Direct answer
Direct Answer - ** An AI detector does not measure "originality" in the creative sense; it measures two separate, narrower things: how much of your text matches existing sources (similarity) and how much of your text was likely machine-generated (AI writing). Turnitin reports these as two independent indicators, so a document can be 0% similar and still carry a high AI score, or 40% similar and carry no AI flag at all [1]. "Originality" in detector language therefore means unmatched and human-attributed text, not "good ideas" — and neither number is a verdict on your integrity [2].
What Does "Originality" Mean to an AI Detector?
In Turnitin's architecture, originality is not a single score. The similarity indicator compares your submitted text against a comparison database of student papers, web pages, and publications, and reports the percentage of text that matched [2]. The AI writing indicator runs a separate language model over qualifying prose and estimates the share of text that appears machine-generated [1]. These pipelines share a report but not a method, which is why the two numbers can point in completely different directions.
The AI indicator also has a confidence floor that students frequently misread. When detection confidence falls below Turnitin's threshold, the report displays *% instead of an exact figure, signalling a low-confidence signal rather than a clean bill of health [1]. A *% result is best read as "no strong evidence either way," and it is the single most common source of confusion in student emails to integrity offices.
Crucially, neither indicator assesses originality of thought, argument, or research design. Turnitin's own guidance frames the reports as evidence for human judgement — an instructor reads the match list and the flagged passages in the context of the assignment brief, not the headline percentage [1][2]. An AI detector is a text-comparison instrument; the interpretive work stays with the marker.
Does a High Similarity Score Mean My Work Is Not Original?
No — a high similarity score usually means your document contains a lot of identifiable text, not that it contains stolen text. Reference lists, block quotations, standard methodology phrasing, module titles, and assessment rubrics all match the comparison database and all inflate the percentage [3]. Turnitin's own help documentation notes that instructors can exclude quotes, bibliography, and small matches precisely because those categories distort the headline number [2].
A second, less obvious cause is self-match. If a draft has previously been submitted to the same institutional account, the new submission can be compared against that earlier version and report matches to itself, pushing the score up without any external copying at all [3]. Students who re-check a revised draft and see the number jump often assume the worst when the cause is simply a stored prior submission.
The practical reading rule is to look at the match list, not the match total. A 30% score concentrated in a bibliography and three correctly cited quotations is a different document from a 30% score spread across unattributed body paragraphs, and only the second is an originality problem in any meaningful sense [2][3]. The percentage is an index into the evidence; it is not the evidence.
How Can I Check My Own Originality and AI Score Before Submitting?
You can see both indicators on your own draft before the deadline, and doing so converts an anxious guess into a specific, fixable list. Turnitin's published guidance on report interpretation is written for exactly this purpose — treating the report as feedback that supports integrity rather than as a punishment mechanism [4]. Reading your own match list early means you can attribute a source, rephrase an over-quoted passage, or add the citation you forgot while there is still time to act.
The order of operations matters. Fix similarity issues first, because rewriting a passage to improve attribution also changes the text the AI detector sees [4]. Then re-read the AI-flagged sections and ask whether the flagged prose genuinely reflects your own drafting process — long, uniformly paced, syntactically smooth paragraphs are the pattern detectors respond to, and they are also the pattern students produce when they paste model output without editing [1][4].
Previewing is also the only way to learn how your discipline's conventions interact with the detectors. A law essay and a lab report carry different baseline similarity, and no generic percentage threshold survives that comparison [2][4]. Treat the first report you see as a calibration run, not a grade.
Reading the report is the easy part; getting an accurate one before your deadline is where most students get stuck. Turnitin0 runs your draft through a non-repository check and returns the same two PDFs your marker would see — the AI detection report and the similarity report — usually in under 15 minutes, so you can act on the flags while the work is still yours to fix. Over 100,000 reports delivered to more than 20,000 students in the US, UK, Canada, Australia, New Zealand, and Ireland, with a 4.9/5.0 satisfaction rating. A single check is $3.80, and a 10-check pack brings it to $2.75 per check with no subscription.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
Get Real Turnitin AI & [Similarity Report](https://www.turnitin0.com/guides/us/ai-subscription)
FAQ
Is an AI detector the same thing as a plagiarism checker?
No. A plagiarism or similarity checker compares your text against a source database, while an AI detector estimates whether prose was machine-generated [1][2]. They run on different logic and produce different indicators on the same report.
Why does my report show *% instead of a number?
Turnitin displays *% when AI detection confidence falls below its threshold, which means the signal is weak rather than absent [1]. Treat it as inconclusive and look at the highlighted passages instead of the headline figure.
Can a fully human-written essay still get a high similarity score?
Yes. Heavy quotation, standard methodology phrasing, and reference lists all match the database legitimately [3]. The match list shows whether the overlap is cited material or unattributed prose.
Does a low AI score prove the writing is original?
It proves the detector found little machine-like text, which is a narrower claim [1][4]. Originality of argument is judged by your instructor against the assignment brief, not by a percentage.