Turnitin's Originality product has their new AI Detection tool embedded into it, so faculty and administrators can test it out. It is totally separate from the Similarity checker and works a different way, and the percentages will be different. Also, this new AI tool's percentage is only based on long-form prose writing sections of the papers, excluding tables, bullet points, and lists.
However, Berkeley's Academic Integrity Policy (2025) states that:
A faculty member who suspects an academic integrity violation shall contact the student in question to discuss the situation, giving the student an opportunity to explain and/or defend himself/herself and present evidence if he/she has any. In academic integrity matters concerning the suspected misuse of AI, a faculty member may not assert AI detection as direct evidence of a violation (until further notice). AI detection programs are currently not reliable enough to bring disciplinary charges against a student, without other corroborating evidence. In the absence of direct evidence, a faculty member should use the discussion to probe the student’s comprehension of the material and their comprehension of the wording, phrasing, development, and structure of their own submission. The faculty member should gauge whether the student engaged in a legitimate intellectual process to arrive at the work submitted and whether the student is able to demonstrate proficiency regarding the assignment learning objectives independently from the work submitted. (pg. 4, section D; emphasis added)
Faculty may also want to review the resources provided by Berkeley's Center for Curriculum, Instruction & Technology (CCIT) and ask them for help adapting assignments to avoid AI misuse.
More info from Turnitin:
- How an AI checker can support quality, original writing
https://www.turnitin.com/blog/how-an-ai-checker-can-support-quality-original-writing - FAQs about how it works, the requirements, how it was tested, how to interpret the results, and its accuracy: https://www.turnitin.com/products/features/ai-writing-detection/faq
They also may need to reconsider how they determine what is cheating:
Links & Files
Media
Understanding false positives within Turnitin’s AI writing detection capabilities
https://youtu.be/4e9zM2MZvRQ?si=pB8hAO51Nz1nCLJ_