Key insights from Agency in an AI-enabled world
As artificial intelligence becomes increasingly embedded in public sector work, a central question emerged at ANZSOG and the Forum of Federation's recent tri-national conversation Agency in an AI-Enabled World: how can governments harness AI's benefits while retaining control over critical decisions, data and public accountability?
The ANZSOG online event brought together experts from the ‘middle powers’ of Australia, Aotearoa New Zealand and Canada for a discussion covering data sovereignty, partnerships, and options for middle power co-operation and open-source international standards for AI use.
Speakers Emma MacDonald (Director of the Centre for Data Ethics and Innovation (CDEI), Yannick Newton (Director of Intergovernmental Cooperation, Institute for Citizen-Centred Service) and event host, ANZSOG’s Kate Carruthers (Head of Data, Analytics and AI), presenter of our AI Tools & Ethical Practice program, explored what responsible AI adoption could look like in practice, with a focus on government agency and control over key decisions.
Ms Carruthers pointed to the need for governments to build partnerships around AI, and asked the panel what capabilities or decisions should remain held internally, given rapid geopolitical change.
“Realistically every government agency now has data in US-based clouds even if it is resident in your own jurisdiction. That data is subject to the Cloud Act so the US government can reach in and grab that data,” she said.
Ms MacDonald said that governments needed to think more deeply, not just about the technological aspects of AI, but also the infrastructure around it.
“We need to understand the layers that sit within that, and where the accountability and decision-making sit,” she said.
“Working together has been pretty beneficial for everybody, and particularly New Zealand. We're quite a small nation and we need to be trading and talking to the world to get these conversations going. But we also need to know which are the bits that are really important to our economy, and our society, and the areas where we need to maintain critical oversight on the decisions,” she said.
“A lot of these conversations will come back to how we're protecting our data, and we need to get more serious about the protection of it.”
Mr Newton said that governments were dealing with a potential loss of sovereignty, and needed to build new partnerships with each other to address this.
“There was a defining moment for me last year when we had a Microsoft Director of Public and Legal Affairs testify to the French Senate saying that they could not guarantee that French citizen data would never be transmitted to US authorities,” he said.
“What struck me was that we're not talking about multinationals, we're talking about very large, globally reaching national companies.
“If we take a look at the state of our current geopolitics and the state of trade, we need to diversify our partnerships and work with others more broadly. When I think about AI specifically there's a pace here that we've never seen before, so we really have to do something different.”
Is AI already improving the quality of government?
The rapid spread of AI contains both risks and opportunities for governments, and discussions covered both the appropriate safeguards and whether AI was already improving the quality of government.
Ms Macdonald said that New Zealand was not – so far – seeing the productivity gains they had expected.
“They're not saving as much time and resources as they thought. There's definitely some benefits that we will be able to see, particularly in education and health and being able to flatten out the world for people and be able to support them,” she said.
“But we need to be looking at whether we are seeing better, more equitable decisions. Do we have humans with authority in the loop, and if anything bizarre starts coming out of the models have we got eyes on that? Who makes the actual critical decisions? If it’s just a person hitting a ‘Yes’ button, then we know from many technology revolutions that humans will just lean in and be like ‘oh yeah it's probably right’ we'll just tick it off.”
Mr Newton agreed that human decision-making was key but that AI was already showing promise at speeding up public sector work.
“Where AI plays well within government is the ability to accelerate our collaboration. If we have a number of departments that are going to issue grants and contributions is there a way that we can solve that problem? With AI the speed at which we can gather that information allows us to come together around the problem.”
What does AI mean for public servants’ daily work?
The panel took questions from the online audience, who were concerned about the impact of AI on their work and whether public servants could have the time and confidence to question AI’s outputs
Ms MacDonald said that public sector leaders needed to be able to support staff in order to have ‘safe places to play’ with AI.
“In New Zealand, our digital agency has created a number of pieces of training at the different levels for people from senior leadership all the way through the work chain to be able to find how to use AI for their own purposes and is encouraging people to be able to test that,” she said.
“AI has to sit at the most senior part of our organisations so we can understand what is critical for us, as nations, to be able to hold tight and where we partner and outsource. It's not for the 'tech bros' anymore. It's for every single person that works in government.
Mr Newton said it should be a priority for all public servants to get their heads around AI.
“We all face it daily in the tools that we use, on social media. We all need to be well aware of what it is, how to use it, how to use it well. If I would encourage one thing, it would be to have training on AI for all public servants and to do it very quickly to make sure that everyone's got that foundational base in place,” he said.
The session concluded with a call for stronger AI literacy, safe experimentation and leadership accountability. Speakers emphasised that responsible AI adoption will depend not only on technical capability, but on the ability of public servants and institutions to ask the right questions, understand the risks and use AI in ways that strengthen public trust.
Continue the conversation: AI Tools & Ethical Practice
For public sector professionals looking to move from strategy to application, AI Tools & Ethical Practice is the ideal next step. This practical course explores how to use AI tools effectively while navigating ethical considerations, governance requirements and responsible decision-making. You'll build the confidence to apply AI in your work while ensuring transparency, accountability and public trust remain at the centre of every decision.