Understanding Public Agencies' Expectations and Realities of AI-Driven Chatbots for Public Health Monitoring

Advances in artificial intelligence (AI) offer the potential for chatbots to support public health monitoring by automating tasks traditionally performed by frontline workers. While introducing AI impacts public agency workers across decision-making, administration, and monitoring roles, the perceptions of workers regarding these technologies and their actual impact on labor are underexplored. We examine the case of CareCall, a large language model (LLM)-driven chatbot used to monitor socially isolated individuals, by interviewing 21 public agency workers across 13 sites involved in its adoption and rollout. We find that CareCall helped expand public reach but increased burdens on frontline workers due to insufficient resources and new labor demands, such as handling lapses in user engagement. We discuss how implementing LLM-driven chatbots in public health contexts can complicate decision-makers’ articulation work and impose additional maintenance work on frontline workers. We recommend AI chatbots in this space leverage public infrastructure and incorporate fallback mechanisms.