AI4RA workshop companion for NCURA Region VI/VII 2026

Sessions

Three hours, two parts.

Nathan opens with ninety minutes on how AI works and where it fits in research administration. After a break, Bert leads an hour on thinking well with AI: an ethical discussion mini-workshop and the NARC reasoning puzzles. The last half hour is open discussion. Slides and activities will be linked here as they are ready.

Learning objectives

By the end of the session, participants will be able to:

  1. Explain how large language models generate output, which tasks they perform well, and the ways they fail.
  2. Distinguish research administration activities that benefit from AI assistance, those that depend on institutional data infrastructure, and those that should remain human-driven.
  3. Use open-source AI4RA tools hands-on: a prompt engineering trainer, a document extraction and review workflow, and a data lakehouse architecture.
  4. Evaluate AI-generated output and judge institutional data readiness before relying on AI-assisted results.
  5. Identify appropriate patterns for verification and human oversight.
  6. Lead a structured ethical discussion about AI use that can be taken back to your own institution.

Session plan

Choose a part to open its guide and slides.

Part 1. Nathan Layman. 90 minutes

Where AI fits in research administration

An overview of modern AI systems: how large language models generate output, the tasks they perform well, and the ways they fail. Then a sorting exercise: which research administration activities benefit from AI, which depend on institutional data infrastructure, and which should remain human-driven. Hands-on throughout with open-source AI4RA tools.

  • How LLMs work: token prediction, context, and why models hallucinate
  • Promptulus: a prompt engineering trainer
  • Vandalizer: a document extraction and review workflow
  • The data lakehouse: preparing institutional data for AI, and why your data model matters more than any AI model
Slides coming soon
Part 2. Bert Baumgaertner. 60 minutes

Thinking well with AI: ethics and reasoning puzzles

A philosopher's guide to using AI responsibly. A structured ethical discussion mini-workshop, built on a discussion tool participants can take back to their own institutions, plus NARC, Narrative Augmented Reasoning Challenges, where the goal is to write the narrative that gets a model to do the right thing.

  • Ethical discussion mini-workshop: a format you can run yourself
  • Keeping a human accountable for every AI-assisted decision
  • Puzzle activity: steering a model with narrative, and seeing where it still fails
  • What verification and oversight should look like before you rely on a result
Slides coming soon
Wrap-up. 30 minutes

Break, discussion, and take-home plan

A short break between the two parts, then open discussion to close. Participants leave with one concrete task to try, one evaluation rule for checking the output, and a policy stance they can bring back to their own institution.

  • Questions for both presenters
  • Decide what to automate, augment, or leave alone
  • Workshop evaluation survey