Skip to main content Skip to main navigation

Publikation

Empowering Academic Exploration with LLM-Powered Robots: Supporting Students in Topic and Research Question Development

Georges Arnaud Kouayim Bonga; Ilona Buchem; Niels Pinkwart
In: Richard Balogh; David Obdrzálek; Nikolaos Fachantidis. International Conference on Robotics in Education (RiE). Pages 126-137, Springer Nature Switzerland, 4/2025.

Zusammenfassung

Formulating research questions can pose a challenge for students in higher education, particularly in interdisciplinary settings where students and supervisors need to engage in dialogues across different scientific backgrounds. This study investigates the potential of integrating Large Language Models (LLMs) with social robotics to facilitate this process. Using a humanoid robot Furhat powered by an LLM, students received interactive guidance in formulating research topics and questions. The system combines verbal interaction through the robot with a visual feedback interface, ensuring an engaging and dynamic learning experience. A mixed-methods evaluation with 47 university students examined feasibility, effectiveness, and user experience. Data from pre- and post-interaction questionnaires, video recordings, and semi-structured interviews suggest that the system enhances engagement, fosters curiosity, and supports structured academic exploration. Statistical analysis revealed strong correlations between information gain, topic engagement, and system-induced curiosity, highlighting the system’s effectiveness in stimulating deeper thinking about research topics. Furthermore, qualitative feedback indicated that students appreciated the interactive and adaptive nature of the robot’s responses, which helped them refine their research ideas in a structured manner. The findings demonstrate the potential of AI-driven tutoring in higher education, showing that combining humanoid robots with LLMs can create an effective and engaging research support tool. This study provides valuable insights for the further development of AI-assisted academic guidance and highlights future research directions to optimize human-robot interaction in educational contexts.