‘Odd choices of words’: How an academic’s AI use was exposed by her peers

When a professor specialising in academic integrity wrote an opinion piece that defended universities against claims that artificial intelligence had downgraded the value of their qualifications, academic WhatsApp groups lit up.

Most academics are familiar with the language patterns generated by AI tools from marking student essays in the age of Claude and ChatGPT. “I knew as soon as I read the first paragraph,” one later remarked. The opinion piece was itself written with AI.

Western Sydney University acknowledged in response to enquiries that Professor Cath Ellis, a pro vice chancellor in quality and integrity, had used AI to write the article published by The Sydney Morning Herald and The Age on Sunday, drawing on her previous research in the field.

It argued that students should hold faith with the higher education system, and was submitted in response to an earlier piece by Macquarie University’s Kylie Moore-Gilbert that argued universities were committing “widespread industrial fraud” by accepting money from students and giving them degrees that they did not earn.

But Ellis denied that her piece was written “by” AI. “It was written with AI, and there’s a really big difference there,” she said.

She said she wrote down some thoughts on the train coming into work and then asked AI assistant Copilot to pull them into a coherent structure.

“I think of it very much as a sort of member of my team,” she said. “I really do feel that it’s allowed me to focus more of my time and energy on what really matters, which is the ideas, the thinking … rather than spending a lot of my time writing sentences from scratch.”

Academics pointed to several giveaways that Ellis’s piece was written by AI, including instances where three verbs or concepts are listed together, known as the rule-of-threes.

Moore-Gilbert said her suspicions were raised by the overuse of short, declarative sentences such as: “It is not this. It is that.” She also noticed a large amount of unnecessary jargon that did not make sense when it was drilled down.

“For instance, the repetition of ‘to assure learning’ or ‘artefact’ to me seemed odd choices of words in the context they were used,” Moore-Gilbert said.

The incident exposes the differing norms in the use of AI between institutions and industries.

WSU has embraced AI and encourages its students and academics to use it. A spokesperson for the university said Ellis’s use of the technology was “sophisticated and appropriate” and reflected WSU’s “institutional position of human-centred AI”.

“To write her opinion article, Professor Ellis uploaded 40,000 words of her own original materials into a Copilot Large Language Model (LLM),” the spokesperson said.

“The model summarised her extensive base of knowledge, providing prompts. This was the basis of the early drafts, reflecting Professor Ellis’s own thinking, ideas and opinions built up over more than a decade of dedicated work as a global leader in this field.”

The media office said it was aware that AI had been used to support editing, structure and language refinement based on Ellis’ own ideas. No other opinion pieces offered to this masthead had been produced in the same way.

This masthead has strict guidelines on the use of AI and allows its employees to use it as a tool where there is genuine benefit, but not to write stories for publication.

Ellis said these had never been conveyed to her. “If I’d known, I absolutely never would have sent it in,” she said.

This masthead’s executive editor, Luke McIlveen, said all new contributors would be asked to guarantee that AI had not been used to write or construct articles.

“Our position on the use of AI is clear – it cannot be used to write stories for publication. AI detection programs are useful, but not foolproof. If our editors believe an author’s work is not entirely their own, it won’t be published.”

The article has been removed from the Herald and Age websites.

Herald editor Jordan Baker said the masthead had not been informed of the use of AI in the compilation of the article by either the author or WSU, and it did not meet editorial standards.

The situation has also raised questions around authorship.

“We are still debating that,” said Dr Francesco Bailo, the co-director of the University of Sydney’s Centre for AI, Trust and Governance.

“When we write something, we put our name on it. There are two lines of thought – if you put your name on something, you should have written every single word, or how I write my stuff is my own business, you are not to see how the sausage has been made.”

Tools for transcription, grammar and translation are all AI-powered.

“Even the fact that you’re using a word processor, it can influence your writing. At the end of the day, you’re not assessing the process when you write something. You’re assessing the final product,” Bailo said.

Opinions on this vary – on TikTok, videos of people throwing out AI-generated books abound, while last week, students at the University of Arizona booed former Google boss Eric Schmidt as he attempted to talk to them about AI.

Studies have shown that reliance on AI can lead to cognitive atrophy, weaker critical thinking skills and a decline in creativity.

Curtin University’s Mollie Dollinger, who oversees assurance of learning, said that leaders at universities must find a way to “take people along the journey with us”.

“Using AI in high-profile pieces of writing, whether that’s a journal article or a newspaper article, or a press release for the university, is a choice, and that choice obviously signals how much you’re trying to push the dial,” she said.

Dollinger is among the dozens of signatories to the Castlereagh Statement, a call for a uniform and nationwide approach to AI and education.

The Morning Edition newsletter is our guide to the day’s most important and interesting stories, analysis and insights. Sign up here.