What your instructor is told to do when your AI percentage is high

Turnitin's guidance for educators frames a high AI percentage as the start of a conversation rather than a verdict. It tells instructors to check institutional policy first, states that the tool's purpose is to facilitate conversations rather than to punish, and supplies four suggested opening lines — one of which is an invitation to open the report and go through the highlighted passages together. It also states that the final decision on misconduct rests with the instructor, not the tool.

They sit in the AI writing section of Turnitin's guides, alongside the technical documentation. One is titled "What should I do if the AI writing percentage is high?" and the other "What questions do I ask my student if AI is detected?"

Between them they run to under four thousand characters. Neither is hidden behind a login. And their existence is itself informative: the vendor has thought about the conversation as a distinct thing from the score, and has written a short script for it.

That matters to you because the conversation is where the outcome is actually decided, and because you can walk into it knowing what shape it was designed to have.

The order of operations, in its own words. First: "refer to your institution's policy on plagiarism and the use of generative AI in the classroom." Then: "you'll need a plan for approaching AI-generated text in your classroom."

Then the framing sentence, which is the one worth knowing verbatim:

"Rather than viewing the results of our AI writing detection as a punitive measure, the overarching purpose of the tool is to facilitate student conversations and interventions on the use of AI writing tools that may not otherwise occur, and use it as one strategy in your toolkit."

One strategy in a toolkit. Not a finding.

And the page closes by putting the decision somewhere specific: "The final decision on whether any misconduct has occurred rests with the reviewer/instructor. Turnitin does not make a determination of misconduct, rather it provides data for the educators to make an informed decision based on their academic and institutional policies."

Two practical consequences. Your institution's policy outranks the number, which means finding and reading that policy is not optional preparation. And nobody involved — including the vendor — describes the percentage as deciding anything by itself.

The second page describes itself as offering "conversation jump starters" for "a collaborative conversation with a student regarding potential AI misuse", and tells instructors to personalise them. All four, verbatim:

1. "When we look at your assignment, we can see [errors, apparent instances of AI misuse, etc.] in these places. Let's work together to find ways to avoid this/improve it in the future."
2. "Explain your process for completing this assignment. Let's look at areas that you are proud of and areas that you think could use some extra improvement."
3. "Let's examine your Turnitin AI writing detection report together. Let's look at the highlighted areas and discuss your choices. Let's brainstorm how to revise these sections of text in order to make the writing more authentically your own."
4. "What processes have worked well for you in your other classes? How can we consider implementing similar strategies in this class in order to optimize your work?"

Read them as a set and the pattern is hard to miss. Not one of them asks whether you used AI. Two of them ask about your process. One proposes opening the report and going through it together. One asks about your habits in other classes.

If the second question is going to be "explain your process", then your process is the thing to have ready — not a defence of the number.

Concretely, before the meeting:

1. Be able to narrate how the document got made. When you started, what order you wrote in, what you cut, what you were reading. This is ordinary recall, not evidence, and it is what question two asks for.
2. Have your drafting record to hand if you have one. It answers the same question with dates attached, and we go through where to find it in version history as evidence.
3. Read the report before the meeting if you have been sent it, because question three proposes going through it together and you do not want to be reading it for the first time in the room. Our walkthrough of what each part means is in how to read a Turnitin AI writing report.
4. Find your institution's actual policy. The vendor's own first instruction to your instructor is to consult it, and policies differ enough that the same percentage means different things at different institutions.
5. Have something to say about the specific highlighted passages, not the total. That is the granularity the conversation is designed around.

Being honest about the limits of the above, because overreading it would be its own mistake.

These are two short help-centre pages, not a procedure your institution has agreed to. Your instructor may never have read them. Your institution's own academic integrity process may be formal from the first email, and where that happens the collaborative framing above is not what you will get.

Nor is any of it a defence. "The vendor says this is meant to be a conversation" is a true statement about the vendor's documentation and a weak thing to say out loud in a meeting, because it argues about the process rather than about your work.

Where the situation is already formal — an allegation, a panel, a written response — that is a different problem with a different shape, and we set out what has actually persuaded reviewers in flagged but you wrote it.

Go back to the third jump starter: "Let's look at the highlighted areas and discuss your choices."

The highlighted areas. Not the percentage. In the conversation the vendor designed, the unit of discussion is the passage, and the question about each passage is what you chose to do there.

That is also the most useful way to look at your own report before anyone else does. A percentage is not something you can answer for. A specific paragraph is.

Sometimes the answer to "discuss your choices" is that a section genuinely needs rewriting, and then the question is how much of the document goes with it.

HumanPen narrows that by taking the report as the boundary: give it the file and the report, and the highlighted passages are what it edits, while paragraphs nobody marked keep the wording you already defended.

It does nothing for the conversation itself. That part is yours, and the material above is the whole of what we can hand you for it.

Does a high AI percentage mean I have been accused of something? Not by itself. Turnitin's guidance for educators states that the final decision on whether misconduct has occurred rests with the reviewer or instructor, and that the tool provides data rather than a determination.

What will my instructor most likely ask me? Turnitin's suggested opening lines centre on process: how you completed the assignment, which parts you are proud of, and going through the highlighted areas of the report together.

Should I quote Turnitin's own guidance back at my instructor? Probably not. It is accurate, but it argues about how the meeting should be run rather than about your work, and your work is what the meeting is for.

Does my institution have to follow this guidance? No. These are the vendor's help pages for educators. Turnitin's own first instruction is to refer to your institution's policy, which is the document that actually governs the process.