Source : Perth Now news
Australian usage of artificial intelligence is skyrocketing, but so are fears around misinformation, security and cultural bias that come with the technology.
A report published as a partnership between the Australian National University and Google probed adoption and attitudes towards AI, finding almost half of all the nation’s adults use the technology.
That figure is only slated to grow as take-up surges among young people.
But eight in 10 Australians were concerned about the growing use of generative AI in politics.
Worries included spreading misinformation and the risk that sensitive political data could be leaked to the wrong hands.
The report’s lead author Jessica Herrington agreed those concerns were justified.
“There’s lots of examples we’re seeing now with deepfakes and the spread of misinformation online that’s quite concerning,” the ANU researcher told AAP.
She acknowledged many of the risks cannot be fully mitigated, but awareness of any potential issues is a great start.
More than 3500 adults were surveyed, and interviews were conducted with senior leaders of many Australian organisations adopting AI systems in the workplace.
Prominent AI systems include models such as ChatGPT, Google Gemini, Dall-E and Claude, all of which were developed in the US.
Dr Herrington worried Australia is becoming reliant on overseas AI models, which may have cultural bias built into their operating systems.
“All the big models that come out of the US are built on primarily data from American websites so they’re going to be more representative of white, male, American-centric values,” she said.
A more sustainable approach would see the government invest in domestic development of generative AI models, Dr Herrington said.
That could also mean each model has one specific task it focuses on, rather than catch-all larger versions.
“For Australia to have models more tailored to our specific demographics, it might be better to have smaller models that are tailored and tuned to our specific purposes,” Dr Herrington said.
“It would be much more efficient and successful if government partnered with industry to do those kinds of things.”
About two-thirds of survey respondents thought it was the government’s responsibility to govern generative AI usage.
