AI Content Prediction helps you assess the likelihood that parts of a submission were generated using AI. This feature is powered by Pangram, a leading solution for AI content prediction

Authorship result

The Authorship section provides an overall AI authorship result for the submission:

• AI Authorship classification: A summary indicating the predicted level of AI-generated content.
• AI Authorship score: Displayed as a percentage or as a risk level (depending on your institution’s settings).

The result is intended to support your review and investigation. It should be considered alongside other indicators in the submission and the wider academic context.

Detailed findings

Clicking “Details” in the Authorship section opens a detailed breakdown of the AI Content Prediction findings. This view provides additional context to help you better understand how the overall authorship result was determined.

The following indicators are available:

• AI Confidence Score: Shows the confidence level of the prediction based on the analysis performed. This helps indicate how certain the system is about the findings.
• AI Segments: Displays the number of segments within the document that are flagged as potentially AI-generated.
• AI-Assisted Score: Indicates how much of the document may have been created with AI assistance, rather than fully generated by AI.
• Human Score: Shows the portion of the document that is predicted to be written by a human.
• Human Segments: Shows the number of segments identified as likely written by a human.

These metrics are designed to support informed review and should be interpreted alongside the full Originality Report and institutional guidelines.

Reviewing flagged content

Content that is identified as potentially AI-generated is visually highlighted within the document view. Flagged sections appear with a purple indicator along the left side of the text, making it easy to locate and review segments that require closer attention.

These visual markers help users quickly navigate between findings without needing to manually scan the entire submission. Selecting a highlighted section will display the related AI Content Prediction details, allowing you to review the context of the flagged content as part of your overall evaluation.

Segment Details Panel

When a highlighted AI segment is selected, a details panel appears on the right side of the screen. This panel provides additional information about the specific finding, helping you review the flagged content in context.

The panel includes:

• Prediction classification: An explanation indicating how confident the tool is that the selected text may be AI-generated.
• Confidence level: Displays the statistical confidence for this individual segment.
• Word count: Displays the number of words included in the flagged section.

Below these details, the panel displays the exact text identified, allowing you to review the flagged content in its original context.

If you want to know more about the false positive rate of Pangram, please refer to their official article.