Jan Leusmann, Anna Belardinelli, Luke Haliburton, Albrecht Schmidt, Stephan Hasler, Sven Mayer, Michael Gienger, Chao Wang, "Investigating LLM-Driven Curiosity in Human-Robot Interaction", CHI conference 2025, 2025.

In the future, we need seamless and natural collaboration with robots. Currently, robots can only perform tasks that they have been taught either in the development stage or by using techniques like learning from demonstration. However, for humans, the natural way to learn is often through curiosity. Currently, it is unclear how users perceive the curiosity of robots. To address this, we developed a curious and a non-curious character using a Large-Language Model for a humanoid robotic system. We studied these characters in two different scenarios. In a within-subject design user study (N=\red{XX}), we had users interact with these two different characters in one scenario. Both qualitative and quantitative results show that we can change the perceived curiosity of the robot and, thus, show that we can modulate the robot's character.