Guides

Practical notes on AI text detection: what the scores mean, how the tools behave, and what to do about it.

Being flagged

What a detection score means, and what to do when human writing is flagged.

• Why AI detectors flag human writing
  Not a bug that a better version will fix. It follows directly from what these tools are, and some human writing genuinely has the properties they measure.
• Does Grammarly get flagged as AI?
  Grammar checking and AI drafting are different things, but detectors cannot tell them apart, because both leave the same fingerprint.
• Falsely accused of using AI
  What the score behind the accusation actually is, how to get an independent measurement, and where the record of your writing process already exists without you having planned for it.
• How Turnitin's AI detection works
  What the AI writing indicator measures, why it is nothing like the similarity score sitting next to it, and where the number stops being reliable.
• Turnitin's AI false positive rate
  What Turnitin publishes about how often its AI detector is wrong about human writing, and how that compares to a detector calibrated for a low false positive rate.
• What a Turnitin AI score means
  It is not a confidence level. A 40% does not mean the report is 40% sure. It means something quite different, and the difference matters.
• What to do when your writing is flagged
  A practical sequence: work out what the score actually says, get a second measurement, and assemble the record of process that the text itself does not contain.
• Why non-native English writing gets flagged more often
  The measured effect, the linguistic properties behind it, and why it follows directly from what detectors measure rather than from any rule about who wrote the text.

How detection works

The mechanics behind AI text detection, and why detectors disagree.

• How AI detectors work
  There is no database of generated text and nothing to match against. A detector is a classifier making a guess from statistical properties, and that explains almost everything about what it can and cannot do.
• Checking a PDF or Word document for AI detection
  How a file gets turned into text decides part of your score before any model sees it, and most tools will not tell you what they extracted.
• Documents that are part human, part AI
  The most common real case is also the one detectors handle worst, and there is a published number showing exactly how it goes wrong.
• Perplexity and burstiness, explained
  The two terms every AI detection explainer uses and most of them define badly. Both are simple, and knowing what they measure tells you exactly when they fail.
• Why AI detectors are unreliable on short text
  Every detector is several times worse per sentence than per document. It is not a flaw in any particular tool, it is how much evidence a sentence contains.
• Why AI detectors disagree
  Run one paragraph through four tools and you can get 12%, 40%, 61% and 80%. None of them are broken. There are five specific reasons.

Academic writing

Detection in papers, theses and journal submissions.

• AI detection in academic papers
  Academic writing scores higher than almost any other kind, whoever wrote it. Detectors measure regularity, and papers are regular on purpose.
• AI detection and a thesis or dissertation
  A long document written over years, edited by several people, is a specific and unusually awkward case for a classifier.
• AI detection and journal submission
  Publishers screen submissions now, and the register that gets flagged is the one journals ask you to write in.
• Checking an essay for AI detection before you hand it in
  What a pre-submission check can and cannot tell you, and which parts of an essay raise a score without anyone having generated anything.
• Why LaTeX breaks AI detectors
  Paste a .tex file into most detectors and you get a number. That number is mostly measuring your preamble.
• Will my paper be flagged as AI?
  You can find out before you submit rather than after, and the sections most likely to be flagged are predictable enough to check deliberately.

Tools compared

How the detectors and rewriting tools actually behave, measured.

• Comparing AI detectors: four things worth checking
  Everyone advertises 95% plus. Those numbers are not comparable to each other. These four properties are, and they actually separate the tools.
• Are AI detectors accurate? Yes, and the number is close to useless
  Every detector advertises 95% plus. Those figures are not comparable, and accuracy is the wrong thing to ask about anyway. Here is what to ask instead.
• ChatGPT detectors: what they actually detect
  Nothing detects ChatGPT specifically. Detectors read properties of text, and those properties are shared across models, which is why the same tool that catches one family misses another.
• Free AI detectors: what you give up, and what you do not
  Free usually means one of four business models, and which one you are using determines what happens to the text you paste in.
• Turnitin and GPTZero give different answers. Here is why
  Two detectors on the same essay routinely disagree. That is structural, not a sign that one of them is broken, and the disagreement itself is the useful output.

Editing drafts

What makes text read as machine written, and how to edit it.

• AI rewriter: what actually changes
  Most rewriting tools do one of two things. One of them barely changes anything, and it is the one most tools do.
• How to make text sound less like a machine wrote it
  Five changes, each of them measurable, each of them something you can do without altering what the text says.
• Paragraph rewriting
  The paragraph is where rewriting starts working, because it is the smallest unit that can move material between sentences.
• Rewording tools: what they change and what they quietly break
  Most rewording tools are a thesaurus with a text box. That is fine for a marketing email and actively dangerous on anything with numbers in it.
• Sentence rewriter: what one sentence can and cannot fix
  Rewriting sentence by sentence changes less than you would expect, and the reason is that the thing detectors measure lives above the sentence.
• What makes text read as machine written
  The specific, checkable properties that make prose look generated: sentence rhythm, connective density, hedging, and vocabulary that models over-select.
• Why paraphrasing tools do not reduce detection
  Synonym substitution changes the words a detector reads but not the properties it measures, and in several cases it makes a text score higher than the original did.

Teaching

Detection scores in the classroom, and what they can support.

• What AI detectors do teachers use
  Mostly whichever one was already installed. That single fact explains most of what people find confusing about this.
• AI detectors for teachers: what the number can carry
  A detection score is a probability with a published error rate. Here is what that means for a class of thirty, and for a cohort where many students write in a second language.