FAES Science Insight Series: Trust Through Knowledge Grounding: AI Agents in Biomedicine
To Know
About this Class
This series invites Principal Investigators, Senior Scientists, and Senior Clinicians to share cutting-edge research and developments in their fields. Each session includes a 20-30 minute presentation followed by a Q&A or journal club discussion, fostering deeper insights and scholarly exchange. Lunch is provided. Please note this event is only open to members of the NIH community.
Recent advances in large language models (LLMs) have enabled powerful AI agents for biomedical research, yet their adoption in high-stakes settings remains limited by concerns about hallucination, opacity, and reliability. In this talk, I discuss how expert-curated domain knowledge can be used to help mitigate these challenges in general-purpose LLMs. Drawing on real-world systems and case studies such as GeneAgent (Nature Methods 2025), I will highlight design principles for building AI agents that are scientifically sound, interpretable, and suitable for biomedical research and clinical applications.