Direct answer
Turnitin's AI detector is the AI writing detection feature built into the Turnitin Similarity Report, and it flags the proportion of a submission's qualifying prose that its model classifies as AI-generated [1]. It is not a standalone tool and it does not declare authorship — it produces a percentage that only appears once enough of the text triggers the detector [1]. Understanding what that number does and does not mean is the difference between a useful signal and a panic attack.
What Is Turnitin's AI Detector And What Does Its AI Writing Score Actually Measure?
Turnitin's AI writing score measures the proportion of qualifying prose segments in a document that its detection model classifies as likely AI-generated [2]. The model does not read for meaning or plagiarism; it analyses statistical patterns in the writing itself, weighing signals such as perplexity (how predictable the next word is) and burstiness (how much sentence structure varies) [2]. Text produced by large language models tends to be smoother and more uniform across those measures than text a person writes in one sitting.
The score is reported per qualifying segment rather than as a single judgement about the whole paper [2]. Turnitin only evaluates long-form English prose, so short answers, bullet lists, equations, and non-English passages fall outside the calculation entirely [1]. That means a document can carry a visible AI percentage while most of its actual words were never assessed.
Turnitin frames the result as a signal for instructors, not proof of misconduct [2]. The company explicitly advises educators to interpret the indicator alongside context — drafts, conversations with the student, and the assignment itself — rather than treating a number as a verdict [2]. For students, the practical takeaway is that the AI writing score is a screening indicator with real false-positive risk, not a lie detector.
How Does Turnitin Calculate And Display The AI Detection Percentage, And Why Does It Sometimes Show An Asterisk Instead Of A Number?
The displayed percentage is calculated only over qualifying text, so a 25% AI score does not mean a quarter of the entire document was written by AI [3]. It means that within the portion of prose the detector evaluated, roughly a quarter of the segments were classified as AI-generated. Because non-qualifying content is excluded, two papers of the same length can show the same percentage while representing very different amounts of flagged text [3].
When the AI score falls below Turnitin's confidence threshold, the report displays an asterisk — shown as *% — instead of an exact figure [3]. That asterisk is a low-confidence signal, not a clean bill of health, and it should be read as inconclusive rather than as confirmation that the writing is human [3]. Students sometimes assume an asterisk means "zero AI," which is a misreading of what the report is actually saying.
Turnitin also notes that the indicator is not designed to be applied to every kind of writing task [1]. Because the model was trained on long-form academic prose, its output is most meaningful on essays, reports, and dissertations of a few hundred words or more [1]. Applying it to lab notes, code comments, or a two-paragraph reflection will not produce a result you can reasonably act on.
Can Students Check Their Own Turnitin AI Detection Score Before Submitting A Paper?
In most institutional setups, students cannot run their own paper through the Turnitin account their university pays for — access to the Similarity and AI reports is controlled by the instructor and by the assignment's settings [4]. Some instructors deliberately enable a draft or pre-check assignment so students can see their report before the final deadline, but that is a course-level choice, not a default [4]. If your module does not offer one, there is no official student-facing button that produces your own AI score.
That access gap is exactly why pre-submission checking has become common practice [4]. Students who want to know where they stand before the real submission often use an independent checking service that returns a Turnitin AI report and a similarity report on their draft [4]. The value is not in gaming the system — it is in seeing the same kind of signal your instructor will see while you still have time to revise, add citations, or rewrite passages that read as machine-generated.
If you do get a preview, read it the way Turnitin tells instructors to read it: as one input among several [2]. A high AI percentage on a fully human draft usually points to writing that is unusually uniform in rhythm and vocabulary, which is fixable. A low percentage or an asterisk is not a guarantee, and a flagged section is not an accusation [3]. The point of checking early is to make an informed decision about revision, not to chase a number.
Reading about thresholds and asterisks only gets you so far — what actually settles the question is seeing your own Turnitin AI report on your own draft before the deadline does. That is what turnitin0 was built for: students upload a draft and get back the same style of AI detection and similarity reports their instructors see, so the revision decision is made with real evidence rather than guesswork.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is Turnitin's AI detector the same as a plagiarism checker?
No. The similarity checker compares your text against a database of sources, while the AI detector analyses writing patterns to estimate how much prose looks machine-generated [1]. Both appear in the same Similarity Report, but they answer different questions.
Does an asterisk (*%) mean my paper has zero AI?
No. An asterisk appears when the AI score falls below Turnitin's confidence threshold, which makes the result inconclusive rather than a confirmation of human writing [3]. Treat it as "not enough signal to report a number," not as a pass.
What percentage of AI writing gets flagged?
Turnitin's AI indicator only surfaces in the report once at least 20% of the qualifying text is flagged, and the percentage is calculated over qualifying prose alone [1]. Short responses, bullet points, and non-English text are excluded from that calculation.
Can I see my Turnitin AI score before I submit?
Usually only if your instructor enables a draft or pre-check assignment, because report access is controlled at course level [4]. Otherwise students rely on independent pre-submission checking to see a comparable AI and similarity report [4].
Why does my fully human essay still show an AI score?
The detector works on statistical patterns like perplexity and burstiness, and very uniform, formulaic prose can resemble AI output [2]. Turnitin itself describes the score as a signal to be interpreted with context, not proof of misconduct [2].