How the Australian government is approaching AI

Artificial intelligence (AI) can be transformative in government, but its implementation has to be done in a way that drives positive outcomes while mitigating its downsides, according to the head of Australia’s Digital Transformation Agency (DTA).

Speaking at the Gartner IT Symposium/Xpo 2023 in Gold Coast last month, Chris Fechner, CEO of DTA, revealed that the agency has provided the government with “tactical guidance” on how to approach generative AI adoption in the public sector while it figures out how to harness the technology to benefit citizens and businesses.

“It’s to make sure you don’t bleed out while you’re figuring out where the hole is,” Fechner said, adding that the agency is deliberately experimenting with use cases in a safe environment.

“We know it can do really good things – it can provide you with translation services, guide you to services that you're eligible for or take out the complexity of needing to understand large regulatory or legislative environments.

“It can speed up service delivery by eliminating what is still very much a paper-heavy service delivery capability in the Australian government. It can help narrow down and help detect fraud – the opportunities are limitless,” he said.

The enthusiasm around generative AI, however, should be tampered with some of the technology’s risks and what could happen when things go wrong, such as the government’s “robo-debt” scheme that used a flawed algorithm to claw back welfare payments.

More recently with generative AI, Fechner pointed to concerns around surveillance in the private sector, AI hallucinations, and problems with biases that could go unchecked. “We need to make sure that we’re really thinking about how we use this for positive outcomes, but be conscious all the time, not just when we deploy it, of what the potential negatives are going to be,” he said.

Fechner said having capability and leadership in government are going to be key in harnessing the technology. That means deeply embedding expertise in government and working with industry to ensure that the government can deliver increased amenity for citizens and businesses.

“If we can do that, we’re going to get that virtuous cycle function – the better we are doing these things, the more we can focus on extending, not just delivering, what we currently do. That’s going to be where we see the power of generative AI – it’s going to allow us to extend what we can do within resource constraints.”

The DTA is not working from a blank slate. In 2019, the government laid out a principles-based approach on the ethical use of AI, including the respect for human rights, explainability, fairness and accountability, among others. Fechner noted these principles are continuously evolving and that the government is learning from the experience of European and US regulators that are applying similar principles.

Early this year, the DTA also released a draft consultation paper on the government’s data and digital strategy, which includes a commitment to harness analytical tools and techniques, including machine learning and AI, to predict service needs, gain efficiencies in agency operations, support evidence-based decisions and improve user experience. The final strategy and implementation plan is expected to be ready later this year.