The Effects of Digital Transformation Perceptions on AI-Driven Welfare Service Demands Among Older Adults In Korea: The Moderating Role of Digital Literacy

Background: Two major trends of digitalization and aging are intersecting worldwide and technological advancement has led to the introduction of welfare technology. Robots and AI-driven systems are complementing social workers’ roles that were traditionally assumed to enhance the physical, emotional and cognitive capacities of clients, while AI technologies to identify and classify (potential) clients are widely adopted by governments for cost-efficiency. It is noteworthy that scholars pay attention to the antecedents and effects of specific AI-driven welfare services, however, they have heavily focused on policymakers and paid care workers’ perspectives. In this regard, examining demands for AI-driven welfare services is insightful, and it is necessary to evaluate not only humanoids or counseling services but also personalized/recommended automated systems. This study examines whether the digital transformation (DX) perception of older adults will generate demands for AI-driven welfare services and whether understanding of technologies, digital literacy, moderates this relationship between DX and the demands.

Methods: The study used the ‘2025 Seoul Survey (N = 5,000)’ conducted by the Seoul Institute, and 1,497 citizens aged 60 or over were analyzed. AI-driven welfare service demands (DV) were assessed with 10 specific digitalized services (e.g., technology-based education, robot-assisted care services and using big data to identify clients). Each item was measured on a 4-point Likert scale, and the sum of scores was calculated. DX perception (IV) consists of 4 items (e.g., DX will deteriorate job status or employment) with a 5-point Likert-type scale. Digital literacy (Mo) was assessed with 6 items (e.g., from basic technology access to identifying disinformation). After adjusting for age, gender, marital status, educational attainment, single-household status, disability, homeownership status, and social media use, a moderation model was conducted using SPSS 29.0 and SPSS PROCESS macro 5.0 (model #1).

Results: The final moderation model was significant (F = 22.82, p < .001, R² = 0.156 & R² change = 0.005, p < .01). The results demonstrated that AI-driven welfare service demands were positively associated with DX perception (b = .383, p < .001) and digital literacy (b = 0.169, p < .001) among older adults. The interaction term (mean-centered) was significant with AI-driven welfare service demands (b = -0.023, p < .01).

Conclusions: To summarize, we explored the moderating effects of digital literacy on the relationship between DX perception and AI-driven welfare service demands based on the perspectives of potential clients, older adults. Although some ‘silver-surfers’ benefit from digital information and online activities, many older adults are digitally excluded, and they are not capable of using smartphones or self-ordering machines. For intervention, we should consider that older adults are at risks of disinformation and age-biased algorithms. The significant interaction term (DX perception x digital literacy) suggests hands-on training that can enhance digital literacy to critically evaluate technologies for older adults, as well as interventions in welfare services without digital barriers of AI-driven welfare services.