Artificial intelligence and its ability to perform intellectual tasks previously the preserve of humans are improving at a stunning rate, and yet we are complacent or politically paralysed in the face of this transformation.

The likelihood of a massive, maybe permanent, disruption to the job market, and the social and political chaos that would likely follow, is such that all countries need to start considering options for new economic models, social contracts and valuations of human activity.

Artificial general intelligence, which most experts define as being better than humans at any economically useful task, is plausibly just a few years away. Anthropic chief executive Dario Amodei, who is the most forthright of the frontier AI leaders, warned in May that AI could wipe out half of entry-level white-collar jobs and push unemployment to 10–20 percent within five years. Amodei urged the AI industry and governments to stop ‘sugar-coating’ the risks.

Optimists, including governments desperate to reassure increasingly restive electorates, argue that AI will complement and augment human work. New products and services, they say, will find markets created by limitless human demand. New jobs will be created; productivity and prosperity will soar.

The flaw in this argument, which advocates invariably skip over, is that there is no reason why these new jobs must be filled by humans. If AI gets better than us at every present and future task, why would human labour be competitive, outside of a few specific exceptions?

Optimists invariably point to historical precedents. Steam power, electrification, mechanisation and factory automation each created their own episodes of panic. But after periods of disruption and change as production adjusted to the innovations, the jobs that disappeared were more than made up for by new demands and new jobs, driving economies to greater productivity and growth.

But these are poor comparisons. Previous innovations replaced our muscle power and our organised manual labour. None came close to automating the work done by our hitherto peerless brains. Whatever our appetite for new products and services, AI tools that can work round the clock to produce anything better than we can must surely push the value of our labour down towards zero. This is a categorically different technology.

Granted, thoughtful counterarguments exist. Oxford economist Daniel Susskind argues we might value certain products precisely because they were made by humans, for instance if they embody an authenticity that automated production processes lack. However, he acknowledges it is uncertain how many decent paying jobs this type of work will create. Competition between humans in sport or games will still be valued, but by definition rewards will go only to the elite.

Other jobs, such as politicians, judges and military officers, might be deemed morally necessary to be filled by humans. But it would be an odd society that only works creating, interpreting and protecting laws for one another. Caring professions, the value of which arguably comes from the belief that the carer has empathy, might provide endless work. But even this is uncertain. The boom in AI counselling already suggests it’s the outcome rather than the process that matters.

Susskind and others have also pointed to the economic theory of comparative advantage. They argue that human labour, rather than sitting inefficiently idle, could relieve AI of tasks that would pose an opportunity cost on its time—the same reason two countries both benefit from trade even if one has an absolute advantage in making things both countries need. Yet powerful AI isn’t like a trading nation that has limited industrial capacity. Computing capacity is expanding by orders of magnitude as trillions of dollars flow into data-centre construction globally, with forecasts they will spread to unpopulated areas.

So where does this leave us? If we manage it well, AI’s ability to provide us with all goods and services—including the ones we haven’t thought of yet— ought to bring a permanent golden age as we are freed to pursue meaning for our lives in other ways.

It will be a tragedy if we bungle this opportunity.

But to fashion this future, nations individually and collectively need to rethink the political and economic models that have existed since humans have been around. A universal basic income will only be the beginning. Governments will need to find ways to ensure that the fruits of production don’t cluster around trillionaire owners of the technology. Each society will need to forge a consensus that a person’s worth is not measured by their economic output. And we need to overhaul education from early childhood to prepare future generations to find meaning in pursuits other than paid work.

Decoupling the output of our waking hours from our place in society and our sense of self-worth is a social experiment like none we’ve ever tried. It could be the dawning of a new age in which we permanently transform what it means to be human and to be a valued member of a society. We can find entirely new things to do. The possibilities are beguiling—they just won’t fit the economic and social models that have existed throughout history.

At our present pace, we’re headed for a chaotic disruption that could bring dystopia. But we’ve got at least a shot at utopia if we start preparing for this metamorphosis with brave, honest political conversations.