Turnitin AI writing detection
Released: March 2023
Turnitin is a AI writing detection tool that has been accused of being highly inaccurate and easily tricked to avoid detection, according to research studies and a media investigation.
An April 2023 Washington Post investigation found that Turnitin's AI writing detection tool is inaccurate over half the time, despite the company claim that its system is 98 percent confident of its assessments. The Post tested the system on five high school students, and discovered that it wrongly assessed most of a range of human-written, AI-generated, and mixed-source essays, even if it was good at identifying AI-generated text when the whole input had been created by ChatGPT.
A July 2023 study by a group of researchers led by Debora Weber-Wulff, professor of media and computing at the University of Applied Sciences, HTW Berlin, assessed 14 detection tools, including Turnitin, to identify words generated by ChatGPT. The researchers found that the tools were unable to detect text that had been slightly rearranged by humans and altered using paraphrasing 42% of the time. Weber-Wulff said the tools 'don't work' and 'don't do what they say they do.'
It is unclear how Turnitin verifies the accuracy of its software or how it predicts text generated by different language models. Educators are unable to opt-out of using the software.
The company points out that the results of its tool should not be used to accuse students of cheating.
Sector: Professional/business services
Purpose: Detect AI writing
Technology: NLP/text analysis; Neural network; Deep learning; Machine learning
Transparency: Governance; Black box; Marketing