Welzijn.AI: Developing Responsible Conversational AI for Elderly Care through Stakeholder Involvement

Stakeholder involvement is essential for responsible AI in elderly care: The study demonstrates that engaging diverse groups—including elderly users, general practitioners, patient federations, and researchers—reveals critical perspectives that single-discipline development would miss. This multi-stakeholder approach helped identify both opportunities (combating loneliness, detecting behavioral patterns) and necessary safeguards.

Elderly users prioritize empathetic and varied interactions over advanced features: The proof-of-concept evaluation found that elderly participants valued conversational AI that was empathetic and capable of varying its responses, suggesting that responsible design for this population should focus on relational qualities rather than technical sophistication alone.

Technology alone is insufficient—implementation context matters: Stakeholders emphasized that while Welzijn.AI could help activate elderly individuals' social networks, the system must be paired with human support structures. This finding highlights that responsible AI development requires considering how technology integrates into existing care ecosystems, not just the technology itself.

Systematic value assessment should guide AI development: The study used structured methods (Hundred Dollar Method, SWOT analysis) to identify and rank what different stakeholders value in the system. This demonstrates that responsible AI requires explicit, documented processes for understanding and balancing competing values rather than relying on assumptions.