For public sector leadership on artificial intelligence, set your sights on Defence

Powered By

When it comes to modernising government agencies, machine learning and artificial intelligence – or AI – provide a range of exciting opportunities. AI can make predictions about potential non-compliance and decide on the optimal response. From a service perspective, it can identify what people need and help them get the right information, the right service and the right support. 

“This can have a profound impact on people’s lives – we’re talking anything from greater convenience to literally life-changing support,” says Darren Menachemson, chief for digital societies and chief ethicist with ThinkPlace.

Despite Menachemson explaining the applications for AI-augmented regulation and servicing are “almost endless”, the technology is not being used to its full potential. “In truth, most implementations of ‘AI’ are not being driven by machine learning,” he says. “Instead, they are just sophisticated applications of more traditional analytics.”

The exception is the Department of Defence. An AusTender keyword search for “artificial intelligence” shows a range of investments in research and development of AI applications since 2015 – including research into AI algorithm assurance, AI and ethics, and a range of proof of concept projects in partnerships with research institutions and industry partners. 

In addition, programs have been established for further partnerships and investment in AI, including the Next Generation Technologies Fund, Defence Artificial Intelligence Research Network and Centre for Advanced Defence Research and Enterprise.

“Artificial intelligence is a key technology enabler for Defence and society at large,” a Defence spokesperson explains. “To effectively harness AI, we need to ensure we pick the right tool for the job at any given instance. “This forms a body of work in itself that Defence is seeking to solve. How do we create an AI that is able to appropriately select the right approach to solve a problem that it faces?”

In addition to using AI within autonomous systems, Defence is researching AI to assist in planning, interpretation of remotely sensed data and other ways to use it within the cyber domain. “We’re also looking at ways that we can accelerate AI uptake, including software frameworks to scale algorithmic development.”

For partners, including industry, Defence is seen as the place to be for research and investment in AI.

“We definitely see Defence as being one of the leaders in AI in the nation,” says Zygmunt Szpak, executive director of Insight via Artificial Intelligence. “Part of the reason for this is that Defence understands that adversaries are doing something in the space, and the nature of Defence is you have to be on top of the latest stuff. Otherwise, you just can’t defend yourself. 

“That’s one of the reasons why we find that a lot of our interesting work is with Defence.” 

Is the Defence investment in AI transferable?

The defence context for AI is “far more austere and challenged” than in other areas in the public sector, Defence says. Its approaches need to operate in environments subject to deliberate adversarial action. But despite this, knowledge gained through Defence research could transfer to other areas of the public sector. 

“Cyber is a great example,” Szpak says. “This is an area of AI Defence is pushing the envelope on and can quite easily be applied in an industry context.” 

Work on explainable AI systems, which better integrate machine learning with human insights and interpretation, is another area with a wider impact on the public sector and industry.

“Organisations will be able to look at key algorithms developed in Defence to port and deploy them in a different space,” says Szpak. “You’re never going to get a particular product, but that knowledge is transferable.” 

But the reverse wouldn’t be true. 

“As these challenges are almost unique to the Defence context, we can be considered a leader in the innovation, science and technology development associated with dealing with these types of challenges,” Defence says. 

“Additional robustness and resilience come with a cost, meaning that the approaches developed to operate in such challenged environments may not be optimal when challenges aren’t present.”

What agencies can learn from Defence

For the public sector, learning about AI from Defence can be a challenge – after all, they operate in a highly classified environment. Having a foot in the door with Defence also helps, with communities of practice providing forums to discuss research.

“There’s a community of interest in Defence around machine learning and AI that I provided a talk to,” Szpak says. “As part of that group, there were some people from other public sector agencies. That tells me that already there is some kind of transfer of knowledge happening between researchers in different parts of the public sector.”

Attending workshops on artificial intelligence hosted by experts working with Defence, such as Szpak, can also help agencies gain the full potential of AI for their operations. Defence is keen to be more open with insights and be seen as a public sector leader, with conferences a key forum for collaboration.

In July, the Defence Science and Technology Group held the Australian Defence Science, Technology and Research Summit, which featured AI as one of three tracks. And a range of AI experimentation programs, including the Army’s Robotics and Autonomous Systems Implementation and Coordination Office, Air Force’s Jericho and Navy’s Warfare Innovation Navy Branch, are all promoting their work publicly.

With expertise in the application of AI, Defence has some key insights to support its successful implementation within any agency.

“To succeed at harnessing AI, the required capabilities include not only access to sufficient computer resources but also people who understand the current technology offerings and ensure it is only applied to the type of problem appropriate for that technology’s maturity,” says a Defence spokesperson. 

“Understanding these limitations also assists in identifying areas in which innovation is required to advance AI, helping to produce more robust solutions. Defence has an AI Centre, which serves in this capacity across the department.”

Read this playbook for guidance through the four steps of the IDEA approach that Microsoft has developed: Digital Transformation Playbook: Four Elements to a Successful Strategy.