AI will transform employment, not eliminate work

Listening to the current debate about artificial intelligence, it is hard to escape the impression that soon we will all have plenty of time to sit in deck chairs looking worried while a server farm somewhere in Malaysia takes over the rest of civilization's duties.

Despite the prevailing sense of doom, however, there are exceptionally good reasons to believe that AI will not create a chronic employment problem.

Do not misunderstand me: the transition is unlikely to be a leisurely Sunday outing with coffee and pastries. It will be turbulent, and many of us will end up doing something entirely different from what we expected. But we will not run out of economically valuable things to do. Humanity is notoriously bad at leaning back and declaring itself satisfied. In the longer run, most of us will still be working — just on new, stranger and probably much more enjoyable things.

History offers an excellent antidote to anxiety. We have spent centuries inventing machines whose sole purpose was to make us redundant. The steam engine displaced much of our muscle power. The tractor sent farmworkers packing. The assembly line took over craft production, and the computer assumed the office work that people had spent generations turning into a virtue.

Yet human labor proved surprisingly difficult to kill off.

Perhaps the world's most stubborn evidence comes from Britain, where economic historians have patiently reconstructed national income all the way back to 1270. What is fascinating is not that the figures fluctuate — they do, through pandemics, civil wars, the collapse of empires and industrialization. It is how consistently human labor has held its ground.

Depending on which historical model you favor, labor's share of the economic pie remained largely between 50 and 68 percent over six centuries. In the 1950s, Britain's labor share edged above 70 percent. Despite microchips, globalization and the internet, it has hovered around 58-60 percent since 2000.

Consider the inventory: steam power, railways, gas lighting, electricity, internal combustion engines, assembly lines, computers and fiber-optic cables. Every wave eliminated human tasks on an industrial scale. Yet people refused to disappear as a factor of production. We simply found something new to do. Again and again.

This brings us to a classic paradox: why did electricity or computers not end up accounting for 90 or 99 percent of the world economy, given that they revolutionized practically everything?

Because successful technology has an unavoidable habit: it makes whatever it produces dramatically cheaper, often almost free. Electricity did not conquer the world by swallowing half of global GDP. It did the opposite: it became so astonishingly cheap and ubiquitous that it enabled thousands of other industries. The value accrued beyond the wires.

AI looks set to become the most extreme example of this mechanism we have yet seen. Global production of AI tokens is currently growing roughly fivefold a year. Meanwhile, the unit cost of generating a comparable response is falling by around 80 percent annually. During a transitional phase, total AI spending can naturally grow explosively, simply because demand is surging and cathedrals of silicon and cooling equipment must be built. But the gravitational pull of abundance always wins. When machine intelligence becomes an endlessly available, inexpensive commodity, such as running water, the AI sector itself will not swallow the entire economy. Its main function will be to make the rest of the world vastly bigger, faster and more varied.

What, then, will we do on Tuesday mornings when computers think faster than we do?

For a start, agentic AI gives us the world's largest creative sandbox. Where launching a project once required a board of directors, three department heads and half an accounting firm, a single person can now sit at a kitchen table with a handful of specialized AI agents that write code, design products, analyze markets and respond to customers. The walls of bureaucracy are crumbling, and the power to create is being democratized.

But few people dream of running 15 virtual software companies at once. Alongside the machine economy, we will therefore see the growth of what I call the passion economy. The phrase has two sides. One is the desire to create: to write books, obsess over sourdough, compose music, design clothes, build quirky garden benches or organize concerts. As machines take on the heavy lifting of keeping our infrastructure running, they free up time and resources for the things we love.

The other side is care. We care about our children, older people, one another and the natural world. A robot can undoubtedly be programmed to turn a bedridden patient or deliver medication reminders with mechanical precision. But it cannot provide genuine human presence. Human companionship does not become worthless because robots exist. On the contrary, it becomes a luxury precisely because it cannot be mass-produced on a server.

That leaves us with a liberating conclusion: the more abundant machine intelligence becomes, the more valuable the things that remain scarce — relationships, authenticity, humor, trust, empathy and presence. Machines can handle the calculations; we will take care of the rest.

Do the experts agree?

When you arrive at a strong thesis, it is always healthy to check whether the people who make a living analyzing these questions disagree completely.

Look at long-term analyses from institutions such as McKinsey, Goldman Sachs, Morgan Stanley, Gartner and Boston Consulting Group, and the conclusions are remarkably consistent. They expect a substantial increase in productivity, but they do not predict permanent mass unemployment.

McKinsey does estimate that hundreds of millions of tasks will be automated, but its models almost always end with economies returning to full or near-full employment. Goldman Sachs expects the effect on net unemployment to be relatively modest and transitional, while Gartner explicitly predicts that, over time, AI will create more jobs than it eliminates.

The next 10-15 years will undoubtedly be a roller-coaster ride. Occupations will disappear overnight, people will have to learn new skills halfway through their careers, and lawmakers will trail reality by at least five steps.

But the disappearance of particular jobs does not mean the disappearance of work.

We are not looking at a future in which people are condemned to passivity. We are looking at a world with plenty of work, but in which we do something entirely different. AI will eliminate an enormous number of the tasks filling our calendars today — but it will not eliminate our insatiable urge to invent new things to do.

And thank goodness for that.

The author is a Danish entrepreneur, investor, and bestselling author known for his work on economics, technology, and global trends. He is the co-founder of Atlas Global Macro and the founder of Supertrends, a company specializing in future trends forecasting. He has written numerous books exploring financial markets, innovation, and the future of society, including Business Cycles and 2050: When Machines Do the Work and Humans Do the Living, among others.

The views don't necessarily reflect those of China Daily.

If you have a specific expertise, or would like to share your thought about our stories, then send us your writings at opinion@chinadaily.com.cn, and comment@chinadaily.com.cn.