Robots as Social Catalysts to Foster Human-human Interactions

PhD dissertation; research; robots; human-computer interaction; machine learning


As artificial intelligence (AI) devices become more common in our homes, concerns about their potential harm to human-human connections arise accordingly. This dissertation aspires to study the responsible design of AI agents as social catalysts to purposefully enhance human-human interactions. Deeply motivated by this calling, it aims to shed light on the following three overarching research questions. Can we become more socially connected with one another through the facilitation of a socially embodied agent? What social capabilities do these embodied agents need to acquire as social catalysts? What approaches should we take to responsibly design, develop and evaluate computing systems that enable positive social interactions between a human group and an embodied agent?

To investigate the three questions, this work proposes a cross-disciplinary framework to holistically design and evaluate personalized AI agents aimed to foster human-human connections, via a multidisciplinary lens and a mixed-method integrative approach. This framework helps resolve the identified key roadblocks and illuminate future directions for research on personalized AI agents in human group interactions across application domains (e.g., education and healthcare). To empirically demonstrate the framework’s applicability to real-world scenarios and illustrate its usage step by step, this dissertation focuses on the at-home robot-facilitated parent-child interaction.

High-quality, reciprocal parent-child interactions critically help children develop socially, emotionally, cognitively, and linguistically. Not every child, however, has access to socially and linguistically enriching adult-child exchanges at home. Hence, exploring the robot-facilitated parent-child interactions also has potential societal impact. A series of human subject studies and experiments on computational modeling were conducted in this dissertation to examine all key aspects of this interaction scenario, namely a robot’s social-affective perception, human-centered context-awareness, and behavior adaptation in the case of robot-facilitated parent-child interaction.

Overall, this dissertation opens perspectives on the possibility of designing embodied AI agents to perform the role of social catalysts in human groups, inspiring future work to examine the potential and challenges of AI-catalyzed group interactions from both technical and ethical views. As sociable AI devices—from personal voice agents at home to autonomous vehicles—rapidly proliferate, humans increasingly interact with AI agents in an ecology that contains other humans and other AI agents. Hence, this work also helps advance the social sophistication of embodied AI agents that live with humans in this emergent human-agent ecology, as well as the understanding of the social and behavioral mechanisms underlying this ecology.