2 days ago
AI not a 'straightforward' fix for ailing productivity
Australia has been warned against the "seductive" pull of artificial intelligence as the federal government looks to the technology to help solve its productivity woes.
AI is expected to take centre stage during the second day of the government's economic reform roundtable, alongside regulation and competition.
Though he recognised its risks, Treasurer Jim Chalmers has previously said AI could be an economic "game changer" to boost Australia's ailing productivity and lift living standards.
But Monash University human-centred computing lecturer Jathan Sadowski warned the story was not so simple.
"AI changes the nature of work but it doesn't straightforwardly make work more efficient or more productive," he told AAP.
"It produces all kinds of new problems that people need to adjust to: they need to fill in the gaps with AI, or they need to clean up the mess after AI does something in not the right way.
"A lot of organisations are simply not prepared to do the hard work necessary to implement AI."
To use AI well, businesses would have to change their practices so they can complement the capabilities of the technology, which generally requires significant infrastructure work, capital investment and human labour.
The technology also works best when it is purpose-built using specific, high quality data for an organisation's specific subject area.
This means having lots of smaller scale technologies, which runs counter to the prevailing understanding of AI.
"There's this real push towards universal models - something like ChatGPT - it's the one model to rule them all, one solution to every problem," Dr Sadowski said.
"It means that you can sell the technology to every market and from the government's point of view means that all you have to do is implement this one solution.
"There's something very seductive to that because it tells a good story ... but it doesn't produce good technology."
Research about AI's impact on productivity shows mixed results.
A recent CSIRO study of 300 employees found one in three did not report productivity benefits, and the majority that did expected the improvements to be better than what was delivered.
Analysis published in the US National Bureau of Economic Research showed unclear results at the organisational level and it can also be difficult to disentangle the impact of AI from other factors.
