According to the recently-published Australian Community Attitudes to Privacy Study 2026, “Australians expect organisations to be open about how AI is used and how decisions are made, including being informed when AI is being used (79%, up from 71% in 2023).” The Study further found that nearly all expressed their desire that some conditions should be put in place before AI is used.
It comes as Pro-Vice Chancellor of Western Sydney University, Cath Ellis admitted using AI, ironically enough in a piece highlighting the dangers of excessive reliance on large language models.
While use of out-of-place language may have led to her being outed, we know for certain that use of AI is widespread among the students, staff and faculty of Australian universities.
What’s at risk is not just the ability to write, but the ability to think. While AI may act as an accelerant for certain types of fundamental research, as a writing tool, it will tend to homogenise and sultify academic communication.
Worse, a position may well be reached where much of the content on the internet becomes unusable as it becomes undetectably interlaced with first, second and third order parroted texts, sullied by the epistemological leaven of generative AI.
At the very least, there should be a simple, intuitive signpost to both bots and human readers indicating the provenance of published academic material.
That’s where the Authorship Classification Scheme comes in.
The ACS sets a high bar for its HumanAuth and HumanAuth+ standards. For the latter, specifically, it’s not enough that no AI be used in the creation of the text; secondary sources consulted ought also to be AI-free.
Eventually even AIs will crave these sources in an effort to avoid the endless regression of AI taught by AI texts.