How Can I Help You? Didactic Alignment for RAG Chatbots in Education

Abstract

The challenge of providing good educational support despite limited resources has been concerning educational science and didactic research for many decades. With the extensive development in artificial intelligence (AI) in recent years, AI chatbots are coming into focus as a possible solution. It is evident that AI chatbots have been used in research and teaching to date, but often without defining the role and the purpose of the chatbot within a specific didactic scenario. Using AI chatbots for their own sake rather than on the basis of didactic embedding in education leaves uncertainty concerning the extent to which these tools contribute to achieving learning objectives. We developed a theoretical framework for the didactic embedding of Retrieval-Augmented Generation (RAG) chatbots in education and derive three different roles that a RAG chatbot can take on in teachinglearning settings and developed didactic scenarios and tested them with regard to their technical feasibility by implementing exemplary scenarios in RAG systems. Our conceptual framework provides a didactic foundation for future research, development, and practical utilization of RAG chatbots in education.