PAIRR: peer and AI review + reflection

Recommended for: Any course with a substantial written assignment that goes through drafts

Background for the instructor

Students who are handed AI feedback tend to treat it the way they treat a spell checker: read once, apply, move on. PAIRR breaks that habit by making AI the second reader rather than the first, and by making the comparison between the two kinds of feedback an assignment in its own right. The order matters. A peer reads for whether the argument lands; a model reads for whether the text looks like the kind of text it has seen. Students who meet the second reader first rarely notice the difference.

The approach comes out of writing instruction but nothing about it is specific to writing. Any draft that can be reviewed by a peer can run through the same five steps.

Preparation

Write the AI review prompt yourself rather than leaving students to improvise one; the quality of the second round depends on it. Decide in advance which tool students use and what happens to their text in it.

Procedure

1. Students read and discuss a short text on AI and language, including whose English a language model treats as standard.
2. Students carry out a peer review of each other's drafts, on paper or in the VLE, before any tool is opened.
3. Students run the same draft past an AI reviewer using the prompt you supplied.
4. Students write a reflection comparing the two sets of comments: where they agreed, where they conflicted, which one understood the assignment, and what each reader could not see. They are asked to talk back to the feedback rather than accept it.
5. Students revise, and state which comments they acted on and which they rejected.

Discussion questions

• Where did the peer and the model contradict each other, and which of them was right?
• Which piece of feedback did you reject, and what made you confident enough to reject it?
• What did your peer notice that no model could have noticed?
• Whose English did the model appear to treat as correct?

Assessment opportunity

Grade the reflection, not the revision. What you are looking for is evidence of judgement about feedback, which is the transferable skill here.

Reference

• Peer and AI Review + Reflection (PAIRR), University Writing Program, UC Davis. Developed by Carl Whithaus and Marit MacArthur with Emerson Case (CSU Bakersfield), Julie Gamberg (Glendale Community College), Hogan Hayes (Sacramento State), Anna Mills (College of Marin), Lisa Sperber and Hao-Chuan Wang (UC Davis), and funded by the California Education Learning Lab. Piloted with 654 students in 2023–24. writing.ucdavis.edu/pairr
• The project uses MyEssayFeedback, a not-for-profit tool with an open library of instructor-written prompts.
• Discussed in The Norton Guide to AI-Aware Teaching (Vee, Watkins & Bruff, W. W. Norton, 2026).