Two-thirds of Australians say there are not enough laws or regulations preventing artificial intelligence from being abused, and fewer than half trust AI to be used safely at work.
A report by consultancy firm KPMG and the Australian Information Industry Association (AIIA), the industry peak body, found Australians expect AI to be regulated and want an independent regulator to monitor the technology as it booms into mainstream society.
"While regulatory frameworks have been developed to tackle issues related to privacy, there has been little progress towards a more holistic framework that incorporates AI," the authors found.
The authors found Australians were "over the hype" of AI and had moved into the "trough of disillusionment" and found that same uncertainty spilling over into conversations with industry and government, who were uncertain whether to approach AI with "outright resistance" or to explore the opportunities it presented.
Industry Minister Ed Husic is due to release a review into recent rapid developments in AI and how the government should respond to the emergence of technologies such as Chat-GPT, Midjourney and others that could disrupt many industries.
Mr Husic has been contacted for comment.
KPMG estimated AI could add 1.2 million new "technology" jobs over the next decade and 5.3 million more jobs created by advancements in technology over that time — but noted most Australians believed disruption from AI would eliminate more jobs than it would create.
The firm estimated that $370 million was added to Australia's economy each year by AI and expected that could explode to be worth more than $315 billion by 2028, as forecast by the CSIRO.
AI is already well established across government, mining, healthcare and other major industries and has been used to dramatically shorten the time to process tax returns, anticipate bushfires, assist in diagnoses using medical imaging and detect fraud, among many other things.
But in recent months, powerful new programs have become available to the general public, prompting fresh concerns AI could easily be abused.
Programs like Open AI's Chat-GPT, which recently was hooked up to the internet, allowing it to search for flights and make purchases, and image generator Midjourney, which had to ban prompts of arrested celebrities after artificial photos of former president Donald Trump being arrested went viral, are advancing at a pace that has caught governments globally off guard.
Industry group recommends National AI register and close watch on EU
The AIIA said Australia was now at a point where there was a need for "guidelines" and "guardrails" but noted the difficulty for government in regulating AI because it had so many different applications.
The group says the absence of a specific policy on AI was a major obstacle to the sector, and that while existing laws could be used in cases where AI causes harm, those laws do not ensure AI technologies are designed or used safely.
It recommended Australia pay close attention to the AI Act in the EU, which, if passed, would become some of the first comprehensive laws regulating AI anywhere in the world.
The AI Act would ensure AI ethics and principles are protected, take a risk-based approach to regulation that categorises the potential for technology to cause social harm, and would accompany laws to create an easier process for victims of AI technologies to bring civil suits.
The EU is also planning to establish a public database of AI technologies to assist in monitoring the sector.
Speaking at the Australian National University, Shadow Communications Minister David Coleman said the government needed to develop a national plan for AI.
"We need to be careful that as regulators, as governments, we don't step on that too much," Mr Coleman said.
"But the bottom line is, we do need a national plan to deal with this issue because it's going to be immense."
The report authors said the federal government should consider establishing a national AI register that tracks technology and its uses, a designated responsible owner for those systems, and identified risks of the technology.
