First impressions of a humanoid social robot with natural language capabilities

Abstract

Concurrent developments in robotic design and natural language processing (NLP) have enabled the production of humanoid chatbots that can operate in commercial and community settings. Though still novel, the presence of physically embodied social robots is growing and will soon be commonplace. Our study is set at this point of emergence, investigating people's first impressions of a humanoid chatbot in a public venue. Specifically, we introduced "Pepper" to attendees at an innovation festival. Pepper is a humanoid robot outfitted with ChatGPT. Attendees engaged with Pepper in a bounded interaction and provided feedback about their experience (n = 88). Qualitative analyses reveal participants' mixed emotional resonance, reactions to Pepper's embodied form and movements, expectations about interpersonal connection and rituals of interaction, and attentiveness to issues of diversity and social inclusion. Findings document live responses to a humanoid chatbot, highlight the affective, social, and material forces that shape human-robot interaction, and underscore the value of "in the wild" studies, creating space and scope for user-publics to express their perspectives and concerns. Such insights are acutely relevant as we enter this next era of social robotics.