University cheats on notice after launch of ChatGPT detection software

Australia’s universities will gain access to new technology designed to crack down on cheats using ChatGPT but some top institutions are shunning the software as teachers look to redesign tests to combat the rise of artificial intelligence.

Most universities nationwide will on Wednesday have the option of using popular anti-plagiarism software service Turnitin to detect whether a student has used a chatbot to help write an essay or complete an assessment.

But some of the country’s biggest institutions including the University of Sydney, Monash University and Deakin University have said they will not use the software – at least initially – and are instead ramping up other detection methods to catch students using ChatGPT to write papers.

Academic integrity expert at the University of NSW, Cath Ellis, said there is a “real fear” the detection tool could lead universities to falsely accusing students of using ChatGPT to do their work.

“We could also end up with a massive tidal wave of referrals coming through from academics that we can’t handle, many that could be false accusations,” she said.

“Turnitin are releasing this tool, but the perception among the higher education sector is that the type of testing that has been done hasn’t been effectively communicated.”

James Thorley, regional vice president of Turnitin, claims the company’s new tool can identify if a student has used an AI chatbot in their work with 98 per cent confidence. He said about 780 high schools in Australia used Turnitin and will have access to the new tool.

“Banning ChatGPT isn’t feasible long-term,” said Thorley. “This detector isn’t just about maintaining academic integrity but is also about understanding how AI writing tools are changing the future of assessment,” he said.

However, Benjamin Miller, an English lecturer at Sydney University, said he is opting to redesign assessments for his first year students to deal with chatbots.

“I immediately started thinking about the ethics of using ChatGPT in academic writing, and I was surprised how well it could write and how widespread its use is,” he said.

“I now give students a sample of writing that ChatGPT has created, and they will be tested on how well it analyses and demonstrates critical thinking. They are also tested on how they exceed the capabilities of a chatbot.”

“ChatGPT isn’t great at analysis and evaluation, and often doesn’t connect ideas across paragraphs, so you can often pick up if it’s been used that way.”

A spokesperson for Sydney University said the institution would not be using Turnitin’s new AI detection feature immediately.

“We aim to avoid making major changes to our systems mid-semester, and without adequate testing or visibility, or time to prepare staff,” they said.

‘We could also end up with a massive tidal wave of referrals coming through from academics that we can’t handle, many that could be false accusations.’

The university said it would be reviewing the feature’s capability to see if it would help markers when assessing if a student’s work was original.

The university is also ramping up face-to-face supervision during oral exams and more pen-and-paper assessments.

Monash University has also decided against using Turnitin’s tool “given the technology is in its infancy”.

“The university is focusing on educating students around responsible and appropriate use. Our goal is to help students understand how generative AI can be used ethically and responsibly in their learning and assessment, and how it may be used effectively in their current or future careers,” a university spokesperson said.

Deakin University said claims Turnitin’s tool has a 98 per cent accuracy rate in the detection of AI-generated text have not been verified by the institution, and flagged concerns it had been trained using out-of-date AI text generator models.

“Until the university can test its efficacy, Deakin has chosen not to apply the tool in the marking of student assessments,” said Associate Professor Trish McCluskey, Director, Digital Learning.

“This is to protect student data and is in line with the approach adopted by a growing list of global education providers, and we expect many Australian universities will follow our lead.”

However, UNSW said the feature would be available for academics to consider cautiously, but that staff would not be relying on it in any way.

Deputy vice-chancellor academic at UNSW Merlin Crossley said the solution to dealing with ChatGPT was to work more closely with students and improve assessments in the senior years when there are smaller classes, and lecturers are better acquainted with their students.

Crossley believes the rise of ChatGPT could lead to a decline in contract cheating – where students pay external companies to ghostwrite essays or complete work – which universities have faced at record levels in the past two years.

“Students can’t cheat their way through a degree. They can take shortcut in the early years, but ultimately a lot of the problems only occur in the early years of students courses.”

 

Story first published on The Sydney Morning Herald

Link to original story

Author: 
Lucy Carroll and Adam Carey, The Sydney Morning Herald