Skip to main content

Turnitin AI Writing Detection

  • Updated

*The University did not renew the AI detection due to updated terms in the TurnItIn AI contract. 

We continue to have TurnItIn

 

1. Overview

Turnitin’s AI Writing Detection tool identifies an overall percentage of likely AI-generated text in a student’s submission, with a report highlighting text segments our model predicts were AI-generated.

2. How to Access the AI Report

The AI detection report is integrated directly into the Turnitin Feedback Studio.

  1. Open the student’s submission in Canvas.
  2. Click the blue Similarity Score to open the Feedback Studio.
  3. In the side navigation bar, look for the blue AI indicator (an icon with a percentage).
  4. Click the percentage to open the full AI Writing Report.

 

Note: If you see an asterisk (*%) instead of a number, it means the detected AI content is below the 20% threshold. To avoid false positives, Turnitin does not highlight text for scores below 20%.
 

3. Technical Requirements & Limitations

For the AI detector to run, the submission must meet the following criteria:

  • Word Count: Minimum of 300 words of prose text.
  • File Format: Must be .docx, .pdf, .txt, or .rtf.
  • Language: Currently optimized for English (additional language support is currently in rolling release for 2026).
  • Content Type: The detector only scans long-form prose. It does not reliably detect AI in:
  • Bullet-point lists or tables
  • Computer code
  • Short-form responses (<300 words)
  • Poetry or creative writing with unconventional structures

4. Interpreting the Results

The AI score is a data point. It is a data point to inform your professional judgment.

  • Low Score (0-19%): No text is highlighted and generally considered "human-written" or containing minor AI assistance.
  • Mid-to-High Score (20-70%): Specific segments will be highlighted in blue. Review these areas: Do they sound different from the student's usual voice? Are they overly perfect or generic?
  • AI-Paraphrased Sub-score: New for 2026, this flags text that was likely generated by AI and then run through a "humanizer" or paraphrasing tool.

5. Frequently Asked Questions (FAQ)

Q: Can a student see their AI score? 

A: No. By default, the AI Writing Report is only visible to instructors and administrators to ensure scores are interpreted with context before being shared.

Q: Does it detect "Grammarly"? 

A: Basic grammar tools usually don't trigger high scores. However, advanced "AI-rewrite" features in Grammarly or Microsoft Editor may contribute to a higher percentage.

Q: What do I do about a False Positive? 

A: If a student's work is flagged but you believe it is original, consider their drafting history or hold a brief "viva" (oral exam) to discuss the content. You can report false positives to the Helpdesk to help improve the model.

Resources: 

Need further help? Contact the Helpdesk at helpdesk@arcadia.edu.

Comments

0 comments

Article is closed for comments.