Direct answer
An AI originality checker is a tool that runs two separate measurements on the same document: a similarity check that compares your text against web pages, publications, and previously submitted papers, and an AI-writing check that estimates how much of your prose looks machine-generated [1]. The "originality" figure you see is the similarity index, not an AI score — the two numbers come from different models, answer different questions, and can move independently of each other [2]. A paper can score 0% similarity and still be flagged heavily for AI writing, which is why reading only one number gives a misleading picture of your risk [3].
What Does an AI Originality Checker Actually Measure?
The word "originality" is where most confusion starts. In academic-integrity tooling, originality checking means source matching: the system fingerprints your submission and compares it against an index of web content, scholarly publications, and student papers submitted previously, then highlights any passages that overlap with those sources [1]. The output is a similarity index — a percentage of your document that matched something else — together with colour-coded highlights and links to the matched sources. Nothing in that process evaluates whether you wrote the sentence; it only evaluates whether the sentence already exists somewhere in the index.
AI writing detection is a genuinely different measurement. Turnitin's detection model analyses linguistic patterns associated with large language models and returns a percentage of qualifying prose it considers AI-generated, along with a sentence-level breakdown showing which passages triggered the signal [2]. Because it reads style rather than sources, it can flag text that matches nothing in the index at all. That is the single most important thing to internalise: a clean similarity report is not evidence of human authorship, and a high similarity score is not evidence of AI use [2].
The two layers therefore answer two questions that students routinely collapse into one. Similarity asks "does this text already exist elsewhere?" AI detection asks "does this text read like a model wrote it?" A document drafted entirely by ChatGPT on a niche topic can return 0% similarity and a high AI score; a heavily quoted literature review can return 30% similarity and 0% AI. Treating "originality score" as a synonym for "AI score" is the most common misreading, and it is the one that gets students into trouble [1][2].
How Is the Originality Percentage Calculated, and How Should You Read It?
The similarity index is a ratio, not a verdict. Turnitin divides the matched text by the total word count of the submission, which means the number is sensitive to length, to how much you quote, and to whether your bibliography and references are included in the calculation [1]. A 12% score on a 4,000-word dissertation and a 12% score on a 900-word essay are not equivalent risks. When you open a report, read the match list before you read the headline number: a score driven by your reference list and three properly cited block quotes is a completely different situation from a score driven by six uncited paragraphs in your discussion chapter.
The AI percentage needs the same scepticism, with an extra wrinkle. Turnitin reports an overall percentage of qualifying text flagged as AI-generated and highlights the specific sentences it attributes to a model [3]. Crucially, when the AI signal falls below the detector's confidence threshold — under 20% — the report displays a *% placeholder rather than an exact figure. That asterisk means low-confidence signal, not verified human writing. Students frequently read *% as a pass, then discover their instructor reads it as an unresolved flag. The report is explicitly probabilistic: both false positives and false negatives are documented, and Turnitin itself advises against using the score as the sole basis for an integrity decision [3].
Practical reading order, then: check the AI percentage and the highlighted sentences first, because those are the passages you would have to rewrite; then check the similarity index and its match list, separating legitimate citation from genuine overlap; then check whether either number sits near a threshold your institution actually enforces. Score bands only mean something relative to your own department's policy, and policies vary widely between universities and even between modules [3].
Can You Preview the Same Report Your Institution Sees Before You Submit?
Normally, no — and that is the structural problem. Turnitin checks are generated when a student submits through the course assignment in the institution's learning management system, so the first time most students see their similarity index or AI percentage is after the deadline has passed and the report has already been attached to their submission [4]. There is no student-facing button inside the institutional system that says "run this on my draft and let me revise." Good practice guidelines increasingly ask institutions to tell students in advance when AI detection is switched on and how results will be used, precisely because the alternative is students discovering the rules only through a penalty [4].
That gap is why pre-submission checking exists as a category. The useful version of it reproduces the institutional output rather than approximating it: the same similarity index conventions, the same AI percentage, the same sentence-level highlighting, the same *% low-confidence behaviour — so that what you revise against is what your marker will actually read [4]. Anything that returns a single invented "originality score" without a match list or a sentence breakdown gives you nothing you can act on, because you cannot see which paragraphs caused the number.
Understanding the report conventions before you submit also changes how you revise. If your AI percentage is concentrated in three transition-heavy paragraphs, the fix is rewriting those paragraphs in your own voice, not paraphrasing the whole document. If your similarity index is inflated by quoted material, the fix is citation formatting, not rewriting. Knowing which lever to pull is only possible if you have seen the report layout in advance [1][4].
Reading the numbers is one thing; seeing your own numbers before the deadline is another. turnitin0 runs a pre-submission check on your .docx, .pdf, or .txt draft and returns the Turnitin AI detection report and the similarity report together as two downloadable PDFs — the same report format your professor opens in the LMS, delivered in under 15 minutes in 98% of cases. The file is checked without being added to Turnitin's student paper database, so your draft stays yours, and you can delete it from your account afterwards.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is an originality score the same as an AI score?
No. The originality figure is a similarity index measuring overlap with existing sources, while the AI score estimates how much prose reads as machine-generated; they come from separate models and can diverge sharply [1][2].
Why does my report show *% instead of a number?
Turnitin displays *% when the AI signal falls below its confidence threshold, which is under 20% — it signals low-confidence detection, not confirmed human writing [3].
Can a paper be 0% similar and still be flagged for AI?
Yes. AI detection reads linguistic patterns rather than sources, so text that matches nothing in the index can still carry a high AI percentage [2].
Will my pre-submission check get stored in Turnitin's student paper database?
With turnitin0 it is not: the draft is checked without being added to the student paper database and is not shared with third-party databases, and you can delete the file from your account.
Should I rewrite everything if my AI percentage is high?
Usually not. Target the highlighted sentences and paragraphs the report identifies, since the score is concentrated there and rewriting unflagged, properly cited material rarely changes it [3].