How Australian Public Servants Can Start to Build AI Muscle

In this article we discuss:

• Training Plans: Personal and Organisational
• Treating AI Like a New Team Member
• Taking Responsibility for Your New Power
• Small Steps You Can Take Today
• Why Little and Often Lead to Progress

“AI can be, quite frankly, overwhelming.”

Judging by the number of thumbs up, this was a feeling that resonated strongly with an audience of 700 public servants when I presented at APS Innovation Month*. The topic was ‘Innovating finance with AI’, which we know is going to reshape how we operate.

AI is all around us in our day-to-day lives – even if we don’t always realise it. When I put my destination into Google or Apple Maps, it’s AI and machine learning that tells me the fastest way to get there based on current traffic and roadworks. AI is working behind the scenes in every music or podcast recommendation. Every web search. Every social feed. And many work interactions. Across the private sector, organisations are increasingly moving from AI pilots to widespread use.

There’s a strong imperative to harness this technology in Australia’s public service, where AI adoption has historically lagged. According to Adapt research, Government CIOs are spending 2% less on R&D than their private sector counterparts, hindering the adoption of cutting-edge technologies like AI and advanced analytics.

So how do we get started? How do we build our knowledge and confidence to the point where we can proactively apply and leverage the benefits of AI, every day?

The technology is evolving at a rate of knots and, because of this, no one single training course is the panacea. The pressure to respond is constant and the risks feel high. As more new tools and capabilities proliferate our personal and work environments – most people find AI completely overwhelming.

This is not just the case in the public sector. Research finds the greatest barriers to all organisations adopting AI tools is less about cost or risk – and much more about people not having the bandwidth. In Deloitte’s CFO Sentiment report of senior finance executives in Australian listed companies, the top two barriers were competing priorities (43%) and a lack of skills (34%).

In other words, it’s easier for busy people to keep on doing what they know – rather than make time for learning something new.

But make time we must.

I’ve never run a marathon. Like the speed of AI development, the thought of it overwhelms me. The furthest I’ve gone is 10km – only to be reset to zero by injuries or competing priorities, built back to 5km, and back again to zero!

Just like gearing up to run a marathon, acquiring AI fluency requires careful preparation and regular practice. Nor can we expect to run the full 42km immediately. If our starting place is the sofa, we have to start with 5km at walking speed and gradually increase the challenge of pace and distance – building ability and confidence over time, and mitigating the risk of injury or set backs.

It’s a concept at odds for mid to late career professionals, who may feel the time of deep learning and foreign concepts is behind them.

But we have to go slow before we can go fast.

Training Plans: Personal and Organisational

When it comes to building AI capability in our public service teams, I like to think of it as a training plan with two levels:

• Personal training – Each of us, as both human beings and public servants, needs to start building our own AI muscle. That might mean signing up for an APS Academy course, listening to podcasts in the car or experimenting with AI tools on non-consequential tasks. I started by asking an LLM to suggest a lower carb nut-loaf recipe after I realised my bread intake had exploded since having a young human in my household. That experience quickly taught me the art of prompting: be specific and clear, break complex tasks into steps, provide examples of what you like. It’s OK to have fun! The important thing is to start somewhere safe and low-risk.
• Organisational training – At the agency-level, we can create communities of practice, where questions can be asked (even anonymously) and answered by experts. At Workday, people post anonymous questions on Slack and our AI experts produce videos in answer – tools that also form a learning library. Agencies also need to build a culture of curiosity and psychological safety by making space for experimentation and being overt about learning from missteps. When we make it OK for people to ask “silly” questions and learn from failure, they are more willing to try something new, to admit they don’t know, and to share their experience of a mistake that leads to learning.

Treating AI Like a New Team Member

One of the biggest mindset shifts is learning to treat AI as a teammate. Too often I hear, “I tried the AI, but the recommendations weren’t useful, so I turned it off.”

In response, I challenge people to change their thinking. If a graduate or new starter joined your team and their first idea wasn’t right, you wouldn’t exclude them from future meetings. You’d coach them, provide feedback and help them understand the contexts and nuances that you’ve learnt from experience.

The same applies to AI. Without feedback, it will never improve. With guidance, it can deliver increasing levels of value.

And it can help you with your own training. When you get to the desired result after numerous prompts, try asking your LLM, “How could I have asked this in a different way to get the same outcome faster?”

Taking Responsibility for Your New Power

Building AI muscle isn’t only about technical skills. It’s also about understanding how to avoid a clever AI tool breaching ethical standards, privacy or trust. And those in the public sector have an added duty of care.

At the Public Sector Network’s recent Local Council Roundtable, one of the key questions raised was: Do we have the public’s buy-in and consent to even play in this space? As custodians of public trust, we need to ensure that AI is deployed in ways that are transparent, fair and responsible.

That means making sure your agency has guardrails in place, being clear on when and how AI should be used, and asking the right questions before rolling out internal or citizen-facing tools. The Standard for AI transparency statements provide guidelines to support our decision making in this space.

Small Steps You Can Take Today

If training for a marathon starts with a walk around the block, then AI adoption starts with small, low-risk experiments. You and your team might like to:

• Listen to podcasts. Some of my favourites are: The AI in Business Podcast to discover how organisations are applying AI across the front, middle and back office, or the Workday Podcast – available on Apple and Spotify – to understand how organisation are more broadly building transformation capabilities.
• Take advantage of free and low-cost training from the APS, professional bodies like CAANZ or our Workday AI Masterclass series.
• Understand your agency reporting obligations by leveraging learning and policies published by the Digital Transformation Agency, or your relevant state or local government policy and framework.
• Share what you learn with your team as you practice with non-consequential outcomes and available tools.
• Establish cross-agency networks to share learnings and avoid reinventing the wheel.
• Use existing government frameworks and standards to guide safe and responsible practice.

These early steps may feel small, but they build your ability and confidence to take on larger challenges. It might seem a strange way to go about acquiring a career-defining capability but, as Einstein said, “You can’t use an old map to explore a new world.” And we’re in a new world now – where even learning is different and ongoing.

Little and Often Lead to Progress

Anyone who’s trained for a marathon (or any sort of endurance activity) knows that consistency matters more than speed. The same is true for AI. There are no short cuts to mastery. But if we keep practicing and learning, we will rapidly acquire the skills to make ever more sophisticated use of these powerful tools.

As AI-powered services become standard practice, each public servant has a role to play: to build our own AI muscle, to support our teams to do so, and to act as responsible custodians on behalf of the communities we serve.

While the task may seem daunting, the prize for getting to the finish line – a more capable, efficient, and citizen-centred public service – is well worth the training effort.

* You can view my presentation and others like it at the APS Members Community Platform.