Abstract Background and Objectives This study investigates how underserved individuals—older adults, low-income middle-aged men, and adults with disabilities—develop long-term relationships with an AI-powered socially assistive robot (SAR) named Hyodol. We examined how these relationships evolve over eight months and why emotional outcomes differ across three groups. Research Design and Methods Using Colaizzi’s phenomenological method, the research team analyzed 13,403 conversational segments from 12 highly engaged SAR users (four from each group). Iterative coding procedures with high inter-rater reliability yielded 10 thematic categories capturing functional, social, and emotional dimensions of SAR interactions. Results Across all groups, interactions with the SAR demonstrated a consistent relational evolution from functional devices to companions to develop emotionally reciprocal patterns of communication. Each group displayed a distinct interaction profile: older adults emphasized routine companionship, middle-aged men combined emotional expression with entertainment and exploratory questioning, and persons with disabilities used the SAR to scaffold both emotional support and social participation. Discussion and Implications The results reveal that human–robot relationships develop meaningfully, but not uniformly across three underserved groups. The paradox observed among middle-aged men challenges assumptions that higher engagement and relational deepening may heighten emotional vulnerability. These findings highlight the need for population-specific relational models and adaptive SAR design that considers diverse emotional needs and social contexts.
Lee et al. (2026) studied this question.