Abstract Background and Objectives The digital divide and limited AI literacy pose significant barriers to technology adoption among older adults in low-resource communities. This study investigates the potential of socially assistive robots (SARs) to promote social engagement and psychosocial well-being by analyzing interactions with the AI-driven Hyodol SAR. Research Design and Methods Multimodal data—including log-based usage patterns and voice recordings—were collected via SAR-embedded sensors. Pre- and post-intervention surveys provided demographic and health information. Human-robot conversations were classified into nine emotional and topical categories, along with six types of activity participation. K-means clustering was employed to identify distinct user personas reflecting engagement patterns. Results Of the participants, 44.6% engaged in conversation with the SAR, and 30.2% discussed activity participation. Three user personas emerged: Social Engagers (28.35%) balanced social and personal interactions with positive emotional tone; Independent Reflectors (41.79%) showed high conversational engagement; Emotionally Expressive Users (29.85%) demonstrated the highest overall SAR usage, including tactile and content-based interactions. While some clusters exhibited numerical reductions in loneliness and depression, these changes did not reach statistical significance. Discussion and Implications These findings suggest that SARs can complement caregiving services for older adults in low-resource communities. By integrating narrative data with quantitative survey responses and usage logs, this study advances methodological approaches in AI-driven gerontological research. The results highlight opportunities for persona-based customization, AI-adaptive learning, and emotion-informed care in future SAR development.
Choi et al. (Thu,) studied this question.