs, as well as a need for clarity concerning the legal frameworks that control how LLMs collect and use teacher and student data. We discuss the findings of the work and elaborate on its implications and areas for future research in the context of hybrid teaching intelligence.