AI Proof Of Value In Government Starts With Scale And Safeguards

In this interview, Benji Crooks, Marketing Director at Public Sector Network, speaks with Peter Meere, Director, Office for Artificial Intelligence at the Department of Treasury and Finance, about where AI is making a measurable difference across government, how agencies should think about proof of value, and what it takes to select and use AI tools responsibly. Peter will appear at Government Innovation Showcase South Australia as part of Government Innovation Week South Australia on Wednesday, 10 June 2026, presenting "AI State Briefing: AI Proof of Value & Government Service Innovation"  

Benji Crooks: Great. So first of all, if you could just introduce yourself—your name, your title, and the company you work for.

Peter Meere: I am Peter Meere, Director of the Office for Artificial Intelligence. We’re a new state government entity set up to help state government do AI well.

Benji Crooks: Excellent. And obviously the big topic is AI at the moment. I’ve just come from the Victoria event and it was AI I was thinking—from every entity possible at the event. From your perspective, where do you see AI as being making real change at the moment across governments?

Peter Meere: The primary area we’re seeing AI make a real difference in government is freeing up frontline service staff time.

Our staff in hospitals, in education and schools, and in police face heavy administrative workloads. AI is great at automating routine tasks, freeing up their time so they can spend it on high-value, human-facing operations.

Benji Crooks: Excellent. And I guess you’ve kind of just answered the question I’m about to ask—but when we talk about proof of value in AI, what does that success actually look like for government agencies?

Peter Meere: Success for us in the proof of value concept looks like moving beyond experimentation.

We’re seeing a significant amount of activity across government focused on proof of concept—experimenting with different ideas. What we’re trying to do, from an Office for Artificial Intelligence perspective and a South Australian Government perspective, is move from isolated value to more systemic value.

That looks like rolling out AI solutions at greater scale, ensuring all the necessary safeguards are there, and at the end of the day being able to measure some sort of return on investment from implementing the AI system.

In practice, that looks like freeing up people’s time, delivering better quality services, and overall being a better and more productive government.

Benji Crooks: Absolutely. And we talked a little bit about safeguards there, but with all the different AI products that are out there in the market—and there’s so many every day coming out—how do you identify the best fit for use within the government?

Peter Meere: It varies. I’m of the view that what is the best fit for government depends on what we’re trying to do with it.

What needs to be in place in a policing context and what needs to be in place in a health context is very different to what needs to be in place for routine government administrative work.

As government does more and more AI applications across its work, the criteria we need to be focused on are the appropriateness of the AI model that is being used, what government data and other data it’s drawing from, and the appropriateness of its outputs in use.

The remainder of it goes to the human factors. The things we teach people and build capability on—what to watch out for, what they need to be wary of in using the tool—and, as always, maintaining human judgment at the end of the day.

Benji Crooks: You come face to face with a lot of these every single day, but what are some of the common misconceptions that people have about AI within government?

Peter Meere: Probably not in a government context—these are more societal implications of AI generally.

Some of the big misconceptions we see vary between either it’s absolutely right all the time or it’s wrong all the time. In fact, it’s neither of those things. It’s somewhere in the middle.

At the end of the day, the current crop of large language models are high-powered probability machines, and we need to treat their outputs as such. That means we always need to apply a level of human judgment and discernment in what it’s actually giving us.

As a workforce in the public sector, what we’re trying to do is move beyond substitution to a place where we’re integrating these tools as part of our everyday work, and we’re using our judgment, taste, and discernment to understand when it’s appropriate to use them, how to use them well, and what outputs they’re producing for us.

Benji Crooks: I’ll just end with this one question then. What is one thing you’re hoping the attendees will take away with them after you’ve spoken at Government Innovation Day?

Peter Meere: The one thing I’m hoping all government employees take away from the Government Innovation Showcase is curiosity.

Government has these tools already. The trick for government now is thinking about how you can apply those in your own work effectively.

Then thinking beyond that: how might you apply AI to improve your business practices in the business unit you work in?

For more senior people and managers: how might you apply AI to deliver a better service out into the community?

And then for our agency and more senior bureaucrats: how might we use AI to build better government services for the community—more cohesive government services?

That’s really the call to action that I’ve asked many people since coming into this role. It’s the one that’s going to serve us well into the future by always ensuring that our AI use is anchored in public value and benefit to the community.

Hear Peter Meere live

Hear Peter Meere at Government Innovation Showcase South Australia as part of Government Innovation Week South Australia on Wednesday, 10 June 2026. Peter’s session will outline what “proof of value” should look like for government AI initiatives and how AI can support service innovation at scale.