Considerations for Agentic Research in Public Health (CDC guidance)

The Centers for Disease Control and Prevention offers an informational resource with guidelines for state, tribal, local and territorial public health agencies on the use of agentic research tools. The document focuses on deep research, an agentic artificial intelligence capability that autonomously plans and executes multi-step tasks. It emphasizes the importance of human oversight and clear expectations when using any artificial intelligence tools. The resource aims to help agencies support evidence-based decision-making, improve efficiency and accelerate early-stage research and planning. It complements the Considerations for GenAI in Public Health. Deep research tools can conduct multi-step research across the web, analyze findings and generate citation-based reports with transparent reasoning. They can ask clarifying questions, refine their approach and synthesize information from multiple sources in real time. CDC's internal exploratory evaluation found the tools well-suited for literature synthesis on emerging health threats or intervention strategies, policy and legal scans across jurisdictions, strategic planning and scenario analysis, public health communications and message development, and comparative analysis of programs, interventions or regulations. The guidance details when deep research is appropriate. This includes tasks with clearly defined scope, rapid information synthesis from publicly available sources and projects where expert validation of outputs is possible. The document advises against use with restricted data sources such as paywalled journals or internal systems, sensitive or regulated data including personally identifiable information or protected health information, situations requiring statistical rigor or reproducible analytics, cases needing professional judgment such as legal or clinical decisions, and when comprehensive literature reviews or vague prompts are involved. Prompting considerations include defining scope and deliverables clearly with details on timeframe, geography, topic and desired format. They also cover supplying examples or templates, breaking down complex tasks into smaller sequential prompts, refining prompts through clarification and planning for subject matter expert review of both prompts and outputs.