The chief executive of the CSIRO’s Data61 unit Adrian Turner has called on Australia to build sovereign capabilities in ethical artificial intelligence, saying the technology has the potential to drive competitive advantage across other industries.
“Economically, every industry will become more data driven. Machine learning and AI sits in the centre of this and can give us that competitive advantage,” he said during his opening keynote at a Committee for Economic Development of Australia AI event on Friday.
“We need to lean in to this and create sovereign capabilities. We can’t import values-based systems for ethics. We need sovereign AI capability, just as we need sovereign cybersecurity capability.”
According to Mr Turner, CSIRO has calculated that if Australia increases its use of ethical machine learning based systems and apply it to industries such as finance, mining and energy, the country has the potential to tap into a $315 billion GDP opportunity.
He warned, however, if Australia fails to lead in ethical AI, the impact would be devastating.
“If we get it wrong, what we see is a global platform company come in and dominate our industry sectors and extract the value out of our value chain,” he said.
“We’re optimistic that we have all the ingredients here and we think AI machine learning is at the core. We think ethics and the way we approach and deal with ethics is at the core. We think the opportunity to build new tools could put Australia in a really strong place in a global context.
“This opportunity is perishable for the country. There is urgency we need to move now,” Mr Turner said.
Mr Turner’s calls to arms comes as public consultation on Australia’s ethics framework on artificial intelligence wrapped up at the end of May, which saw over 100 submissions. The feedback that was received would be used to inform the Australian government’s approach to AI ethics in Australia.
Standards Australia policy manager Jed Horner, who joined Mr Turner on a panel discussion during the event, said that setting ethical AI standards had to be a joint effort.
“The hard work will begin when committees comprising of corporate, government, academia, civil society come together. They all need to be around the table, but whatever we do has be commercially realisable and real, and it must resonate with the broader community,” he said.
“We need a strike consensus, which is not what we always see with purely government-driven processes or purely market-driven processes.”
Mr Horner said despite differences, he’s optimistic that it would be possible for all different parties, domestic and internationally, to work together, pointing to how information security standards sets a positive precedent.
“We actually do have international agreements to cybersecurity standards. It’s called ISO27001. It might not solve the world’s problems but it’s used by our major banks … [and it’s used and mandated across the NSW government, so we do have precedent where the corporate world, where experts in computer sciences, and different governments do play together,
“And that’s the historical example to follow with AI.”