A guest post by Chris Scott on Kiwiblog:
Every so often, New Zealand produces a piece of public policy that doesn’t really belong to the left or the right — it simply works. ACC is the classic example. When it arrived in the 1970s, it wasn’t universally adored, but it solved a real problem in a way both sides could live with. The left valued the universality and fairness; the right appreciated the end of endless litigation and the stability it brought to business.
It was, in its own way, a kind of political magic pill: a rare moment where the country managed to design something that delivered outcomes both sides could accept, even if for different reasons.
But those moments are rare. Not because politicians don’t want them, but because formulating a political magic pill is hard. It requires a system capable of producing outcomes that are more than just tolerable to both sides — outcomes that are genuinely better, cheaper, faster, and fairer all at once. That’s a tall order when the machinery you’re working with is creaking along on 1970s logic.
Our public sector isn’t bloated because it’s generous. It’s bloated because it’s inefficient. Every agency has its own systems, its own databases, its own processes. The only way to keep that fragmented machine running has been to keep adding more people, more layers, more checks. We’ve ended up with a government that’s large because it’s slow, not because it’s ambitious.
This is where AI enters the picture — not as a replacement for human judgement, but as a tool for finally upgrading the operating system the state runs on. Most of what clogs up government isn’t meaningful work; it’s friction. It’s re‑entering the same information into multiple systems. It’s reconciling mismatched data. It’s waiting for approvals that only exist because the underlying machinery can’t coordinate itself.
Clear that friction, and the whole structure becomes lighter. Not because you’ve cut it back, but because it no longer needs to be so heavy.
And once you start imagining a state that actually works properly, you realise something interesting: the old political trade‑offs begin to dissolve. Lower taxes versus better services. Faster decisions versus fairness. A leaner state versus a more capable one. These aren’t iron laws of politics; they’re symptoms of an outdated operating system.
A modern, intelligent one gives you the ingredients to formulate a new magic pill — one that could deliver:
But those moments are rare. Not because politicians don’t want them, but because formulating a political magic pill is hard. It requires a system capable of producing outcomes that are more than just tolerable to both sides — outcomes that are genuinely better, cheaper, faster, and fairer all at once. That’s a tall order when the machinery you’re working with is creaking along on 1970s logic.
Our public sector isn’t bloated because it’s generous. It’s bloated because it’s inefficient. Every agency has its own systems, its own databases, its own processes. The only way to keep that fragmented machine running has been to keep adding more people, more layers, more checks. We’ve ended up with a government that’s large because it’s slow, not because it’s ambitious.
This is where AI enters the picture — not as a replacement for human judgement, but as a tool for finally upgrading the operating system the state runs on. Most of what clogs up government isn’t meaningful work; it’s friction. It’s re‑entering the same information into multiple systems. It’s reconciling mismatched data. It’s waiting for approvals that only exist because the underlying machinery can’t coordinate itself.
Clear that friction, and the whole structure becomes lighter. Not because you’ve cut it back, but because it no longer needs to be so heavy.
And once you start imagining a state that actually works properly, you realise something interesting: the old political trade‑offs begin to dissolve. Lower taxes versus better services. Faster decisions versus fairness. A leaner state versus a more capable one. These aren’t iron laws of politics; they’re symptoms of an outdated operating system.
A modern, intelligent one gives you the ingredients to formulate a new magic pill — one that could deliver:
● lower taxes
● smarter, leaner regulation
● better public services
● a more adaptive education system
Not because of ideology, but because the machinery finally supports the outcomes we’ve been arguing about for decades.
Of course, the moment you say any of this, people worry about jobs. And fair enough — that fear sits under everything. But the future isn’t a mass firing. It’s a gradual shift. Most reductions happen through natural attrition, as they always have. And the work that remains becomes more human, not less. AI is good at the routine; it struggles with judgement, nuance, and the messy edge cases where real people live.
The future of work looks more like shorter weeks, more flexibility, and jobs that focus on people rather than paperwork — supported by “pocket mentors” that help workers learn, plan, and navigate their careers. This isn’t something to fear. It’s something to prepare for.
And preparation means getting specific. It’s easy to sketch broad ideas about a smarter state, but at some point you have to put a real proposal on the table. So here’s one worth considering.
Right now, we use 13% of our renewable electricity to produce aluminium — a low‑margin commodity in a world where the highest‑value product is compute, and the intelligence that runs on top of it. If we’re serious about building a modern operating system for government, we should also be thinking about the infrastructure that supports it.
Tiwai Point — with its enormous, steady supply of renewable power — is a natural candidate for a national‑scale data and compute hub. Not a government‑run monolith, but a public–private partnership with a hyperscaler like Microsoft, AWS or Google. They bring the capital and expertise; we bring the clean energy and the strategic intent.
A project like that would:
● turn low‑value electrons into high‑value compute
● anchor a sovereign capability we currently lack
● support the AI systems needed for a smarter state
● and lay the groundwork for a new kind of economic engine
This is what I mean by formulating a political magic pill. Not slogans, not wishful thinking — but structural changes that deliver outcomes both sides of politics have wanted for decades: lower taxes, smarter regulation, better services, and an economy built on high‑value capability rather than low‑margin commodities.
And that’s really the point. ACC showed we can build systems that work across ideological lines. We did it once by accident. With the right tools, we might be able to do it deliberately.
And if we do — well, that opens the door to a much bigger conversation about what New Zealand could become when intelligence, not aluminium, becomes our next great export.
3 comments:
I admire your vision and positivity Chris.
But unfortunately it will remain unrealised.
The government can’t (or at least seems unwilling to) make best use of internet technologies which already exist. What makes you think they’re likely to be early adopters of AI?
We have all had to deal with government department websites which ask us to download, print, complete, and post (yes, SNAILMAIL) a form, when a web form could have been used.
And we have to use our own online banking to make payments to their account. What’s wrong with a credit card, a web form, and a payment gateway?
We have to phone (and wait ages for it to be answered) just to make a time to see someone. Why can’t they add appointment scheduler software to their websites?
Let’s face it, with budget cutbacks, technophobic and risk-averse public so-called “servants”, government agencies have become technical laggards when they could be leaders.
AI-driven efficiency? Yeah, nah. Never happen.
I'm not convinced. How about New Zealanders build a smaller, smarter state. What it takes is process development and quality control, and humans have done that in the past.
Copilot keeps reminding me that they are my assistant. I find copilot incredibly useful and I expect other AI could be even more so. But AI does not have human intuition.
It remains that the best language for programming AI is English. Expecting AI to fix our problems is avoidance of responsibility. Government could halve the size of the Public Service without AI, they just don't have the gumption and motive to do so. AI will not change that.
The first problem is turkeys don’t vote for Christmas. The bureaucrats don’t want to lose their cushy jobs.
The second problem is the crippling redundancy bill because the bureaucracy had employment conditions long since gone from the private sector including big fat redundancy pay outs.
The third problem is we have a 3 year parliamentary term and while the costs of change will hit within that term the benefits will not arrive within it.
The fourth problem is our politicians are overwhelmingly going to choose self interest over the interests of the people.
Our politicians will chose to ride this into the ground and will only make change when this is the least painful option for them.
Post a Comment
Thank you for joining the discussion. Breaking Views welcomes respectful contributions that enrich the debate. Please ensure your comments are not defamatory, derogatory or disruptive. We appreciate your cooperation.