Students’ Use of Artificial Intelligence Leaves University in Puzzling Plagiarism Pickle

Words by Ethan Manera (he/him)

With the rise of AI, students are turning to ChatGPT to write assignments, despite Te Herenga Waka—Victoria University of Wellington’s concerns surrounding academic integrity. 

At the start of 2023, VUW moved quickly, releasing a statement warning students that using ChatGPT would “likely” be a violation of the university’s academic integrity policy.

But their ability to detect and regulate the use of AI is unclear. 

Jo*, a third-year student, says she’s used ChatGPT to help with essay plans and processing notes for multiple assignments. “I haven't handed in anything that is directly written by the AI, but it certainly generated valuable ideas and points. If you know how to use [it], why shouldn't you?” she said.

Jo* received an A grade for an essay that used AI supported structure and planning.

Dr Simon McCallum, a senior lecturer at VUW with a PhD in AI, has been alerting the university to the possible consequences of students using AI. But he says they need to act fast to ensure VUW has appropriate systems in place to combat ChatGPT.

McCallum says staff need to be clear with students about whether or not they allow the use of tools like ChatGPT. “My main concern is that many lecturers are still in denial about the significance of this.”

Turnitin claims to have developed new AI detection software. However, its effectiveness is unproven, with tests showing that AI written text can evade detection. GPTZero is another AI writing detection tool which is being used in universities, but McCallum says this is problematic.

“All the detection tools are basically a hunch [...] there's a reasonably high chance that one of these tools will falsely detect human writing as AI,” he said. 

McCallum expressed concern that innocent students could face punishment due to false AI detection. 

A university spokesperson told Salient that “ChatGPT should only be used if specifically directed by lecturers and referenced in assignments”, and said that VUW “is not encouraging staff to use technologies like GPTZero to combat the use of AI; detection technology is too unreliable as yet.”

“Instead, we are encouraging staff to begin adapting their assessments to embrace AI technology and enhance the learning experience wherever possible. This means creating assessments that are less vulnerable to simple AI-based circumvention,” they said.

*names have been changed