Redistributed

For decades, universal basic income was a parlor game — a thought experiment debated by philosophers, a footnote in economics textbooks, an idea that united libertarians and socialists in mutual suspicion. Then AI started automating white-collar work, and suddenly everyone from Silicon Valley CEOs to presidential candidates was talking about writing checks.

The shift has been rapid. In 2017, UBI was a fringe proposal associated with utopian futurists and a handful of development economists. By 2020, Andrew Yang had built a presidential campaign around it. By 2025, OpenAI had funded one of the largest cash transfer studies in American history. The question is no longer whether UBI deserves discussion. It is whether UBI — as proposed by the people most likely to implement it — would actually redistribute anything, or merely provide political cover for an unprecedented concentration of wealth.

The Tech Billionaire Turn

The most visible UBI advocates are the people building the systems that threaten to make it necessary. This is not a coincidence, and it is not entirely cynical — but it is worth examining carefully.

Sam Altman has been the most persistent. As president of Y Combinator, he funded early UBI research. As CEO of OpenAI, he has proposed "universal basic compute" — the idea that every person should receive a share of AI compute or model access as a form of wealth redistribution. The logic: if AI becomes the primary engine of economic value, then distributing access to AI is more fundamental than distributing cash.

Altman has also backed Worldcoin, now rebranded as "World" — a project that scans people's irises using a chrome orb to create a unique biometric identity, then distributes cryptocurrency tokens. The stated goal is building the infrastructure for global UBI distribution: a verified identity system that could deliver payments to anyone on earth without duplication or fraud. The unstated reality is that a private company is building a global biometric database, asking the world's poorest people to trade their iris scans for tokens of uncertain value. The dystopian undertones are not subtle.

Elon Musk has intermittently endorsed UBI, or what he prefers to call "universal basic income" rebranded as "universal high income" — the idea that AI abundance will be so vast that everyone will receive more than they need. This is less a policy proposal than a hand-wave: trust the technology, and the economics will sort themselves out. It is also conveniently unburdened by details about who pays, how much, or what happens during the transition.

Andrew Yang's 2020 presidential campaign was the inflection point. His "Freedom Dividend" — $1,000 per month for every American adult, funded partly by a value-added tax — brought UBI from academic seminar to cable news debate. Yang's framing was explicitly tied to automation: AI and robots were coming for trucking jobs and retail jobs and call center jobs, and the responsible thing was to provide a floor. He did not win, but he permanently shifted the Overton window. UBI is now a mainstream policy option in a way it was not before.

The pattern is consistent. Tech leaders who are building the systems that automate work are also the loudest voices proposing a stipend for the people whose work is being automated. The question this raises is uncomfortable but necessary: is this genuine concern, political inoculation, or both? The honest answer may be that it is genuinely both — that people can sincerely believe in a policy that also happens to serve their interests. What matters is the policy's design, not the motives of its advocates.

What the Evidence Actually Shows

Unlike most policy debates, UBI has been tested. Not at scale, not permanently, but enough to challenge both utopian and dystopian assumptions.

Finland (2017–2018): The Finnish government gave 2,000 unemployed people €560 per month with no conditions. The results were modest and instructive. Participants reported better wellbeing and lower stress. Employment effects were small — recipients were slightly more likely to work than the control group, but the difference was marginal. The experiment was not renewed, partly due to political opposition, partly because the results did not produce the dramatic employment boost that proponents had promised (Kangas et al., 2019).

Stockton, California (2019–2021): The SEED program, launched by then-Mayor Michael Tubbs, gave 125 residents $500 per month for two years. Recipients were more likely to find full-time employment than the control group. They spent the money on food, utilities, and transportation — the pedestrian necessities that poverty makes precarious. Mental health improved. The program demonstrated that cash transfers work as poverty alleviation. It did not demonstrate that UBI could replace the social safety net or address structural inequality (West & Castro Baker, 2023).

GiveDirectly in Kenya: The largest and longest UBI experiment, giving over 20,000 people roughly $0.75 per day. Early results showed increased consumption, asset accumulation, and modest business formation. But the context — extreme poverty in rural Kenya — is so different from AI-driven job displacement in wealthy countries that extrapolation requires caution.

The OpenAI-funded study (2023–2025): OpenResearch, funded by Sam Altman, gave 1,000 participants $1,000 per month for three years, with a 2,000-person control group receiving $50 per month. The results, published in 2025, told a complicated story. Recipients worked slightly less — about 1.3 fewer hours per week. They reported modestly improved wellbeing and life satisfaction. They spent more on housing, transportation, and healthcare. But the study also found that $1,000 per month did not fundamentally alter participants' economic trajectories. It smoothed hardship without eliminating it. It provided breathing room without providing security (Jaroszewicz et al., 2025).

The honest summary of the evidence: UBI helps. It reduces stress, improves wellbeing, and does not make people stop working. But at the levels tested — $500 to $1,000 per month — it is a supplement, not a transformation. It treats symptoms of economic insecurity without addressing causes.

The Intellectual Debate

The case for UBI is straightforward. It eliminates the poverty trap — the perverse incentive structure in means-tested benefits where earning more money can cost you more in lost benefits than you gain in wages. It provides security during economic transitions, giving workers time to retrain or relocate without destitution. It recognizes the value of unpaid care work — the childcare, eldercare, and domestic labor that is socially essential and economically invisible. And it is administratively simple compared to the labyrinthine welfare bureaucracies it could replace.

The case against is more varied, and more interesting.

The left critique comes from thinkers like Aaron Benanav, whose Automation and the Future of Work argues that UBI without structural change is a palliative. If you give everyone $1,000 per month but leave the ownership of AI systems, data, and compute in the hands of a few corporations, you have not redistributed power — you have merely redistributed enough cash to prevent unrest. Benanav's deeper point is that the "automation crisis" is overstated. The real crisis is not that robots are taking jobs but that capitalism chronically underproduces good jobs. UBI addresses neither the power imbalance nor the structural underemployment (Benanav, 2020).

This is the critique that should trouble UBI advocates most. A stipend funded by general taxation while AI profits remain private is not redistribution — it is cost-shifting. The public bears the expense of maintaining a consumer base while corporations capture the productivity gains. It is welfare for capital dressed as welfare for workers.

The right critique focuses on dependency and incentive effects. If people receive income without working, will they stop contributing? The evidence suggests the effect is small — people reduce work hours modestly, if at all. But the philosophical objection runs deeper: work provides purpose, structure, and social connection. A society of idle stipend-recipients is a society in trouble, regardless of material comfort. This concern is not frivolous, even if its proponents often deploy it cynically to oppose any form of redistribution.

The practical critique is about math. A UBI of $1,000 per month for every American adult (about 260 million people) would cost roughly $3.1 trillion per year — more than the entire federal discretionary budget. Even aggressive taxation of AI profits would not cover it. The numbers only work if UBI replaces existing programs (Social Security, Medicaid, food stamps), which would leave many vulnerable people worse off than before.

The Redistribution Question

Here is where the funding mechanism matters more than the policy itself.

UBI funded by taxes on AI companies — through levies on AI-driven profits, taxes on compute usage, or data dividends paid to the people whose information trained the models — would represent genuine redistribution. Value created by AI, derived partly from collective data and publicly funded research, would flow back to the public. This is the logic Mariana Mazzucato articulates in The Entrepreneurial State: public investment creates the conditions for private innovation, so returns should flow back to the public (Mazzucato, 2013).

UBI funded by general taxation while AI profits remain private is something else entirely. It shifts the cost of economic transition onto workers and taxpayers while the beneficiaries of that transition — shareholders, executives, platform owners — keep their gains. As with stock buybacks, the question is not whether value is created but who captures it.

Several funding mechanisms have been proposed:

Robot taxes. Bill Gates proposed this in 2017: if a human worker earning $50,000 pays income tax, a robot doing the same work should generate equivalent revenue. The EU considered and rejected a robot tax. The administrative challenge is defining what counts as a "robot" — is a spreadsheet a robot? Is autocomplete? The concept is directionally right but technically slippery.

AI windfall taxes. A tax on profits above a threshold for companies deploying AI, capturing the surplus generated by automation. This is more administrable than a robot tax but faces the same lobbying opposition as any corporate tax increase.

Data dividends. The concept, popularized by California governor Gavin Newsom and economist Glen Weyl, treats personal data as labor. If your data trains AI models that generate billions in revenue, you deserve a share. Alaska's Permanent Fund — which distributes oil royalties to every resident — provides a model. Data is the new oil; perhaps its returns should be similarly socialized (Posner & Weyl, 2018).

Compute credits. Altman's "universal basic compute" proposal: rather than cash, distribute access to AI systems. This has a certain logic — if AI is the new infrastructure, then access to it matters more than cash. But it also conveniently ensures that the value stays within the AI ecosystem, flowing through systems that companies like OpenAI control.

Alternatives Worth Considering

UBI is not the only redistribution mechanism, and it may not be the best one. Several alternatives address the power question that UBI sidesteps:

Shortened work weeks. If AI doubles productivity, the same output could be produced in half the time. Rather than eliminating half the jobs, reduce everyone's hours. The four-day work week trials in the UK, Iceland, and elsewhere have shown maintained productivity with improved wellbeing. This redistributes time, not just money — and it has the advantage of having been tested at scale in functioning economies.

Sectoral bargaining. Rather than individual company-level unions, sectoral bargaining sets wages and conditions across entire industries. This would ensure AI productivity gains are shared with workers through higher wages rather than funneled to shareholders. It works in Germany, Denmark, and much of Europe.

Sovereign wealth funds. Norway's Government Pension Fund invests oil revenues for public benefit. A similar fund, capitalized by AI windfall taxes, could generate returns distributed to all citizens — a permanent, growing income stream rather than a fixed stipend subject to annual appropriation. Alaska's Permanent Fund, which has paid dividends to residents since 1982, demonstrates that this model works in the American context.

Worker ownership. If AI transforms the firm, workers should own a share of the transformed firm. Expanded ESOPs, mandatory profit-sharing, or cooperative ownership models would distribute AI gains at the source rather than requiring after-the-fact redistribution through taxes and transfers.

Each of these addresses something UBI does not: the underlying distribution of power. UBI leaves ownership structures intact. These alternatives challenge them.

What This Means for You

If you are watching the UBI debate from the outside, here is what matters:

Follow the funding. Any UBI proposal is only as redistributive as its revenue source. UBI funded by AI windfall taxes redistributes. UBI funded by cutting Social Security does not. The policy is the funding mechanism, not the check.

Be skeptical of tech-funded solutions. When the people who profit most from automation propose to give you a stipend, ask what they are not proposing. Are they offering to be taxed? To share ownership? To accept regulation? A stipend without structural change is a concession designed to foreclose more ambitious demands.

The pilot results are real but limited. Cash transfers help people. This is established. But $1,000 per month does not solve the problem of an economy where a handful of companies capture most of the value that AI creates. The question is not "does UBI work?" but "is UBI sufficient?"

Alternatives exist. Shorter work weeks, sectoral bargaining, sovereign wealth funds, and worker ownership are not utopian fantasies — they exist in functioning economies today. The conversation should not be "UBI or nothing" but "what combination of policies would actually redistribute the gains from AI?"

The UBI debate has moved from fringe to mainstream because AI has made the question of economic redistribution unavoidable. That is progress. But the risk is that UBI becomes the ceiling of ambition rather than the floor — a minimum concession that substitutes for the deeper structural changes that genuine redistribution requires.

Here is a stipend. Now please do not ask who owns the machines.

Sources

Benanav, Aaron. Automation and the Future of Work. Verso Books, 2020. Publisher
Jaroszewicz, Ania, Jon Jachimowicz, Oliver Hauser, and Elizabeth Rhodes. "The Effects of a Guaranteed Income: Findings from the OpenResearch Study." OpenResearch, 2025. OpenResearch
Kangas, Olli, Signe Jauhiainen, Miska Simanainen, and Minna Ylikännö. "The Basic Income Experiment 2017–2018 in Finland: Preliminary Results." Ministry of Social Affairs and Health, Finland, 2019. Julkaisut
Mazzucato, Mariana. The Entrepreneurial State: Debunking Public vs. Private Sector Myths. Anthem Press, 2013. Publisher
Posner, Eric A. and E. Glen Weyl. Radical Markets: Uprooting Capitalism and Democracy for a Just Society. Princeton University Press, 2018. Publisher
West, Stacia and Amy Castro Baker. "Stockton Economic Empowerment Demonstration: SEED's Guaranteed Income Results." SEED, 2023. SEED

General

Universal Basic Income and AI: The Debate Gets Real