Exploring Older People's Preferences in the Design Features of Virtual Agents: A Comprehensive Analysis

By engaging users through natural language and nonverbal cues, virtual agents have demonstrated their potential to offer convenient and efficient support to older individuals in various scenarios. The visual appearance and interactive features of virtual agents are pivotal aspects of human-computer interaction. The main objective of this paper is to establish design guidelines tailored to the preferences of older users. To accomplish this, we began by conducting a com- prehensive case study, reviewing virtual agent designs geared towards older individuals from the past two decades across various databases. These cases were classified into three categories based on the seriousness of virtual agent application scenarios and the smartness of their systems: high seriousness and high smartness, moderate seriousness and high smartness, and low seriousness and low smartness. Subsequently, we identified the design preferences of older users for each category. It's noteworthy that the case study highlighted a scarcity of research on virtual agent feature preferences for moderate seriousness and high smartness. To address this gap, we further conducted in-depth interviews and integrated the findings with the insights from the case study, providing practical design guidelines for future virtual agents research targeting older users.