Direct answer
Direct Answer — An AI detector is a tool that identifies whether a piece of text was generated by artificial intelligence. Turnitin's AI writing detection, built into the same platform as its similarity checker, analyzes sentence-level patterns such as perplexity and burstiness to distinguish human-written content from text produced by large language models (LLMs) like ChatGPT, Claude, and Gemini [1]. The detector returns an overall percentage score along with highlighted sections that indicate which parts of a submission were likely AI-generated, helping instructors maintain academic integrity in the classroom [1].
How Does the Turnitin AI Detector Analyze Writing?
Turnitin's AI detector does not simply flag entire documents as "AI" or "not AI." Instead, it segments each submission into smaller writing units — typically a few sentences at a time — and evaluates every segment independently for signs of AI generation [2]. This sentence-level approach allows the detector to identify specific passages that exhibit patterns common to LLM output, such as uniform sentence complexity and low variation in word choice [2].
The analysis relies on two primary linguistic signals: perplexity and burstiness. Perplexity measures how predictable a sequence of words is — AI-generated text tends to be more predictable (lower perplexity) than natural human writing. Burstiness captures the natural variation in sentence length and structure that human writers produce, which LLMs often lack [1]. By combining these signals, Turnitin can highlight the exact sentences or paragraphs it suspects were AI-generated, and instructors can view those highlights alongside the overall score in the AI writing report [2].
Turnitin's detection is designed for English and Spanish submissions and targets a false positive rate of less than 1% for entire-document analysis [1]. This means that when the detector flags a passage, there is a high degree of confidence. However, Turnitin advises that the AI indicator should be used as a conversation starter rather than a punitive measure, encouraging instructors to discuss writing processes with students when flags appear [2].
What Do Turnitin AI Detection Scores Mean and How Are They Displayed?
The Turnitin AI detection score — called the AI indicator — is displayed as a percentage from 0% to 100%, reflecting the portion of a submission that the detector believes was generated by AI [3]. A score of 0% means no AI-generated text was detected, while higher percentages indicate a greater proportion of flagged content. Critically, scores below 20% are displayed as an asterisk (%) rather than as a single-digit or low double-digit number [3]. This design choice prevents misinterpretation of low-confidence flags: if only 3% or 12% of a document is flagged, the system shows % rather than an exact low number that could be misread as a definitive finding.
The only explicit low numeric outcome a student typically sees is 0% [3]. Everything else below 20% falls into the asterisk bucket. This display convention aligns with Turnitin's guidance that the AI indicator is not a definitive "cheating detector" but an observational tool meant to prompt dialogue between instructors and students about proper source use and original writing [3].
The AI writing report is presented alongside the similarity report in Turnitin's Feedback Studio. Instructors see both the overall percentage and color-coded highlights over the flagged portions of the text [2]. The report does not identify which specific AI tool generated the text; it only indicates that the writing patterns are consistent with AI generation from LLMs in general [1].
Can Students Preview Their Turnitin AI Detection Score Before Submitting?
In the standard institutional workflow, students cannot preview their Turnitin AI detection score before submitting an assignment. The AI writing report is generated after a paper is submitted to an instructor's Turnitin-enabled assignment dropbox, and the results are visible only to the instructor [4]. This means students receive no advance warning about how their writing might be flagged before the formal submission is graded.
However, pre-submission awareness of AI detection is important for students who want to ensure their work reflects their own original thinking. Some third-party services offer the ability to upload a draft and receive a Turnitin AI detection report before the final submission is made to an instructor [4]. These preview reports use the same underlying detection indicators and show the same percentage score, flagged highlights, and overall AI indicator that an instructor would see after submission.
Previewing the AI score before submission gives students the opportunity to review flagged passages, understand why certain sections were identified as potentially AI-generated, and make revisions if needed [4]. This proactive approach aligns with academic integrity best practices: rather than discovering a high AI score after submission, students can address concerns in advance. Services like Turnitin0 provide this preview capability by generating real Turnitin AI and similarity reports from draft uploads, helping students check their work before the official submission deadline.
At Turnitin0, we understand that navigating AI detection can be stressful — especially when you are unsure how your draft will score before submitting it to your instructor. That is why we offer a real Turnitin AI and similarity checking service that lets you preview your reports in advance, giving you full visibility and control before your final submission.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
1. Is the Turnitin AI detector accurate?
Turnitin targets a false positive rate of less than 1% for entire documents and analyzes submissions at the sentence level using perplexity and burstiness signals [1]. While no detector is perfect, Turnitin's approach is widely used in higher education as a reliable indicator that prompts instructor-student discussion rather than serving as a definitive judgment [3].
2. Why does Turnitin show *% instead of a low number?
Scores below 20% are displayed as an asterisk (*%) to prevent misinterpretation of low-confidence flags [3]. Showing an exact single-digit percentage could be misleading, so the asterisk indicates that while some text was flagged, the amount is below the reporting threshold. The only explicit low numeric score a student typically sees is 0% [3].
3. Can I check my AI score before submitting to my instructor?
In the standard institutional workflow, no — the AI writing report is generated only after submission and is visible only to the instructor [4]. However, third-party services like Turnitin0 allow you to upload your draft and receive a real Turnitin AI report before the official submission, giving you a preview of how your work would be scored [4].
4. What types of AI writing does Turnitin detect?
Turnitin's detector identifies text generated by large language models (LLMs) including ChatGPT, Claude, and Gemini [1]. It does not identify which specific AI tool produced the text — only that the writing patterns are consistent with AI generation from LLMs in general [1].
5. Does a 0% AI score guarantee my paper is completely human-written?
A 0% score means no AI-generated text was detected by Turnitin's analysis [3]. However, no detection tool offers absolute guarantees. Turnitin recommends using the AI indicator as part of a broader assessment of academic integrity, not as the sole determining factor [3].