Turnitin AI Detection Algorithm
I have observed that the AI generated text identification in Turnitin is not accurately reflecting the source of the content. In reviewing multiple assignments, research papers of my students, and even my own work, which are entirely original and not created using any AI tools or large language models, the system still indicates a high AI generation percentage. Furthermore, Turnitin appears to flag text that is directly taken from books or academic sources published prior to 2018 as AI generated, which is factually incorrect. This misidentification creates an unfair situation for students and researchers, as it could lead to false accusations of using AI generated content despite the work being fully original. I request that Turnitin reviews and improves the accuracy of its AI detection system to prevent such errors and ensure fair evaluation.
1 reply
-
Hello there,
We invite you to join our #AskTurnitin: Month-Long Q&A with the Turnitin Team on Navigating AI in Teaching and Learning event! Our team members, Patti and Gailene, would love for you to join our ongoing discussion about AI in the classroom.
And truly, you’re asking this in exactly the right place. These are the kinds of conversations so many educators are navigating right now, and we really appreciate you bringing them forward.
We’d love it if you could share your question directly in the thread so we can continue the conversation there!
- Kat, Turnitin Team