In the weeks leading up to the election, I spent some time pulling together the AI policy statements of New Zealand’s political parties. I did it out of personal interest because I wanted to understand what our politicians were saying about a technology already changing our lives, our work, and our classrooms – and also so that I could participate in conversations with colleagues from an informed perspective.

The exercise left me with a collection of positions (you can access my summary here), but not a clear picture of where we are heading as a nation. Some parties emphasise innovation and light-touch rules; others propose new oversight, stronger protections, or limits on infrastructure. The Good Ancestors AI Election Scorecard adds a useful risk-focused view of where parties stand. But a scorecard, like a collection of policy statements, can only take us so far. What I found myself looking for was a shared direction – how can we make the most of AI’s potential while preparing people for change and protecting them from harm?

That question feels especially timely as parties work through the formation of the next government. The choices made now could shape New Zealand’s AI direction for years. And they matter to educators in particular, because the decisions we make about AI will help shape the world our children inherit.

A compass with three points

As I think about this I keep returning to the Brookings Institution’s Prosper, Prepare, Protect framework. It offers a practical way to hold three responsibilities together: make it possible for people and communities to benefit from AI; prepare learners, workers, and institutions for a changing world; and protect people from harm. Brookings developed this framework for education, but the questions it raises belong in national policy too. Brookings’ report argues that AI can enrich or diminish learning depending on how it is used, and sets out actions for governments, technology companies, educators, families, and others.

• Prosper means more than attracting investment or adopting new tools quickly. It asks whether AI helps solve real problems, strengthens public services, supports creativity and productivity, and shares its gains fairly. It also means asking what infrastructure AI requires, and who pays the costs in energy, water, and public resources.
• Prepare means building the knowledge and confidence to live and work well alongside AI. We need informed public servants, capable businesses, supported teachers, and students who understand what AI can do, where it falls short, and how it can affect their choices. Preparation is not simply training people to use the latest tools. It is helping them judge when those tools serve a human purpose.
• Protect means putting clear, enforceable safeguards around the uses that can cause harm. That includes privacy, discrimination, scams and deepfakes, children’s wellbeing, security, and the use of algorithms in decisions that affect people’s lives. It also means independent capability to understand the systems we rely on, investigate serious incidents, and work with international partners.

Protection should not become the whole AI conversation. But credible protections are part of what makes confident innovation possible. The goal is not to eliminate risk before acting – that would be impossible. It is to make deliberate choices, know who is accountable, and give people the means to challenge decisions that affect them.

Guardrails are part of the route

In my thought piece Leading Like a Wayfinder in an AI enabled world, I draw on the 3P framework and use a navigation metaphor for leadership through change. A wayfinder needs a destination, knowledge of conditions, and the ability to adjust course. The same thinking belongs in national AI policy. We need room to explore, but we also need shared bearings and a way to recognise when we are drifting into danger.

A dedicated group or agency could help build that capacity, provided it has a clear purpose, the expertise and resources to do the work, and meaningful input from the people affected. It should help government understand emerging risks, coordinate across portfolios, evaluate high-impact systems, and work constructively with researchers, civil society, Māori, educators, and AI companies. Engagement with companies matters because they build and operate many of these systems. Independence and public accountability matter because their commercial interests do not always match the public interest.

The question is not simply whether New Zealand needs an “AI office” or an “AI safety institute”. Those names can become a distraction. The question is whether someone is properly equipped and accountable for helping the country understand the technology, prepare for its effects, and act when safeguards are needed.

Our tamariki must not be a footnote

For educators, this is not an abstract policy debate. AI is already changing how learners find information, create work, and relate to digital systems (hence the motivation for writing my thought piece). The choices we make about assessment, privacy, access, and classroom use will influence young people’s learning and development.

Brookings’ findings are a useful prompt for taking that responsibility seriously. AI’s benefits depend on how it is used, while over-reliance can affect learning, wellbeing, relationships, and privacy. That is a reason for thoughtful design and preparation, not a reason to abandon the potential benefits. Its recommendations include involving educators and students in design, building AI literacy, preparing teachers, and setting clear expectations for ethical use.

This connects directly with the questions in my three posts titled “Signals, noice and the children in our classrooms” pt.1: reading the headlines, pt.2: watching what governments do, and pt.3: remembering it isn’t about us. The first asks us to look past the noise and test what has actually been committed to. The second asks us to look for institutions, capability, and accountability. The third turns our attention back to the children whose futures are shaped by the decisions adults make now.

Education cannot be left to respond after national settings have already been decided. Teachers, school leaders, boards, students, and whānau need a place in the conversation from the beginning. They understand the everyday contexts in which AI’s benefits and harms are experienced.

Make the next conversation count

It is my view that the new government should bring parties together around a durable national direction for AI, one that can survive a change of ministers or a new technology cycle. It should set out what we want AI to help us achieve, how we will prepare people for change, and what protections must be in place. It should also explain who is responsible, how progress will be measured, and how the public can raise concerns.

Citizens and educators can help make that happen. We should be asking our local MPs what the government will do to build independent AI capability, how educators and young people will be involved, and how the public will know whether commitments are being met. Ask whether policy covers benefits and preparedness as well as risk. And ask what mechanisms will keep the work transparent and accountable.

The challenge is to make AI serve a future we have chosen, rather than letting the future emerge by default. Prosper, prepare, protect: three connected responsibilities, and a useful compass for finding our way.

More reading:

• Lead like a wayfinder in an AI-enabled future – advice and guidance for school leaders
• Signal, noise, and the children in our classrooms: Part 1 – Reading the headlines
• Signal, noise, and the children in our classrooms: Part 2 – Watch what governments do
• Signal, noise, and the children in our classrooms: Part 3 – It isn’t about us
• A new direction for students in an AI world: Prosper, prepare, protect (Brookings research)
• Who gets a say in the AI conversation?