NCURA 2026 Region VI/VII Regional Meeting: The Good, the Bad, and the Ungoverned: Evaluating AI Outputs at Scale – Artificial Intelligence for Research Administration

Tuesday, November 10 | 4:00 PM | Platte | Intermediate

Presenter: Barrie Robison, University of Idaho
As RA offices use AI for document review, data extraction, drafting, and policy interpretation, they need practical ways to evaluate AI outputs across many tasks, documents, prompts, and tools. This session shares an emerging AI4RA evaluation framework that covers:

Testing many examples
Comparing results to validated answers
Checking consistency across repeated runs
Tracing outputs back to their sources