Turnitin switch on AI detection tool amid ChatGPT concerns

EdTech giant Turnitin turns on artificial intelligence detection capabilities today for a trial release as USyd and major universities pour cold water on the new technology for the time being.

American plagiarism detection giant Turnitin is rolling out artificial intelligence detection capabilities in an attempt to combat academic integrity breaches through the use of technologies like ChatGPT.

The move is billed as a “trial release” on a voluntary basis where universities have the option of opting out of AI detection services on a temporary basis. It is understood that the University of Sydney are among a number of universities across the country to have opted out of Turnitin’s enhanced capabilities for now. 

Currently used by universities around the world to check students’ work for originality, Turnitin claims that the enhanced capabilities is capable of distinguishing between AI-generated from human writing. 

It claims that the technology has a low false positive rate and that the technology is “constantly evolving” to cope with increases in AI-generative tools’ complexity.  

In a statement, Turnitin claims that the company’s AI detection tools has “98% confidence and a less than one percent false-positive rate”. 

“Today, we are pleased to announce the launch of our writing detection capabilities in Turnitin Feedback Studio (TFS), TFS With Originality, Turnitin Similarity, Simcheck, Originality Check, and Originality Check+.”

We have been very careful to adjust our detection capabilities to minimize false positives and create a safe environment to evaluate student writing for the presence of AI-generated text,” said Turnitin chief executive Chris Caren. 

Responding to questions from Honi regarding Turnitin’s enhanced AI detection service, a University of Sydney spokesperson said that the University is “currently investigating the new AI detection feature in Turnitin”.

“We’re taking a holistic approach to AI in our teaching and learning – aware of both the opportunities and the threats it represents. AI can help students learn and will be used in jobs of the future,” the spokesperson said. “We need to teach our students how to use it [artificial intelligence] effectively and legitimately.”

Meanwhile, USyd Students’ Representative Council President Lia Perkins said that Turnitin’s enhanced capabilities risks students undergoing time-consuming and onerous investigative processes for false positive reports. 

“The SRC has a serious concern about the possibility of false positives in Turnitin’s testing and the detrimental effect this will have on delaying the already broken system and causing undue stress,” said Perkins.

“USyd should focus on teaching and support that prevents breaches in academic integrity, as well as teaching students how to use new technology as a tool, not as a replacement for learning academic skills.”

Academics and experts have expressed concerns regarding the accuracy of Turnitin’s AI detection tools and workload burdens associated with increased volumes of academic integrity reports. 

Major universities including Cambridge have opted out of the service. According to Turnitin, two per cent of its customers have requested the service not to display the tool’s AI writing score on students’ work. 

British Columbia Institute of Technology Librarian and EdTech expert Dr Ian Linkletter warns Turnitin that the company “should delay the launch” of the AI text detection tools “until it is proven safe and accurate”.

“1 in 16 is an outrageous false positive rate,” Linkletter said. 

“The convenient thing about not letting anyone verify your claims is that you can pretend you didn’t know better. Turnitin can no longer do so, and should delay the launch of their AI Text Detector until it is proven safe and accurate. 1 in 16 is an outrageous false positive rate.”

Linkletter’s former institution, Canada’s The University of British Columbia (UBC), has chosen to opt out from Turnitin’s enhanced AI detection capabilities for the foreseeable future, citing “lack of ability to review and validate the feature”, “results not available to students” and “potential bias in the detector” on non-native English speakers as reasons behind the university’s hesitation.

“We are taking a pause to have time to review this new feature from Turnitin. This pause also provides a chance for broader discussions at the institution around the capabilities, limits and risks of AI detector tools (including this one but also others), and their value for academic integrity purposes.”

Turnitin can encounter issues when it comes to essays and works produced by a mix of human and AI sources where Turnitin “misidentified the individual sentences or missed the human part entirely.” 

Turnitin is widely used by universities across Australia and other countries, with the company stating that more than 10,700 institutions are currently registered with the service.

Filed under: