How Colleges Can Help Faculty and Staff Learn to Use AI

Faculty and staff need practice, coaching, and clear responsibility for AI use in teaching, research, and administrative work.

By RUDI · Published November 1, 2025
Updated September 22, 2026

Start with a task people recognize.

A faculty member preparing discussion questions and an admissions team answering routine inquiries need different kinds of AI support. Each needs to choose appropriate source material, check the result, and decide whether the work improved. A campus subscription gives them access. Applying it takes time and judgment.

Begin with a recurring task and a concrete standard for success. An instructor might use approved course readings to draft discussion questions, then check whether the questions reflect the text and serve the learning objective. This is an illustrative starting point, not a reported campus result. The instructor remains responsible for what reaches students.

UNESCO’s guidance for generative AI in education and research calls for institutions to validate pedagogical and ethical suitability and build human capacity. That gives campus leaders a practical question: who will help people assess a proposed use before it becomes routine?

Teach the controls and the judgment.

Onboarding should explain the approved account, access settings, permitted information, and where to ask for help. Discipline-specific practice should then cover what a plausible but incorrect answer looks like and how to check it.

In a workshop, ask participants to bring a permitted task, compare the generated result with its sources, and record what they changed. Follow up after they have tried the process in their own work. A person who can identify a missing qualification or an invented citation has learned something a login count cannot show.

Use fictional or institution-approved material for initial exercises. Keep student records, unpublished research, and personnel information out of those exercises unless the institution has explicitly approved the data and environment. An appealing demonstration does not establish that a workflow is suitable for sensitive work.

Organize support around the three Cs.

Compliance: Publish usable instructions: approved tools and data, required review, disclosure expectations, and a route for questions. Give every AI workflow a responsible human owner.
Competency: Help faculty and staff frame tasks, select sources, verify outputs, and recognize when to stop. Offer coaching, peer review, and examples from the work they actually do.
Cost: Budget for licenses, accessibility, staff time, coaching, technical support, and evaluation. Include the time spent correcting output when comparing a pilot with the existing process.

The three Cs guide for people leaders explains how these responsibilities connect. Revisit them together when the tool, the information it accesses, or the task changes.

Give faculty and staff a dependable place to get help.

The teaching and learning center, library, IT team, academic leadership, and data or privacy owners each bring part of the expertise. Name a lead who can coordinate their work and route questions. A new center is one option; a faculty fellow with protected time or a shared service across institutions may fit a smaller program.

Support needs a schedule and capacity. Set office hours, maintain examples with review dates, and provide a way to report a confusing or harmful result. Student fellows can contribute examples and peer support, with continuing staff responsible for access and program continuity.

When a product changes, identify which campus workflows depend on the changed feature. Retest those examples and update the guidance. This is more manageable than asking every instructor to track every AI announcement.

Measure whether the work improves.

Choose a small pilot and record a baseline: time to complete the task, quality of the result, corrections required, and effects on the people receiving it. Compare equivalent work and include review time. Faster preparation is useful only if the material still serves its purpose.

For teaching, separate staff efficiency from student learning. Producing a lesson faster does not demonstrate that students understand more. Ask what evidence would justify continued use, and what would cause the team to revise or stop the workflow.

RUDI’s RESPECT framework provides a review lens: Responsible, Equitable access, Safety, Privacy, Efficient and effective, Control, Transparency. Check who owns the result, who can access the support, and how affected people can question or correct an AI-assisted outcome.

Fund the people who make adoption workable.

A workshop can introduce a useful practice. Ongoing support gives people somewhere to bring the failures, exceptions, and questions that appear afterward. Budget for that work, give the program lead time to do it, and use the pilot evidence to decide what deserves broader adoption.