Cook County, which encompasses Chicago and some five million people, did something rather unusual with its 2026 budget. It made permanent a programme that hands 3,200 households 500 US dollars a month, no strings attached. Across rural Kenya, some 23,000 adults have been receiving monthly cash transfers since 2018, part of a 12-year experiment run by GiveDirectly, an American charity. Wales paid former foster children 1,600 UK pounds a month. Catalonia handed 800 euros per month to 5,000 adults. Finland ran the world’s first randomised national trial in 2017–18, giving 2,000 unemployed people 560 euros a month.
These scattered experiments share a common ancestor: universal basic income (UBI), an idea that has ricocheted between utopian fantasy and serious policy proposal for half a century. Milton Friedman liked a version of it. The concept is simple: give every citizen a regular, unconditional cash payment, enough to cover basic needs. No means tests, no bureaucratic gatekeeping.
What has changed is the urgency. Artificial intelligence is eating white-collar work with a speed that makes previous waves of automation look positively genteel. The IMF reckons that 40 per cent of jobs worldwide face meaningful exposure to AI, rising to 60 per cent in advanced economies. Goldman Sachs estimates that generative AI could affect 300 million full-time jobs globally. The World Economic Forum projects 92 million roles displaced by 2030, offset (it insists cheerfully) by 170 million new ones. Whether those new jobs materialise, and who gets them, are questions policymakers have been conspicuously reluctant to answer.
Previous upheavals followed a reassuring pattern. Workers displaced from farms found factories, and those displaced from factories found offices. AI scrambles the logic. It targets the cognitive, not the menial (legal assistants, financial analysts, content writers, medical coders). An SHRM survey found that 15 per cent of American employment (roughly 23 million jobs) already involves the automation of at least half of all tasks. Geoffrey Hinton, the computer scientist credited as AI’s godfather, warns of “massive unemployment” and a world in which the rich get richer while most workers get poorer.
Pilots are not policy
Into this breach steps UBI, an looking less fantastical by the year. Finland’s experiment found a small positive effect on employment and a large one on wellbeing: recipients reported higher life satisfaction, lower stress and greater confidence than those in the control group. In Kenya, GiveDirectly’s mammoth trial recorded substantial economic expansion in recipient villages, such as more enterprises, higher revenues, and a structural shift away from subsistence farming. Labour supply did not fall. In Stockton, California, a two-year pilot giving 125 residents 500 US dollars a month saw recipients more likely to find full-time work than a control group. America’s largest study, of 1,000 low-income people receiving 1,000 a month across Texas and Illinois, found increased spending on food and housing, with employment dipping by a mere 1.3 hours a week.
Sceptics will note, fairly, that pilots are not policy. Evelyn Forget, a Canadian economist, has told Newsweek she doubts the statistical robustness exists to scale regional results to national predictions. A 2026 meta-analysis by the American Enterprise Institute found a net employment effect of +0.8 per cent across 30 randomised pilots, but -3.2 per cent in the four largest ones. That gap ought to give pause.
Then comes the bill. In America, a 12,000 US dollars-per-year payment to every adult would cost roughly 2.4 trillion US dollars annually, almost as much as the entire existing safety net. The ILO has estimated UBI costs at between 20 and 35 per cent of GDP for most rich countries. Opponents wonder where the money comes from when American national debt stands at 36 trillion US dollars. Proponents counter that the gross cost is misleading, given that much of the money is a tax rebate returned to the same people who paid it, and the real redistributive burden may be closer to three per cent of GDP.
Not Utopia
A more modest path exists. Partial UBI (enough to take the edge off, not enough to live on) could serve as a foundation rather than a replacement for existing welfare. Something in the range of 300–500 US dollars a month, roughly what Cook County is testing, would cost a fraction of a full UBI and could be funded by closing tax loopholes, redirecting inefficient subsidies, or taxing the AI that displaces the workers. Sam Altman, the boss of OpenAI, has floated a tax on AI-generated productivity. It is a neat proposition: the machines that take the jobs pay for the safety net.
The political obstacles remain formidable. The right dislikes the expense; the left worries about dismantling targeted programmes in favour of untargeted cash. Both have a point. A UBI that replaces unemployment benefit and housing assistance could leave the most vulnerable worse off. The trick lies in designing it as a supplement, one that sits alongside existing protections rather than in place of them.
No country has introduced a full UBI. Two (Mongolia and Iran) have tried partial versions, both since restructured. The fact that Cook County has made its pilot permanent tells its own story. So does the sheer global sprawl of experimentation: the Stanford Basic Income Lab tracks well over a hundred pilots worldwide. When an idea attracts backing from Friedman’s disciples and from trade unions, from Kenyan villages and Illinois county boards, dismissing it as utopian starts to look like the less serious position.
AI may not produce the employment apocalypse that its most anxious observers predict. But it is already producing a rapid, uneven redistribution of economic power from those who sell their labour to those who own the algorithms. A floor under everyone’s income will not fix that imbalance. It might, at least, stop people falling through it.
Photo: Dreamstime.
