ORISE Symposium 2026 Preview: James Dewar, Ph.D., says treat every AI prompt like an experiment

James Dewar, Ph.D., assistant professor of biochemistry at Vanderbilt University, uses Large Language Models (LLMs, a type of generative artificial intelligence) in his research. In a recent article published at nature.com, Dewar writes that to get more from AI researchers should treat every prompt like an experiment. In this episode of the ORISE Featurecast, Dewar talks in more depth about his article and gives a preview of his upcoming ORISE Symposium Seminar on the use of LLMs for science and research. He also discusses and shares some of the tools he has developed to help students and researchers better understand how to use LLMs effectively. For more information and to register for the ORISE Symposium, visit https://orise.orau.gov/news/events/symposium.html Read Dewar’s Nature Careers article: https://www.nature.com/articles/d41586-026-02083-6 Some of the links mentioned in the conversation are listed below: Papers showing AI automating the early stages of drug discovery https://www.nature.com/articles/s41586-026-10644-y https://www.nature.com/articles/s41586-026-10652-y Paper comparing AI to physicians at clinical reasoning https://www.science.org/doi/10.1126/science.adz4433 Preprint from Dewar on AI use for scientists: https://zenodo.org/records/19177101 Preprint from Dewar on how to teach with AI: https://osf.io/preprints/edarxiv/t8mea Dewar’s personal website with free tools: https://www.jmdewar.com/ This is the specific meta prompt tool he mentions: https://prompting.jmdewar.com/meta Links not mentioned in the Featurecast but that may be of interest: Dewar’s ORCID and Linkedin: https://orcid.org/0000-0003-4153-3818 https://www.linkedin.com/in/james-m-dewar/ Links to the graduate biomedical programs at Vanderbilt University School of Medicine. https://medschool.vanderbilt.edu/igp/igp-program-overview/ https://medschool.vanderbilt.edu/qcb/qcb-program-overview/