Can AI Pay for a Universal Basic Income?

AI may create enough new wealth to pay for a universal basic income, but only if governments can tax that wealth where it lands. PwC’s estimate that AI could add up to $15.7 trillion to the world economy by 2030 measures the size of the gain; it says nothing about who will collect it.

This article sets out what the evidence says about the gain, what an income for everyone would cost, and which tax ideas are on the table. It is written for policy makers and the people who advise them.

How much wealth could AI create?

Two widely cited estimates give the order of magnitude. PwC’s Sizing the prize (2017) projected that AI could add up to $15.7 trillion to global GDP in 2030: $6.6 trillion from higher productivity and $9.1 trillion from new consumer demand. Goldman Sachs (2023) estimated that generative AI could raise global GDP by about 7%, almost $7 trillion, over ten years.

The same Goldman Sachs analysis found that the equivalent of 300 million full-time jobs could be exposed to automation. The IMF (2024) put exposure at almost 40% of global employment and about 60% in advanced economies, though it expects roughly half of the exposed jobs in advanced economies to benefit from AI rather than lose out. Its conclusion for every economy is that social safety nets and retraining are crucial.

These are forecasts, not measurements. They show that the gain could be large; they do not show how it will be shared.

How much would a universal basic income cost?

A simple calculation shows the scale. Sam Altman’s essay counts about 250 million adults in the United States. Paying each of them $1,000 a month, or $12,000 a year, would cost about $3 trillion a year. That is our own arithmetic, for one country, and close to half of PwC’s projected productivity gain for the whole world.

Real proposals are usually smaller or targeted: they phase out with income, replace some existing benefits, or start with a lower amount. The point stands: a universal income is a permanent cost of trillions a year, so it needs a permanent source of money.

Can taxing AI fund a universal basic income?

Three kinds of idea are discussed.

Taxing capital rather than labour. In Moore’s Law for Everything (2021), Sam Altman suggested an American Equity Fund paid for by a tax of 2.5% a year on the market value of large companies, paid in shares, and 2.5% on the value of privately held land. Every adult would receive a yearly distribution, which he estimated at about $13,500 within ten years. He presents it as a conversation starter, not a costed plan.

Paying a dividend from a shared asset. Alaska already does this with oil revenue. Its Permanent Fund Dividend for 2026 is $1,200 per eligible resident, including a $200 energy relief payment. It shows that a universal payment can be run year after year; it also shows how far a resource dividend is from a full income.

Stopping profits from moving abroad. AI profits can be booked wherever a company chooses to put its servers and its intellectual property. The OECD’s global minimum tax rules apply a top-up tax so that large multinationals pay at least 15% on low-taxed income. That protects a country’s tax base; it does not create the new revenue a universal income would need.

What could go wrong?

Analysis: the main risks are economic, not technical.

• Inflation. If incomes rise faster than the supply of goods and services, prices rise. Supporters argue that AI will lower the cost of many goods; that is a hope about the future, not yet evidence.
• Work. The largest US trial so far, OpenResearch’s Unconditional Cash Study, paid 1,000 people $1,000 a month for three years. Recipients worked 1.3 fewer hours a week on average and were 2 percentage points less likely to be employed than the control group.
• Timing. The jobs crisis that would make a universal income urgent has not arrived. The Budget Lab at Yale found no discernible disruption of the broader US labour market in the 33 months after ChatGPT was released. A programme built in a hurry during a crisis is likely to be designed badly.

The benefits are real too: a guaranteed floor gives people time to retrain, move or start something new, and it is simpler to run than many conditional benefits.

What should policy makers do now?

1. Measure before promising. Track which tasks and jobs AI is changing, using labour data, not forecasts alone.
2. Fix the tax base first. Apply the global minimum tax and decide how AI-driven profits and capital gains will be taxed before deciding how to spend them.
3. Run pilots with clear questions. Test amounts, eligibility and delivery on a small scale, with a control group and a baseline survey.
4. Strengthen the safety net you have. The IMF’s advice, safety nets and retraining, does not need a universal income to start.

For how a payment would actually reach people, see How Would an AI Dividend Reach Everyone?. For the evidence from basic income trials, see Universal Basic Income and AI: What the Trials Show.

Frequently asked questions

How much could AI add to the global economy?

PwC’s 2017 study estimated that AI could add up to $15.7 trillion to global GDP in 2030: $6.6 trillion from higher productivity and $9.1 trillion from consumer demand. Goldman Sachs estimated in 2023 that generative AI could raise global GDP by about 7%, almost $7 trillion, over ten years.

How much would a universal basic income cost in the United States?

As a simple illustration, paying $1,000 a month to each of the roughly 250 million American adults that Sam Altman’s essay counts would cost about $3 trillion a year. Real proposals differ in amount, eligibility and how they replace existing benefits.

What is Moore’s Law for Everything?

It is a 2021 essay by Sam Altman. It suggests an American Equity Fund paid for by a 2.5% annual tax on the market value of large companies and on privately held land, paying every adult a yearly dividend he estimates at about $13,500 within ten years. He calls it a conversation starter, not a costed plan.

Can a global minimum tax stop companies moving AI profits abroad?

It helps. The OECD’s global minimum tax rules apply a top-up tax so that large multinationals pay at least 15% on low-taxed income. It does not by itself create a new tax on AI, so a universal income would still need its own revenue.

References

1. PwC (2017), Sizing the prize: What's the real value of AI for your business and how can you capitalise? (archived copy) — Up to $15.7 trillion added to global GDP in 2030; $6.6 trillion from productivity and $9.1 trillion from consumption effects.
2. Goldman Sachs (2023), Generative AI could raise global GDP by 7% — About 7% (almost $7 trillion) higher global GDP over ten years; the equivalent of 300 million full-time jobs exposed to automation.
3. IMF (2024), Gen-AI: Artificial Intelligence and the Future of Work, Staff Discussion Note SDN/2024/001 — Almost 40% of global employment exposed to AI, about 60% in advanced economies; social safety nets and retraining are crucial.
4. Sam Altman (2021), Moore's Law for Everything — The American Equity Fund proposal: 2.5% a year on large companies' market value and on privately held land; about $13,500 per adult per year.
5. OECD, Global Minimum Tax — The GloBE rules impose a top-up tax on low-taxed income of in-scope companies up to 15%.
6. Alaska Department of Revenue, Permanent Fund Dividend — A working resource dividend: the 2026 payment is $1,200, including a $200 energy relief payment.
7. OpenResearch, Unconditional Cash Study — Randomised trial: $1,000 a month for three years; recipients worked 1.3 fewer hours a week and were 2 percentage points less likely to be employed.
8. The Budget Lab at Yale (2025), Evaluating the Impact of AI on the Labor Market: Current State of Affairs — No discernible disruption of the broader US labour market in the 33 months after ChatGPT’s release.