Sydney academic's AI-written opinion piece removed by SMH

A top Sydney academic used AI to write an opinion piece that urged students to “do the work” and not cut corners by using such technology, with the Sydney Morning Herald removing the “unacceptable” piece from its website.

Western Sydney University’s pro-vice chancellor for quality and integrity, Prof Cath Ellis, had an opinion piece published in the Sydney Morning Herald last month, in response to an article from academic Kylie Moore-Gilbert.

Moore-Gilbert had written that she had advised her stepdaughter to think twice before enrolling in university as students could easily outsource their learning to AI, saying students were “being graded on who can write the best AI prompts.”

In response, Ellis wrote in her piece that the “AI problem is real”, but students should still go to university and study properly.

“Don’t cut corners. Don’t outsource your thinking, however tempting that may be. If the system is as fragile as some claim, then genuine effort will not be hidden. It will stand out,” she wrote.

However, the column, when submitted to AI-detector service Pangram, came up as 100% AI-generated.

In response to questions from Guardian Australia, the university said Ellis had used AI in writing the column.

“To write her opinion article, Prof. Ellis uploaded 40,000 words of her own original materials into a Copilot Large Language Model (LLM). The model summarised her extensive base of knowledge, providing prompts,” the spokesperson said.

“This was the basis of the early drafts, reflecting Prof. 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 spokesperson said the use of an LLM to draw on her own expertise and experience “demonstrates a sophisticated and appropriate use” of generative AI.

“Programs like Pangram can detect AI use, but they cannot determine whether that use was appropriate or inappropriate,” the spokesperson said. “The University believes the AI use in this case was appropriate.”

Ellis’s opinion piece did not declare the use of generative AI to write the article as of Wednesday morning.

The editorial policy for Nine, parent company of the Sydney Morning Herald, allows writers to use AI for initial research and prompt ideas but states: “AI will not be used to write stories for publication.”

The policy states where AI-generated material is published, it will be clearly labelled, but says “the use of assistive AI does not require declaration”.

Nine did not respond to questions from Guardian Australia. But in a subsequent article published Wednesday, the Sydney Morning Herald’s editor, Jordan Baker, said the article didn’t meet its editorial guidelines and has been removed.

“The Herald was not informed of the use of AI in the compilation of the article by either the author or Western Sydney University,” Baker said.

“Clearly this is unacceptable and we are investigating further.”

As generative AI tools become more embedded in workplaces across the globe, incidents of media outlets publishing output from AI without disclosure are becoming more common. Crikey removed a series of articles from one writer in March after it was revealed the author had used AI to assist with the proofing of the copy.

Also in March, a freelance journalist admitted to using AI for a book review that echoed elements of one published in the Guardian, leading the New York Times to cut ties with him.