Start with the decision
Use the 2 a.m. paper scenario. Ask what learning must remain visible.
Faculty facilitation guide
Teaching Responsible AI Use with RAFT
A ready-to-run 60-minute faculty development session
Session promiseParticipants leave with a common language for learning-centered AI use and one assignment-level change they can implement immediately.
Learning outcomes
- Distinguish use that augments learning from use that replaces it.
- Explain and apply the four RAFT steps.
- Identify where an assignment needs clearer AI expectations.
- Select one piece of process evidence that makes learning visible.
- Draft concise, student-facing AI-use language.
Run of show
Augment versus replace
Use the GPS analogy and the Learning Impact Test.
Teach RAFT
Research first, augment deliberately, fact-check, and transform through original reasoning.
Work a case
Identify one appropriate use at each step, one replacement risk, and one useful piece of process evidence.
Audit an assignment
Connect the learning goal, AI boundaries, verification, and evidence.
Draft student-facing language
Write concrete permitted, ask/disclose, and not-permitted uses.
Commit and close
Each participant names one course change and one support need.
Assignment audit
- What learning or performance is this assignment assessing?
- Which parts must remain fully human to preserve that learning?
- Where could AI provide productive practice, feedback, or extension?
- What uses need permission, disclosure, or prohibition?
- What process evidence would make learning visible?
- How will students verify and briefly defend the work?
Copy-ready syllabus language
Learning-centered AI use: AI tools may be useful for feedback, practice, brainstorming, and exploration, but they should not replace the thinking or performance a task is designed to develop. For each assignment, follow the stated AI-use expectations. When AI use is permitted, apply the RAFT process: research independently first, augment your work, fact-check everything, and transform the result into your own work. You remain responsible for the accuracy, sources, reasoning, and integrity of anything you submit. If expectations are unclear, ask before using AI.
Facilitation cautions
- Tie every boundary to the learning outcome, not fear of a tool.
- Do not imply that one traffic-light list applies to every course.
- Make room for accessibility, language-learning, privacy, and disciplinary contexts.
- Choose process evidence that supports learning without becoming surveillance or busywork.