Can AI Become a Friend to Older Adults? Exploring Chatbot Interaction Design Strategies to Alleviate Social Isolation

While social isolation has emerged as a critical challenge in aging societies, it remains puzzling how AI (artificial intelligence) can effectively enhance social support and well-being for older adults. This study aims to explore the dynamic interplay between social chatbots and older adults, identifying how social chatbots provide social support and alleviate isolation perception for older adults in day-to-day interactions. A qualitative study was conducted using semi-structured interviews with 24 cognitively intact older adults who engaged with the Replika social chatbot. Following an inductive, iterative thematic analysis protocol and social support framework to categorize emergent support patterns until theoretical saturation was achieved. Three sequential stages of relationship development between older adults and Replika were identified: (1) Exploratory Engagement, dominated by companionship support driven by the chatbot’s availability and instant responsiveness; (2) Further Dialogue, centered on emotional support with increased self-disclosure of sensitive topics, facilitated by transparent privacy policies and context-aware empathic responses; (3) Stable Integration, characterized by synergistic multi-dimensional support, with chatbots transitioning from active social substitutes to embedded life companions. This study advances understanding of human-chatbot interactions in gerontology by uncovering a stage-specific “relationship-supported” model. This work addresses critical gaps in digital gerontology and offers actionable strategies for mitigating social isolation in aging societies. These insights contribute to the ethical development of AI companions tailored to older adults’ psychosocial needs, emphasizing their role as supplementary rather than substitutive social agents.