Last November, I wrote about the nightman — that poor soul trudging around 1920s New Zealand emptying backyard dunnies, whose entire profession was wiped out by indoor plumbing within a generation. The point was simple: nobody mourns their job today. Progress happened, people adapted, and life got better for everyone, including the nightman’s grandchildren. The AI panic, I argued, was the same old story wearing different clothes.
I stand by that argument. But this week gave it a sharp new test — and two developments have surfaced that deserve more than a shrug and a historical footnote.
On Monday, Pope Leo XIV stood before the world’s cameras at the Vatican and declared artificial intelligence the defining moral challenge of our era. At his side was Christopher Olah, co-founder of Anthropic, the company that makes the AI tools many of us use daily. Their jointly released document, Magnifica Humanitas, calls for robust regulation of AI, limits on its military applications, and the protection of human dignity amid the digital revolution.
Meanwhile, 18,000 kilometres away in Wellington, Finance Minister Nicola Willis was doing something rather more concrete. She announced that nearly 9,000 public servants would lose their jobs, that AI would be embedded “as a basic expectation for all public entities,” and that the whole exercise would save New Zealand $2.4 billion. Budget week, it turns out, is also AI week — whether the Pope likes it or not.
The nightman’s ghost is still chuckling. But this week, even he would have raised an eyebrow.
Let’s not be naive about Monday’s Vatican ceremony. Anthropic is currently in dispute with the Trump administration and has resisted Pentagon pressure to weaponise its technology. The company has an obvious strategic interest in cultivating powerful moral voices to provide cover for that resistance. What better ally than the Pope? The Vatican gets to look relevant and engaged with contemporary issues. Anthropic gets a moral imprimatur that money cannot buy.
Olah’s remarks were, to his credit, remarkably candid. He admitted openly that every frontier AI lab — including his own — operates within a web of commercial and geopolitical pressures, and what he called “the older, plainer pressures of pride and ambition,” which can conflict with doing the right thing. He said those outside that system — religious communities, civil society, governments — need to act as a check on the industry.
“We need moral voices that the incentives cannot bend,” Olah told the gathering.
Fine sentiment. It would carry more weight if Anthropic had sought that external moral check before it needed one against a hostile administration. Candour and self-interest are not mutually exclusive — he can be both sincere and strategically positioned. He is almost certainly both.
As for the Pope, neither he nor the institution he leads was elected by anyone outside the Catholic Church. Their bilateral arrangement about humanity’s AI future is not a substitute for genuine democratic deliberation. It is two unelected institutions agreeing that the rest of us need managing. New Zealanders might reasonably ask where their voice is in that conversation.
This is where my nightman argument gets its real stress test this week, and where I want to be honest rather than reflexively dismissive.
Willis’s case has genuine merit in places. New Zealand currently operates 39 government departments and ministries, compared with 16 in Australia, 24 in the United Kingdom, and around 12 in Finland. That is a bloated administrative structure by any international measure, and the argument that digitisation and AI can deliver equivalent services with fewer people is not implausible on its face. Some of those 8,700 roles will be genuine redundancies in a properly streamlined system.
But “AI will do the work” is not a plan — it is a slogan (Her slogans really grate with me!). And Wellington’s own mayor — and former Labour Party leader — Andrew Little said publicly that he was taken completely by surprise by the announcement and that the Government’s AI references lacked detail. As he put it: “It’s not clear what ministers mean when they say they want to use more AI in government, because that on its own is a pretty meaningless statement.”
He is right. Which AI systems? Built by whom, at what cost, with what accountability, when they get it wrong? Who handles the complex cases that no algorithm handles well — the grieving widow navigating WINZ, the small business owner fighting an IRD assessment, the family dealing with a Corrections system that just made an error? What happens to the institutional knowledge that walks out the door with 8,700 people and cannot be recovered once it’s gone?
These are not minor implementation details. They are the difference between a genuine productivity transformation and a politically convenient way to cut costs while leaving the actual work undone — or done badly by inadequately supervised software. Government technology projects in New Zealand have a long and distinguished history of costing more than projected and delivering less than promised. The optimism embedded in that $2.4 billion figure deserves scrutiny that Budget week’s political theatre makes difficult.
Here is something my November article didn’t fully grapple with, and I’ll acknowledge it plainly now.
The jobs most vulnerable to AI — in both the public and private sectors — are entry-level positions. Predictable, repetitive, knowledge-based tasks are exactly what current AI handles well. Those are also precisely the jobs that young people and career-changers have historically used as the first rungs on the employment ladder. You become a competent policy analyst by first spending years doing the grunt work of research, summarising, and drafting. You become a skilled accountant by first grinding through the routine reconciliations. You build professional judgement by doing the unglamorous work first.
The nightmen found new work — but they had time, and it still had entry points. If AI removes those lower rungs before the next generation has climbed them, we don’t just have a short-term unemployment problem. We have a structural problem with how people develop skills and build careers at all. The pipeline itself breaks. New Zealand research already shows entry-level hiring slowing sharply as AI handles what graduates used to cut their teeth on.
That is a new wrinkle the historical analogy doesn’t fully address, and it deserves a serious policy response rather than a Vatican encyclical or a pre-Budget speech.
There is a second issue my original Nightman piece glossed over, and the Willis announcement brings it into sharp relief.
Past technology waves created new jobs broadly across economies and nations. The Industrial Revolution built factories in Manchester and Christchurch alike. The internet created tech industries in dozens of countries. AI’s economic gains are concentrating in a handful of American corporations. New Zealand is not going to produce the next Anthropic or OpenAI. We will absorb the disruption — the job losses, the retraining costs, the social adjustment — and export the profits to San Francisco. The $2.4 billion Willis hopes to save will largely be reinvested in AI tools built overseas, running on overseas servers, owned by overseas shareholders.
That is a distributional problem of real political consequence, and it is precisely the kind of problem that has historically required active government policy to address — not faith in the market to sort it out eventually.
The nightman argument still holds on to the fundamentals. History is right. Technology creates more than it destroys. The doom-merchants will be wrong again on the net outcome. I am not abandoning that position because the current panic is louder than previous versions.
But “history says relax” is not sufficient political analysis in Budget week, when a government is using AI as justification for the largest public service restructuring in a generation, with no detailed implementation plan, no serious discussion of the entry-level career pipeline, and no policy framework for ensuring New Zealand captures any of the value it is being asked to help generate.
The Pope is half right. Willis is half right. The Anthropic co-founder is shrewd. And the ghost of the nightman, watching all of this from wherever nightmen go, is probably thinking that at least his bosses had the decency to tell him directly that his job was gone — rather than announcing it in a speech at an Auckland business lunch.
That is the real budget deficit worth worrying about.
Steve Baron is a New Zealand-based political commentator and author. He holds a BA with a double major in Economics and Political Science from the University of Waikato and an Honours Degree in Political Science from Victoria University of Wellington. A former businessman in the advertising industry, he founded the political lobby group Better Democracy NZ. https://stevebaron.co.nz