This article features early insights from Apolitical’s AI Readiness Check, which has been used by 2,000 public servants globally, including 100 in Australia.

In the last few years, some of the most important stories about the pitfalls and potential of artificial intelligence (AI) in government have come from Australia. As more public servants and their leaders grapple with how to make the best use of AI, there are important lessons to learn from their Australian colleagues.

Lesson 1: Give people guidance

Everyone remembers the New York lawyer who made global headlines when his court filings cited legal precedents which did not exist. Six months later, a much more serious story emerged about an Australian social worker who used ChatGPT to analyse the case files of a vulnerable child potentially at risk of abuse. Aside from the clear breach in data protection, there were also concerns about the fanciful way ChatGPT had misinterpreted important facts of the case — analysis which found its way into the final report submitted in court.

But this wasn’t a story about one individual’s unfortunate use of AI. The more important lesson, which a subsequent investigation emphasised, was the lack of training and guidance which the agency had put in place to support staff. They found that “there were no departmental rules in place about when and how these tools should or should not be used.” In the absence of clear guidance, 100 child protection reports were produced with the help of AI and nearly 1,000 colleagues across the agency were making use of these tools. When we surveyed public servants in June last year, just 35% said they had received guidance on how to use AI at work. In 2025, early data from leaders using our AI Readiness Check suggests that only around half of public organisations have developed detailed advice for staff. That’s an important gap to close.

Just 35% of public servants say they have received guidance on how to use AI at work. (Apolitical community survey, 2024)

Lesson 2: Mindset matters

The story captures a challenge facing governments around the world. Chris Fechner is CEO of Australia’s Digital Transformation Agency, he was listed in Apolitical’s AI 100 list and recently spoke to us about Australia’s approach to AI. He says that governments need to figure out how to “Safely, quickly, and affordably… change our workforce, processes, systems, and governance to… realise the intended benefits of AI while avoiding the downside and unintended outcomes.”

Whether governments can achieve that transformation will, in part, depend on the willingness of public servants to explore new technologies and learn how to embed them into their work. Several frameworks which define the capabilities that public servants need to develop in the age of AI emphasise more psychological capabilities like having a positive attitude toward AI (European Commission) and being more curious, experimental and adaptable (UNESCO). Fechner echoes the same point. He says that public servants (especially leaders) need an “[E]xperimentation and growth mindset.” He says that “AI represents an opening of completely new possibilities and new ways of delivering government for the benefit of all people.” Our early evidence from 2,000 public servants globally shows very high levels (above 90%) of optimism and open-mindedness about AI’s potential in government.

Lesson 3: Empower public servants

One way to build optimism and open-mindedness is to give people exposure to the opportunity. That's what Australia did. Last year, over 7,000 public servants across 60 government agencies participated in a six-month pilot study to see how generative AI could be used in their work. “Staff saved an average of one hour per day on administrative tasks,” Fechner told Apolitical. And of that time saved, “40% was spent on learning and team collaboration,” he said.

An extra five hours a week would be a significant gain for public servants, especially if they could re-invest the time in building new skills. However, evidence from the trial suggests that these kinds of productivity gains will depend on how well public servants are able to use AI. Only 1 in 3 participants on the trial were daily users of CoPilot. And, crucially, those who participated in more training sessions were nearly 30% more confident in their ability to use it.

The lesson from Australia’s AI development in recent years reaffirms an important principle: People must be at the heart of any plan for tech innovation. Or, as Fechner puts it: “[W]e need to dispel the myth that AI is some magical, all-seeing, all-knowing capability that can immediately and permanently solve all existing problems.”

If governments globally want to learn from the experiences and example of a people-powered approach to AI transformation, looking down under is a great place to start.

*Explore the AI Readiness Check, fast, interactive tool for public servants to plot their pathway to AI success in government, provided for free as part of the Government AI Campus. *