Empowering Social Service with AI: Insights from a Participatory Design Study with Practitioners

AI can reduce administrative burden in social services: The study found that generative AI shows promise for streamlining documentation, assessment tasks, and worker supervision—areas where practitioners currently spend significant time that could be redirected toward direct client care.

Overreliance and skill erosion are serious concerns: Practitioners worry that depending on AI tools could cause caseworkers to lose critical professional skills over time, potentially compromising the quality of human judgment that is essential in social service decision-making.

Algorithmic bias poses risks to vulnerable populations: The research highlights that GenAI tools risk perpetuating or amplifying existing biases in social service systems, which could disproportionately harm already vulnerable clients who depend on fair and equitable case assessments.

Participatory design with practitioners is essential: By involving 27 practitioners in co-design workshops and 24 in real-world testing with actual case data, the study demonstrates that social service workers must be active partners in developing AI tools—not passive recipients—to ensure tools address genuine needs and account for on-the-ground realities.

Professional identity and client safety require careful implementation: The study emphasizes that AI integration must preserve what makes social work valuable—human connection, professional judgment, and accountability—while establishing clear guidelines about when and how AI should support (rather than replace) practitioner decision-making.