If AI does the work, who gets paid?

There is a question missing from much of the conversation about artificial intelligence.

We hear about productivity, efficiency and the opportunity for businesses to achieve more with fewer people. We hear that workers need to adapt, learn new skills and prepare for a changing world.

But follow that thinking a little further.

If businesses need fewer workers, and fewer people receive wages, who buys what those businesses produce?

An automated business still needs customers. And customers need some way to pay.

That is where the conversation about AI becomes a conversation about how we organise our economy, who owns its productive assets, and how people get a share of the wealth they help make possible.

An economy needs more than production

Imagine a business that automates a large proportion of its work. It can produce the same goods or services with fewer employees, potentially reducing its costs and increasing its profits.

For that individual business, the decision might make financial sense.

Now imagine similar decisions happening across thousands of businesses.

People who lose income cut their spending. Other businesses lose customers. Suppliers receive fewer orders. Governments face greater demand for support while potentially collecting less tax from wages.

A decision that benefits one company can create problems when repeated across the wider system.

This does not mean automation inevitably causes economic collapse. Lower prices, investment and new jobs can help offset disruption. But those benefits are neither automatic nor guaranteed to reach the people whose income has disappeared.

The ability to produce more does not, by itself, give people the ability to buy more.

Does everyone actually become unemployed?

We should be careful here. A future in which nobody has a job is a thought experiment, rather than an established forecast.

The International Labour Organisation’s 2025 assessment found that around one in four workers globally were in occupations with some exposure to generative AI. Its assessment was that changes to jobs were more likely than complete replacement. Exposure is not a prediction of redundancy.

Still, we do not need everyone to become unemployed for the consequences to matter.

Fewer entry-level opportunities, reduced hours, insecure contracts or downward pressure on wages could leave people struggling even while they remain technically employed.

And retraining, useful as it is, cannot answer every question. It helps people compete for available work. It cannot guarantee that enough secure, adequately paid work will exist.

So it is worth asking what happens if paid employment becomes a less reliable way of distributing income.

Could an economy work without jobs?

In principle, yes — if it could still produce what people need and give people access to it.

An economy involves producing, distributing and using goods and services. Paid employment is one way we organise those activities and distribute purchasing power. People also receive income through pensions, benefits, investments and business ownership.

In a highly automated future, people could receive income through broader ownership of productive assets, public payments, or a combination of the two. Essential services could also be provided directly, reducing how much money households need.

People would continue caring, learning, creating and contributing to their communities, regardless of whether those activities came with a salary.

But AI software alone cannot build a world without human labour. Food, housing, energy, infrastructure and care all involve physical resources and practical constraints. A genuinely jobless economy would require far more extensive automation than the tools we use today.

The useful question is how we would share the benefits as less human labour became necessary.

Income would need another route to households

One option is a universal basic income: a regular payment that does not depend on having a job.

It could provide security and maintain people’s ability to spend when wages are scarce. But saying “introduce a basic income” leaves substantial questions unanswered.

How much would people receive? How would it be funded? Would disability support and other additional assistance remain? Would payments keep pace with living costs?

There is also the question of supply. Giving people more money does not automatically create more homes, electricity or healthcare. Where supply is constrained, additional spending can push up prices.

A credible approach would need to consider income alongside the availability and affordability of essentials.

The International Monetary Fund has argued that sharing AI’s gains will require stronger social protection and changes to taxation, including attention to income from capital as the economy changes.

That brings us to ownership.

Who owns the machines matters

If a small group owns the technology doing an increasing share of the work, that group is positioned to receive much of the financial return.

Everyone else could become increasingly dependent on public support, insecure employment, or the spending decisions of wealthy asset owners.

We could choose to broaden ownership.

Employee ownership, cooperatives and public investment funds are possible approaches. A public fund could hold productive assets and use some of its returns to support services or distribute a dividend to residents.

These models involve trade-offs. Investments carry risk. Funds need accountable governance. Communities need a meaningful say in decisions made on their behalf.

But they ask a necessary question: should people share in technological progress only when someone employs them, or should they also have an ownership stake in its success?

For Aotearoa, that conversation should include Māori ownership and tino rangatiratanga, alongside the ability of Pasifika communities and others facing barriers to investment to participate.

Access to an AI tool is useful. A lasting share in the value it creates is a different kind of opportunity.

We could also share the time saved

Between today’s economy and a hypothetical future without jobs lies another possibility: less paid work per person.

If technology enables a team to produce the same output in fewer hours, some of that gain could support shorter working weeks while protecting incomes.

That would require negotiation, evidence and arrangements suited to different industries. Productivity gains will vary, and some organisations will have more room to reduce hours than others.

Still, it deserves a place in the conversation.

Time saved could mean more rest, more care for whānau, more community participation and more space to live. Those are meaningful returns on technological progress.

The transition needs attention now

Even a workable future model would not make the journey painless.

Rent and mortgage payments continue while people retrain. Families cannot put groceries on hold while policymakers debate a new tax system.

Practical preparation could include stronger income protection, paid opportunities to learn, worker participation in automation decisions, and public investment in essential services.

Governments would also need to examine where revenue comes from. If more income flows to owners and less to workers, tax arrangements must be able to support public needs under those conditions.

For a small country, there is an added challenge when the companies capturing the greatest returns are based overseas. An economy can adopt powerful technology without retaining a proportionate share of its profits.

That makes local capability, ownership and international cooperation part of the same discussion.

What should greater productivity give us?

At Paadia, I want us to be confident enough with AI to ask questions beyond what a tool can do.

Who benefits from its use? Who carries the disruption? Who gets to influence the decisions?

If technology eventually allows us to meet more human needs with less human labour, that could be an extraordinary achievement.

Whether it gives us greater security and freedom will depend on the systems surrounding it: ownership, taxation, public services, worker voice and democratic accountability.

We can have an economy with much less paid employment. Making it work would require reliable ways for people to share in what that economy produces.

If AI does more of the work, people should be able to share in both the wealth created and the time released.

That is a future worth organising for.