It’s 1920s New Zealand and the nightman is about to do his rounds, emptying backyard dunnies across the neighbourhood. It’s honest work. It pays the bills. It feeds his family. Then indoor plumbing arrives, and within a generation, his entire profession vanishes.
Nobody mourns the nightman’s job today. We don’t lie awake worrying about all those unemployed dunny emptiers. Progress happened, people adapted, and life got better for everyone—including the nightman’s grandkids, who certainly aren’t queuing up to empty outdoor toilets for a living.
Yet here we are in 2025, having the exact same panic about artificial intelligence that every generation has had about every transformative technology. The headlines scream about job losses. And somewhere, the ghost of that nightman is having a quiet chuckle.
Let’s start with what’s actually happening. The St. Louis Federal Reserve found that in the United States, computer and mathematical occupations—predictably among the most AI-exposed—saw some of the steepest unemployment rises between 2022 and 2025.
Young American workers are getting hammered. Goldman Sachs reports that unemployment among 20- to 30-year-olds in tech-exposed occupations has risen by almost 3 percentage points since early 2025.
The same research tracking job losses is also tracking job creation. World Economic Forum data shows that globally, while 85 million jobs will be displaced by 2025, 97 million new roles will emerge. That’s a net gain of 12 million jobs worldwide. Not a catastrophe. Not even close.
We’ve been here before. Every single time.
When ATMs were introduced in the 1970s, everyone assumed bank tellers were doomed. Economist James Bessen’s research on the United States shows what actually happened: the number of tellers required per branch fell from 20 to 13 between 1988 and 2004. But banks responded by opening 43% more branches, and teller jobs actually increased. The role changed—they became relationship bankers instead of cash handlers—but the jobs didn’t disappear.
The same thing happened in the 19th century. Power looms automated 98% of the labour needed to weave a yard of cloth. The number of factory weaving jobs increased anyway, because lower prices meant sharply increased demand, and weavers’ remaining skills became more valuable.
Elevator operators, telephone switchboard operators, human computers doing calculations by hand—all gone. Yet we have more phones, more elevators, more calculations, and massively more total jobs than we did when those roles existed. This isn’t an accident. It’s how technology works.
Here’s something most commentary misses: New Zealand might actually be better positioned than most countries to weather this transition.
University of Auckland research points out that roughly 70% of our exports come from agriculture, horticulture, seafood, and forestry. Domestically, leading employment sectors include aged care, physiotherapy, plumbing, and early childhood education. These roles require physical dexterity, sensory judgement, and human empathy—precisely the skills AI cannot yet credibly replicate.
While advanced economies over-invested in finance, bureaucracy, and what some academics politely call “bullshit jobs,” New Zealand’s focus on tangible, value-producing work could be a strategic strength. AI can write policy briefs and generate marketing copy. It cannot change a tyre, fix a blocked drain, care for an elderly patient, or teach a four-year-old.
The AI Forum of New Zealand’s report found that 82% of Kiwi organisations are already using AI—higher than the global average of 78%. Yet just 7% report AI replacing workers. What they do report is AI inhibiting new hires, which is a different problem with different solutions.
Let’s not sugar-coat this. The transition period is brutal for those caught in it.
McKinsey’s analysis shows that during the UK’s first Industrial Revolution, productivity continued to grow while real wages remained flat for about 40 years. An entire generation of workers saw no benefit from the technological revolution happening around them. That’s the uncomfortable truth about progress—the gains aren’t distributed evenly or immediately.
Entry-level positions are getting hammered right now, particularly in tech sectors across developed economies. But here’s what the panic merchants won’t tell you: a Brookings Institution study from October 2025 found no macro-level pattern of increasing AI exposure among the unemployed in the United States. Even transformative technologies like computers and the internet took decades for their impacts to fully materialise in the workplace.
In 2020, “prompt engineer” wasn’t a job title. It wasn’t even a concept. Today, companies in the United States are offering between US$90,000 and US$335,000 annually for people who can effectively communicate with AI systems.
According to US workforce analytics, AI-related roles now account for 9.5% of all new software job listings, with demand continuing to grow. These include AI ethics leads, trust and safety specialists, data analysts who interpret AI outputs, and MLOps engineers who deploy and maintain machine learning systems.
The PwC 2025 Global AI Jobs Barometer found something fascinating: globally, workers in the same job with AI skills now command a 43% wage premium compared to those without. That’s up from 25% just last year. AI isn’t just destroying jobs—it’s creating a new class of highly-paid workers who know how to work alongside it.
The real challenge isn’t job destruction. It’s the skills gap.
The New Zealand Treasury’s analysis warns that our traditionally slow diffusion of new technology and low levels of investment in intangible capital could be a barrier to realising AI’s benefits. We need up to 161,000 new AI workers by 2030, according to CSIRO estimates.
That’s a training challenge, not an unemployment crisis. It’s the same challenge we faced when tractors replaced horses, when computers replaced typewriters, when the internet replaced… well, everything. We adapted then. We’ll adapt now.
The transition period will be messy. Some people will struggle. Some communities will hurt. That’s not acceptable, and it requires real policy responses—better retraining programmes, stronger social safety nets, and genuine support for workers transitioning to new roles. But it’s not the end of work.
Nobody’s coming back from the future to warn us that AI destroyed civilisation and left humanity unemployed. You know why? Because it didn’t happen. It won’t happen. It never happens.
Every generation thinks its technology is different. Special. Uniquely threatening. The Luddites smashed weaving machines. People panicked about ATMs. Experts warned that computers would create mass unemployment. They were all wrong, and the AI doomers are wrong too.
The nightman lost his job. His kids found better ones. Their kids found even better ones. And somewhere along the way, we got indoor plumbing, which beats the hell out of an outdoor dunny.
That’s not blind optimism. It’s pattern recognition. Progress is painful, uneven, and often unfair. But it’s also inevitable, and historically beneficial. The question isn’t whether AI will destroy jobs—it won’t, not on net. The question is whether we’ll support people through the transition, invest in the training they need, and ensure the gains are shared broadly rather than concentrated at the top.
We’ve done this before. We’ll do it again. And fifty years from now, our grandkids won’t mourn the loss of data entry jobs any more than we mourn the nightman’s cart.
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
Peter Hewson says:
Here’s hoping