Four things to bring, and two of them were written by the company that makes the detector.
None of this is about arguing that the tool is broken. That argument loses, and it loses for a specific reason I go into below — the flaw most people reach for was fixed in 2023 and the fix is documented. This is about the difference between a meeting that is about a number and a meeting that is about a document.
The short answer
If you have been called in about a Turnitin AI score on your dissertation, you have four things to bring to that meeting, and none of them require you to claim the tool does not work. The first is a sentence from Turnitin's own documentation saying an AI score should not be used as the sole basis for adverse action. The second is the report itself, which you are entitled to see. The third is your version history. The fourth is another sentence from Turnitin, this one saying that if your text has certain features common in dissertations, the high score should be discounted.
That is the outline. The rest of this article explains where each card comes from, what it does, and what it does not do. None of it teaches you how to deny using AI. It teaches you how to make the meeting about evidence and procedure rather than about a single number.
The distinction matters because the number is the thing you cannot argue with on its own terms. You did not compute it. You cannot inspect the model. What you can do is ask the people who did compute it to follow their own guidance, and bring the material that only you have.
Card one: Turnitin says the score should not be the sole basis
Turnitin's AI writing report guide says:
"Our AI writing detection model may not always be accurate (it may misidentify human-written, AI-generated, and AI-paraphrased text), so it should not be used as the sole basis for adverse actions against a student."
The sentence immediately after it is the one that gets dropped every time this quote is used:
"It takes further scrutiny and human judgment in conjunction with an organization's application of its specific academic policies to determine whether academic misconduct has occurred."
Read both sentences together. The first says do not use this number alone. The second says what should happen instead: further scrutiny, human judgment, and the institution's own policies. That is not a student talking. That is the company that built the tool telling the person holding the number to do more work before drawing a conclusion.
The FAQ page says the same thing in its own words: "the percentage on the AI writing indicator should not be used as the sole basis for action or a definitive grading measure by instructors." And a release note from December 2023 repeats it: "an AI Writing score should not be used as the sole basis for adverse actions against a student." Three pages, same instruction.
There is also a separate page titled How should I review the AI Writing report? which puts it differently:
"It is not meant to provide definitive answers in isolation. More important than any tool is the educator who sees the score and makes decisions balancing this information with their personal knowledge of their students, their work, and institutional policy."
The next sentence completes it:
"When educators look at the AI writing score and utilize it as a single data point rather than a definitive response, then it is being used as intended."
That is the company calling its own score a single data point. Not a verdict. Not a definitive response. A single data point. If your meeting starts with a number and no further discussion, the person running it is using the tool in a way the manufacturer says it should not be used.
What this card does not do: it does not say the number is wrong. It says the number alone is not enough. That is a weaker claim and a sturdier one.
Card two: you have the right to see the report
Turnitin's FAQ states two things that matter to you right now:
"Please note, only instructors and administrators are able to see the indicator."
And on the same page:
"The AI writing detection indicator and report are not visible to students."
The second sentence is followed by one that changes it:
"However, with the PDF download feature, instructors can download and share the AI report with students."
Put those together: you cannot see the AI score yourself by logging in. Your supervisor can. And your supervisor can download the report as a PDF and hand it to you. If nobody has done that, you are sitting in a meeting about a number you have never seen on a page you have never opened.
So the first request is simple. Ask for the PDF of the AI writing report. Not a screenshot of the percentage. The full report, because the full report shows you where the highlights sit, how many words were analysed, and whether the score is a real number or an asterisk.
If the number is between 1% and 19%, Turnitin does not display a percentage at all. It shows an asterisk. A forwarded number in that range is not something the report itself commits to.
What this card does: it moves the conversation from a claim about a number to a discussion of a document you can both look at. What it does not do: it does not tell you what the number will be. You might see a high score. But at least you will know what it is attached to.
Card three: your version history is timeline evidence
If you wrote your dissertation in Word, Google Docs, or Overleaf, the platform kept a version history. That history is a timestamped record of when changes were made to the document over days, weeks, or months. An AI-generated draft pasted in one go looks different from a document that accumulated edits across dozens of sessions.
This is the one piece of evidence that only you have. The institution does not have your version history. Turnitin does not have it. Your supervisor might have seen drafts at various stages but probably does not have the full edit log. You do.
What to bring:
- The version history itself, either exported or shown on a laptop.
- The earliest draft you still have, even if it was rough.
- Any drafts you shared with your supervisor by email or through a shared document, with the dates those were shared.
- Notes, outlines, or annotated PDFs you produced while working on the dissertation. These are the residue of a writing process, and their dates trace a path that does not look like a single generation event.
The point is not to prove a negative. You cannot prove that no AI was involved at any stage. The point is to show that the document has a history consistent with being written by a person over time. That is evidence the score cannot address, because the score only looks at the final text.
What this card does not do: it does not disprove anything. A version history can show that editing happened, and editing is consistent with a document that was drafted by a person, revised with feedback, and polished before submission. It is also consistent with someone pasting an AI draft and then editing it. What makes it useful is that it shifts the question from "is this number right" to "what does the full picture look like," and the full picture is something only you can present.
Card four: Turnitin says some text types get inflated scores
This is the card most people do not know they have. Turnitin's FAQ page lists characteristics of text that tends to produce false positives:
"Sometimes false positives (incorrectly flagging human-written text as AI-generated), can include content without a lot of structural variation, text that literally repeats itself, or text that has been paraphrased without developing new ideas."
The next sentence is the one that matters in your meeting:
"If our indicator shows a higher amount of AI writing in such text, we advise you to take that into consideration when looking at the percentage indicated."
Read what that covers. Content without a lot of structural variation. Text that literally repeats itself. Text that has been paraphrased without developing new ideas. Those three describe a large share of dissertation writing, especially literature reviews and methodology sections. A literature review summarizes existing work, which is a paraphrasing task by nature. A methodology section follows a standard structure with repeated phrasing. Neither is supposed to introduce new ideas, because the new ideas belong in the analysis and discussion chapters.
So if your flagged sections are literature review or methodology, the company that made the tool is telling the person reading the score to discount it for exactly that reason. Not you. The company.
This is also where you should be careful. Turnitin also published a release note in May 2023 acknowledging that false positives had been concentrated in the first and last few sentences of documents:
"Since launch, we have observed a higher incidence of false positive detection in the first few or last few sentences of a document. Many times these sentences consist of introduction or conclusion content written in a generic way. As a result, we have changed our detection logic to help reduce these false positives."
The key phrase is the last one: "we have changed our detection logic to help reduce these false positives." This was a fix announced in 2023. It is not a current weakness you can point to. It is evidence that the tool changes over time, which means two things for your meeting. First, a report generated six months ago may have been produced under different logic than one generated today. Second, arguing that the tool is broken is arguing against a target that has moved. Your position is stronger when you say the tool is improving and your case deserves to be looked at on its own terms, not when you say the tool is useless.
What not to do
Three things will hurt you in this meeting, and all three are tempting.
Do not say Turnitin is inaccurate. The company itself says its model may not always be accurate, and it says the score should not be the sole basis for action. That is your lever. If you go further and claim the tool does not work, you are making a claim you cannot support, and you are picking a fight with the institution's chosen system. The person across the table will dig in. The right stance is that the tool has a documented marginal error rate and documented guidance on how to handle it, and you are asking them to follow that guidance.
Do not refuse to talk. Silence reads as guilt. You have evidence to present. If you walk out, the meeting happens without your version history, without your drafts, and without anyone hearing the sentence about text types. The only way the cards on this list work is if you show up and put them on the table.
Do not bring the old "first and last sentences" argument. As we said, that was a 2023 fix, not a current flaw. If you bring it as evidence that the tool is currently broken, the person across from you can look it up, find the fix, and you lose credibility on everything else you brought.
How to prepare
Before the meeting, gather these things:
- The AI writing report as a PDF. Request it if you do not have it. You need to know what the number is, where the highlights sit, and whether it is a percentage or an asterisk.
- Your version history. Export it or have it ready to show on a laptop. If you used Overleaf, the page history is there. If you used Google Docs, the version history is in File > Version history. If you used Word, it depends on whether you had OneDrive autosave on, but even without it, your drafts and their file modification dates tell a story.
- Your writing process material. Notes, annotated PDFs, outline drafts, email threads with your supervisor, feedback on earlier drafts. Anything that shows the dissertation was not produced in a single sitting.
- A list of specific questions you want answered. Not a script, not a speech. Questions. "Which sections were flagged?" "Was the report generated before or after the May 2023 logic change?" "Does the flagged text fall under the categories Turnitin says can produce false positives?" Questions like these keep the meeting in a space where you are participating in an investigation rather than defending against an accusation.
The meeting is not a trial. It is supposed to be a review, and Turnitin's own documentation says it should involve further scrutiny and human judgment. Your job is to bring the material that makes that review possible.
Where we sit
HumanPen is a document rewriter. We are not going to tell you that running your dissertation through a rewriter will make an AI score disappear, because we do not make claims about future scores. What we can say is what the tool does: you upload the document with its AI writing report, and only the passages the report flagged are rewritten. The rest of the document stays verbatim. You review the output before it becomes anything final.
If you are in this situation because a report already came back high, and you want to revise the flagged sections before your meeting, that is one path. If you would rather go into the meeting with the original document and your version history and let the evidence speak, that is another path. Both are legitimate. The meeting comes first.
Frequently asked questions
Can I refuse to attend the meeting? You can, but it is the worst move you can make. The meeting is where you present your evidence. Without you there, the institution makes its decision based on the score alone, which is exactly the outcome Turnitin says should not happen.
What if the score is 100%? A score of 100% is not evidence that every sentence was generated by AI; it is a statistic over the qualifying prose the model was willing to read, on one date, under one model version. What matters in the meeting is the same: request the full report, bring your version history, and ask the specific questions.
Does version history prove I did not use AI? No. It shows that the document was edited over time in a pattern consistent with human writing. It cannot prove a negative. What it does is give the reviewer something to look at besides the score, and that is its value.
What if I did use AI for part of the dissertation? This article does not teach you how to deny using AI. If AI was used for some part of the work, the meeting is where that gets discussed alongside everything else. The documentation cited here, the version history, and the text-type guidance all apply regardless, because they are about how the score should be interpreted, not about whether AI was used.
Will running my paper through a rewriter fix the score? We do not make claims about future scores. What HumanPen does is rewrite the flagged passages you select. Whether the next report comes back lower is something we will not promise, because the model changes and the report is produced by Turnitin, not by us.
Bring the report, bring your history, bring your questions
If a Turnitin AI report has flagged your dissertation, the meeting is where you present your evidence. If you want to revise the flagged sections beforehand, upload the document with its report and only those passages get touched.
Originally published at HumanPen: https://humanpen.net/blog/accused-of-using-ai-on-dissertation?utm_source=devto&utm_medium=article&utm_campaign=d69
