Computational power and data centres have occupied much of Australia’s discussion around AI. But the nation’s universities have been grappling with other questions beyond these headline-grabbing topics.

AI sovereignty, and how and where Australia can invest to safeguard against potential political and commercial headwinds, is something AI leaders in the nation’s universities are considering, too.

Leaner, purpose-built AI

Whether for the nation or institutions, sovereign AI may not mean pursuing our own frontier models, says Dr Sue Keay, Director of the UNSW AI Institute. 

So rather than advanced general-purpose models, it could mean investing in developments that play to Australia’s strengths. That includes computationally leaner, purpose-built models tailored for the nation’s specific needs.

“There’s a lot of value in us building our own models, they just don’t need to be at that [frontier] scale,” Keay says.

But investment is needed, and Keay says it remains insufficient to grasp the opportunities before Australia. 

“For many of us [in the field], we’re not paying enough attention to developing those specific tools where Australia actually could have our own advantages,” says Keay. “Taking advantage and getting value out of a lot of our data sets that we don’t want to open up to foreign companies.” 

“At the moment, we’re not seeing a lot of investment in that space, and that makes it hard to really support the development of our AI capability.”

Keay says it’s disappointing, because Australia has proven its capability for developing technology and AI in resource-constrained environments.

Next-generation thinkers

Though universities are beginning to develop infrastructure to meet AI needs, Keay says it’s also important they equip the next generation of thinkers with the skills and know-how to develop local capability.

“Australian universities are probably one of the few places where researchers – and people in general – have the opportunity to experiment and develop AI for discovery purposes,” Keay says.

“There’s a lot of value in being able to develop that AI capability, of being able to develop a novel pipeline of research data, and developing a model that can help answer basic scientific questions. Those are exactly the sort of skills that we need in order to not only build our own AI but also to be able to evaluate the AI that we’re importing from other countries.”

Beyond sovereignty

It’s becoming clear that Australian universities should prioritise education and research in AI, with strong support from the Australian government. At the same time, university research can contribute beyond addressing technical questions. Professor Stephen Garton, President of the Australian Academy of Humanities, says there are important questions universities can help to answer.

“If we just focus on data sovereignty … that’s only 20% of the important conversation,” says Garton. He says ethical use, government and policy requirements, the use of AI in pedagogy, its impact on research and the community at large are also important. 

“We can’t stop it, so let’s put in place the appropriate guardrails and policy platforms to make sure we use it well,” says Garton. “And use it ethically and appropriately, and ensure that there are good jobs for people in the sector, new jobs that we haven’t imagined yet.”

– Written by Matthew Ward Agius