Social Robots for Human Companionship: Stigma Perceptions, Social Orientation, and Design Preferences

Abstract

Advances in artificial intelligence (AI) have transformed the capabilities of social robots, enabling them to participate in interactions that resemble human social exchange. Through adaptive learning and personalized engagement, these systems can provide counsel, emotional support, and companionship across a range of social contexts. Individuals perceive the advantages of interacting with a social robot (vs. human) such as being perceived as nonjudgmental, reduced risk of social rejection, accessible, and emotionally responsive interactions. As social robots become increasingly capable of simulating humanlike social presence, these developments raise important questions about the psychological processes that underlie how individuals perceive, evaluate, and adopt social robot as companions. Researchers in marketing, robotics, and computer science have largely focused their attention on facilitating factors that lead to social robot acceptance. Such findings may lead designers to adopt a "one-size-fits-all" perspective. However, far less is known about individual differences in social robot preferences. In addition, the literature is sparse on how social robot designs may influence stigma perceptions. In particular, there is a gap in our understanding of how social robots with anthropomorphic designs may drive inferences of humanlike capabilities, elicit stigma, and the psychological processes by which stigma shapes resistance to social robot companionship. This dissertation examines how social robot design activates psychological mechanisms that influence the adoption of AI-driven social robots for companionship. Essay 1 (Chapter 2) investigates how anthropomorphic design features shape perceptions of a social robot's cognitive, affective, and social capabilities and how these inferences mediate perceived stigma (in parallel). We then investigate a psychological process in which perceived stigma, anticipated stigma, and self-stigma serially mediate adoption intentions for social robots as companions. Essays 2 and 3 address individual differences in social relationship orientations that may signal differential benefits from AI companionship. Drawing on literature in social competence, exclusion, and solitude, Essay 2 (Chapters 3 and 4) develops a scaling methodology that classifies individuals as socially included (I), socially excluded (D), or social excluders (R). Using a multi-stage process, we create and validate a 42-item instrument that distinguishes these social relational profiles. Finally in Essay 3 (Chapter 5) we explore how design preferences for social robots (physical features, anthropomorphic qualities, interactional capabilities, and preferred relational roles) vary by these social relationship profiles. Together, these essays provide a comprehensive framework for understanding how stigma, individual differences, and design considerations may interact and influence the adoption of social robot companions. The dissertation concludes with theoretical, managerial, and policy implications for designing and responsibly deploying AI technologies that support human social needs and address the growing societal challenge of companionship deficits and loneliness.