An AI-Driven Card Playing Robot: An Empirical Study on Communicative Style and Embodiment with Elderly Adults

The global population is aging, presenting growing challenges, particularly as shortages in social-related professions exacerbate these issues. Older age and related circumstances, such as the end of employment, often lead to social isolation and reduced cognitive stimulation, highlighting the need for technology-driven alternatives. This paper explores the use of socially interactive agents (SIAs) to address some of these challenges. Specifically, we implemented a prototype featuring either a physical or virtual embodiment and social-or task-oriented communication, which can autonomously recognize physical cards, game context and instruct moves. One focal point is the implementation, giving a broad overview of the architecture and important features. Equally relevant is the comparison of communication styles and embodiments, both known to influence users' perceptions when interacting with SIAs. We investigated the communication styles and embodiments to evaluate their influence on social presence, enjoyment, warmth, and competence in a user study with older adults. Our results revealed a positive effect on social presence with a physically embodied SIA in use and higher perceived competence for the social-oriented communication style. Furthermore, our findings confirm the proof-of-concept for the card detection AI and the gameplay strategy. This results in a promising tool for entertainment and cognitive stimulation and rewards us with overall positive feedback.