Social Robots for Encouraging Inclusive Participation in Group Discussions among University Students: Exploring the Role of Furhat as a Facilitator

Abstract

Classroom group discussions are a fundamental part of active learning in higher education, helping students develop critical thinking and collaborative skills. However, these interactions are often unbalanced, where a few dominant voices overshadow quieter students, leading to unequal participation. Social robots have recently been integrated into educational settings to enhance engagement and support personalized learning, demonstrating a unique potential to influence social dynamics. While social robots are increasingly used in education, their ability to facilitate complex, multi-party conversations remains a challenge due to the difficulty of interpreting the subtle social cues and complexity of managing multiple speakers simultaneously. This thesis explores the potential of a social robot to serve as a feasible facilitator to promote inclusive participation in small-group academic discussions. To explore this, the study followed a Research through Design (RtD) approach. A user study was conducted with 24 university students organized into 7 small groups. The study utilized a Wizard-of-Oz (WoZ) methodology, where a human operator controlled the Furhat robot, named "Nora," to facilitate the discussion (15 minutes) on technology and human connection. Data was collected using pre- and post-session questionnaires and video recordings of the interactions, which were then analysed. Questionnaires were analysed using thematic analysis, affinity diagramming, and basic statistical methods and the video observations were systematically coded to capture interaction dynamics. The aim was to evaluate students’ perceptions of the robot, the impact of the robot on group dynamics, and the specific design elements that contributed to a sense of inclusion. The results revealed a clear trade-off between ensuring equal participation and maintaining a natural conversational flow. Functionally, the robot successfully enforced inclusive participation, ensuring that every student had an opportunity to speak. Participants rated the robot highly for its objectivity and lack of bias, which helped create a safe environment for introverted students. However, the robot's strict adherence to a scripted turn-taking structure often disrupted the natural rhythm of the dialogue, leading to interruptions that students perceived as rude or discouraging. Regarding design, the findings challenged the assumption that high-fidelity visual and behavioral realism is always superior; participants expressed a strong preference for low-tech physical customization, such as a wig and props, over high-tech facial animations, which were often described as uncanny. The primary contribution of this thesis is a set of design implications for creating effective robotic facilitators for group discussions specifically for university students. These implications advocate moving away from rigid scripts toward adaptive turn-taking strategies that respect the natural pauses in human conversation. Furthermore, the study proposes a "low-tech" design approach, showing that simple physical props are more effective than realistic animations for reducing user discomfort in university settings. Additionally, the research highlights the practical value of humor as a strategy to rebuild trust during technical errors and awkward moments. These insights provide novel knowledge for the fields of Human-Robot Interaction and Educational Technology, suggesting that for a robot to support collaborative learning, it must possess not only the intelligence to manage time but the social awareness to respect the flow of conversation.