Monash University in Melbourne senior lecturer in emerging technologies Prof Jathan Sadowski is in Dunedin to speak at a University of Otago symposium tomorrow.
He told the Otago Daily Times AI technologies such as ChatGPT and OpenAI were not the "neutral" technologies they were sold as.
"Central to the pitch of AI is that it is this almost superhuman force that we can’t resist.
"Rather than using ChatGPT to make our writing more interesting or maybe a little bit easier, we instead outsource that writing of news stories or essays at university to the machine.
"People become editors for the machine rather than the machine becoming an editor for people."
Prof Sadowski said he worried that lack of understanding of AI’s functions would lead to "reorganising where the human becomes subordinate to the machine".
"AI is this pushing everything towards what’s called the convergence to the mean, because the way that the AI works is it produces an average of all of the data that it has ingested and analysed.
"If you only look at the individual outcomes of AI, such as the short story, news article or movie script it writes, it might sound really interesting and creative. But if you look at all of the outputs of the AI, they start looking exactly the same."
Earlier this year, Otago University’s student academic misconduct report revealed that of the 79 investigations of academic misconduct in 2023, 24 were for unauthorised use of generative artificial intelligence (GenAI).
Prof Sadowski said he encouraged his students to be "sceptical" in its usage.
"The most important and the most responsible way to use AI is to use it in a really sceptical and critical way, to understand that the machine is not magical, that the machine does not understand anything.
"AI does not think. It does not know what it’s doing."
Instead, one could use it as a way to edit copy for clarity.
"When I have students who start relating to AI in that way, as something stylistic and superficial, I see that it actually improves their writing, and improves their thinking, because they can understand what it’s trying to achieve."
It would take the next decade to really come to grips with AI.
He also predicted the big technology companies such as Google would not achieve their dream of "big universal generalised forms of AI".
"I’m very hopeful that we will, as a society, position ourselves in a more responsible and effective way towards technologies that for a very long time we’ve given the benefit of the doubt.
"What we need to have instead are a number of much smaller, much more specialised forms of AI that actually work to improve people’s lives, and safeguard the values that we have as a society, rather than these big massive mega systems that try to override people."
The lecture on AI is in Castle Lecture Theatre 2 from 5pm tomorrow, and will also feature a panel of Otago University academics afterwards.
