I use AI every day. It's fun, it's fast and it's useful. I'd be lying if I said otherwise, and a fair slice of my work now involves helping businesses put it to use.

But there's a question I can't shake. If AI does the work people used to be paid for, who is left with the money to buy what all these companies are selling?

AI is replacing skills, and skills are how people get paid

Strip away the marketing and AI is doing one thing. It takes a skill someone spent years learning, such as writing, coding, bookkeeping or design, and makes it available to anyone for a few dollars a month.

That's wonderful if you're the buyer. It's a different story if that skill was your income.

The people building the technology are saying this out loud. In May 2025 the chief executive of Anthropic, Dario Amodei, told Axios that AI could wipe out half of entry-level white-collar jobs and push unemployment in the United States to between 10 and 20 per cent within one to five years. That is a forecast, not a measurement, and plenty of economists think he's wrong. But it came from someone who sells the product.

An economy needs customers with pay packets

Here's the part the spreadsheets miss. Wages aren't only a cost. They're also the money that comes back through the front door.

In Australia, household spending makes up about half of the whole economy, according to the national accounts. Every business sits somewhere in that chain. So do the AI companies.

Amazon, Microsoft, Alphabet and Meta are on track to spend more than US$700 billion between them this year, most of it on AI infrastructure, by Forbes' count in October 2026. That bet only pays off if businesses keep buying AI. Businesses only keep buying if they have customers. And customers need an income.

There's an old story about Henry Ford. In January 1914 he doubled his factory pay to five dollars a day, and the legend says he did it so his workers could afford the cars they built. Historians at The Henry Ford museum will tell you the real reason was to stop staff quitting. The lesson stuck anyway. Workers and customers are the same people.

People won't cheer for the thing that sacks them

People aren't silly. They can see where this is heading, and they won't promote a product that puts them out of work.

When Pew Research surveyed Americans for a report published in April 2025, 64 per cent of adults expected AI to lead to fewer jobs over the next 20 years. Only 5 per cent expected more.

of US adults expected AI to mean fewer jobs over 20 years (Pew, April 2025)

64%

US unemployment within five years, as forecast by Anthropic's chief executive (May 2025)

10–20%

planned 2026 spending by four big tech firms, mostly on AI (Forbes, October 2026)

US$700b+

That mood matters commercially. Klarna, the Swedish payments company, said in February 2024 that its AI assistant was doing the work of 700 customer service agents. By May 2025 its chief executive was telling Bloomberg the cost-cutting had gone too far and service quality had suffered, and the company began hiring people again. Klarna says it's still committed to AI. It just found out customers want a person at the other end.

I've made that argument before, in Software is cheap now. Your customer service is the product.

Productivity for whom?

We keep hearing AI makes us more efficient and more productive. Fine. But what do those words mean when the value comes from a machine and not a person?

If a team of ten becomes a team of two doing the same work, the company's productivity figure looks terrific. The other eight people don't appear in that figure. Their mortgage does, somewhere, and so does the spending they've stopped doing at the local shops.

To be fair, there's a strong counter-argument. New technology has always created jobs nobody predicted. The World Economic Forum's Future of Jobs Report 2025 projects 170 million new jobs worldwide by 2030 against 92 million displaced. I hope that's right. But that's a projection too, and it assumes the people who lose the old jobs are the ones who get the new ones. That part has never happened by itself.

Listed companies can't stop, even if they want to

So why would any company cut the very people who keep the economy turning? Because for a listed company, the output is profit. Not products, not jobs. Profit, and more of it every year.

Milton Friedman put the case in The New York Times in 1970: the social responsibility of a business is to increase its profits. Half a century later that's still the rule the share market enforces. A flat year is treated as a failure. If AI can remove a wage bill, a listed company's board has very little room to say no.

Each company doing that is being rational. All of them doing it at once is how you run out of customers.

Now take away the share price. If your company isn't listed, does it really matter whether it grows? Picture a business that makes the same $10 billion every year for the next ten years. On the stock exchange that's a disaster. In real life it's ten years of wages paid, customers served and owners doing very nicely.

Private owners get to choose. In September 2022 the founder of Patagonia handed the company to a trust and a non-profit so its profits would go to protecting the environment. You don't have to go that far. The point is that an owner who answers to nobody's quarterly forecast can decide that enough is enough. It's one reason I think the future belongs to smaller, owner-run companies.

Time to change what we call success

I'm not asking anyone to stop using AI. I won't, and neither will your competitors.

I'm asking for a different scoreboard. Success should mean a business that lasts, and one that leaves its people, its customers and its community better off. A contented society is a better market than an anxious one.

If you run a business, here's where I'd start:

• Use AI to grow what your people can do, not to shrink the payroll. Take the time it saves and spend it on customers.
• Count more than profit. Track staff kept, customers kept and how quickly a real person answers.
• Decide what "enough" looks like. If you own the company, you set the target. It doesn't have to be "more than last year".
• Keep humans where trust is made. Sales, support and anything that goes wrong.