LLM powered embodied virtual tutor

Acquiring efficient individual support for learning is a significant challenge for university students. However, providing individualized tutoring to each student by tutors is often not feasible. One promising solution to this problem could be embodied virtual tutors (eVTs). These virtual tutors are AI-powered virtual agents that can explain complex concepts, answer questions, monitor learning progress, and offer support to students. Moreover, eVTs can be personalized and trained on expert knowledge, making them valuable for student learning. The project aims to investigate the potential, acceptance, and effectiveness of eVTs for individualized student learning. For this, we implement an eVT by leveraging recent advancements in artificial intelligence (e.g. LLMs) and mixed reality engineering. We will conduct user studies to improve our prototype’s usability. Finally, we will conduct a study to test the acceptance of the virtual tutor. In the future, our system might be applied to other departments and teaching environments.