Interaction Studies - Online First

Elderly care is a key area of interest in social robotics, yet research on the practical applications and acceptance of such robots in real-world group settings remains limited. In this study, we use an ethorobotic approach to explore the potential of a socially assistive robot (SAR) to elicit interactive behaviours, such as orienting, approaching, withdrawing, touching, talking, and smiling, from elderly residents and other non-resident individuals in a nursing home. The elderly residents demonstrated high levels of acceptance and curiosity toward the robot, whereas most non-resident staff tended to ignore it. Residents showed significantly more interest in the robot than non-residents, and both groups maintained a consistent level of interaction over the 11-day observation period. These exploratory observations suggest that it is indeed possible to design socially interactive robots capable of sustaining long-term engagement with elderly individuals.

As Generative Artificial Intelligence (GAI) becomes increasingly integrated into daily life, understanding how users develop trust in these systems while navigating privacy concerns is critical. This study examines how perceived anthropomorphism, privacy concerns, and dependency influence trust in GAI, drawing on Privacy Calculus Theory (PCT) and Media Dependency Theory (MDT). The findings reveal that users trust GAI more when they perceive it as human-like, but privacy concerns reduce trust, creating a trust-privacy paradox. However, GAI dependency moderates these relationships, strengthening the positive effect of anthropomorphism on trust while weakening the negative impact of privacy concerns. Additionally, privacy concerns partially mediate the relationship between anthropomorphism and trust, suggesting that users who perceive AI as human-like worry less about privacy risks. By integrating PCT and MDT, this study offers a comprehensive framework to understand how trust in AI evolves, not just through rational cost-benefit evaluations (PCT) but also through behavioral adaptation based on dependency (MDT). These insights have practical implications for AI developers and policymakers, emphasizing the need for human-centered AI design, privacy safeguards, and ethical guidelines to foster sustained trust in AI-driven interactions while addressing user concerns.