Notes for teachers, lecturers and school leaders on reading how students actually work with AI, instead of guessing from the finished text.
An AI-detection percentage is not evidence a panel can act on. What is: the record of how the work was produced — and how to get one for a single assignment.
Brisk Inspect Writing replays how a Google Doc was written. Cognity records the AI conversation itself. What each one shows a teacher, and when to use which.
When AI can produce any finished text, the finished text stops being evidence of learning. A process portfolio is what replaces it, and it is more useful than the essay ever was.
The IB requires AI-generated work to be credited in the text and bibliography. Here is a nine-point checklist for coordinators turning that requirement into a policy teachers can run.
School-wide AI policies are too broad to teach from. Here is how to write AI rules for a single assignment, with three worked examples a teacher can copy.
A script for the conversation after a student hands in AI-generated text. Not an accusation, not a tribunal. Five questions that lead somewhere.
Grammarly Authorship lets a student prove how a document was written. Cognity lets a teacher set the rules and read the whole process. An honest comparison of what each record contains.
The honest answer: you mostly can't tell from the finished text, and detectors won't fix that. Here is what you can actually look at instead.
Both tools show teachers how student writing was produced. They differ in where the AI lives, who buys them, and what the record contains. An honest comparison.
The IB requires students to credit AI-generated work in the text and bibliography. Here is a plain acknowledgement template your students can use, and how to make it automatic.