Direct answer
An AI tester picture is the visual output of an AI writing detector: a report page showing an overall AI score, a percentage of qualifying text, and colour-highlighted passages inside your document. In Turnitin's report, the score sits in a sidebar while flagged sentences are shaded in the text itself, so you can see exactly which paragraphs triggered the detector [1]. Scores below the 20% confidence threshold are displayed as *% rather than a precise number, which signals a low-confidence result rather than a clean pass [1].
What Does a Turnitin AI Detection Report Actually Look Like, And What Do The Highlighted Parts Mean?
The report opens with a summary panel. It shows a single AI writing percentage for the submission, the number of qualifying words the detector analysed, and a note that the figure reflects the proportion of qualifying text flagged as AI-generated [2]. That distinction matters: a 30% score does not mean 30% of the whole essay — it means 30% of the qualifying prose the detector could evaluate. Reference lists, quotes, and very short fragments are excluded from the qualifying word count, so the denominator is smaller than your total word count [2].
The picture then continues into the document body. Flagged sentences are shaded, and clicking a highlighted segment in the sidebar jumps straight to the matching passage in the text [2]. This is the part students most often misread: the highlight marks a segment the model scored as machine-like, not a segment proven to be machine-written. Turnitin's own report separates AI-generated text from AI-paraphrased text, because a rewritten passage leaves a different statistical signature than a freshly generated one [2].
Colour intensity is also informative. Stronger shading generally corresponds to segments the detector scored with higher confidence, while faint shading clusters near the threshold where the signal is weaker [1]. When the overall figure falls under the confidence floor, Turnitin replaces the number with *% instead of publishing a precise percentage, precisely because a low-confidence signal should not be read as an accusation [1]. Reading the picture therefore means reading three things together: the headline score, the share of qualifying text, and where the shading actually lands.
One practical note on what the picture is not. The AI report is a separate document from the similarity (plagiarism) report, with its own score, its own highlighting, and its own logic [1]. A submission can show 0% similarity and still carry a visible AI score, or the reverse. Students who expect one combined picture are usually looking at the wrong tab.
Why Does An AI Detector Flag Writing That A Student Wrote Or Edited Themselves?
Detectors do not look for authorship; they look for statistical predictability. A language model produces prose that is unusually smooth and low-variance in word choice, and the detector is trained to recognise that pattern [3]. Human academic writing can drift toward the same pattern for entirely innocent reasons — heavy editing toward a formal register, a template-driven structure, or paragraphs rewritten to sound more academic all reduce the variance the detector expects to see [3].
This is why the picture sometimes shows shading on passages a student genuinely drafted. Turnitin and independent researchers both describe false positives as an inherent limitation of AI detection rather than a bug that can be eliminated [3]. The report is designed to be read as a signal that prompts a conversation, and institutional policy — not the number alone — governs what happens next [3]. A shaded sentence is evidence of a pattern, not evidence of misconduct.
Editing history is the usual culprit. Students who draft with an AI assistant and then rewrite sentence by sentence often produce text that is simultaneously more polished and more uniform than either the raw draft or the original human version, which can push segments over the confidence line [3]. The same applies to non-native English writers who use grammar tools aggressively, and to anyone who has tightened prose until every clause has the same rhythm.
The useful takeaway is that the picture is diagnostic, not final. If shading lands on passages you wrote yourself, the fix is usually stylistic variation — varied sentence length, specific detail, a genuine argumentative voice — rather than a rewrite from scratch [3]. If shading lands on text you did generate with a model, the fix is revision of that text, which is exactly what the report is meant to prompt.
How Can I See The AI Detection Picture For My Own Draft Before I Submit It?
The most reliable route is a pre-submission check that returns the same report view your instructor sees. Some institutions enable draft checking inside the institutional system, and some configurations exclude the draft from the student paper repository so it does not create a self-match later [4]. Where that option exists, it is the cleanest way to see your own picture with no third-party handling of the file.
Where institutional draft checking is not available, students commonly use a non-repository pre-check: the file is analysed, the AI and similarity reports are returned, and the submission is not added to Turnitin's student paper database [4]. The value of this loop is timing. Seeing the picture before the graded deadline lets you revise the specific shaded passages and re-check, instead of discovering the result after the window has closed [4].
Re-checking after revision is the part most students skip. A single check tells you where the shading landed; a second check tells you whether your edits actually moved the score or merely shifted the highlight elsewhere [4]. Two checks, one before and one after revision, is the minimum useful workflow for a draft you are worried about.
Keep the file requirements in mind so the picture is comparable to what your instructor will see. Checks generally expect English text, a word count above roughly 300 and below 30,000, and a file under 20 MB in .docx, .pdf, or .txt [4]. Formatting differences between a draft and a final submission can change how much text qualifies for analysis, which in turn changes the denominator behind the score [2].
If you would rather see the actual picture than imagine it, turnitin0 returns the real Turnitin AI detection report and the similarity report together, so you can look at the score, the shading, and the flagged passages on your own draft before your instructor does.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does a Turnitin AI score of *% mean my work is clean?
No. Turnitin displays *% instead of a precise percentage when the AI signal falls below its 20% confidence threshold, which means the result is low-confidence rather than absent [1]. Treat it as a weak signal and read the shading on the document alongside it.
What is the difference between the AI score and the similarity score in the picture?
They come from two separate reports with separate scores and separate highlighting [1]. The AI score reflects machine-like text patterns; the similarity score reflects matching text sources. A low number in one says nothing about the other.
Can the highlighted passages be wrong?
Yes. Detectors measure statistical predictability, not authorship, so human writing that is heavily edited or unusually uniform can be flagged [3]. The report is intended to prompt review, and institutional policy determines how it is used [3].
Will checking my draft first get me into trouble?
A non-repository pre-check does not add your file to Turnitin's student paper database, so it will not create a self-match against your later submission [4]. Confirm your institution's own policy on draft checking before you rely on it.
How long does it take to get the report picture?
With turnitin0, most checks are delivered in under 15 minutes, and delivery is guaranteed within 30 minutes even during queue spikes. Each order includes the AI detection report and the similarity report as two downloadable PDFs.