🔍 Read the full analysis: Ellison’s “Universal High Income” And The Evidence From Basic Income Trials on ThorstenMeyerAI.com
TL;DR
Oracle co-founder Larry Ellison said AI-generated value could be taxed to fund “universal high income,” a proposal he said would take more than a decade. South Korea is already testing monthly payments in rural counties, while trials in the United States, Finland and Kenya show varied effects on work and earnings. Neither those trials nor Ellison’s remarks establish how a nationwide AI-funded income program would be financed or operate.
Oracle co-founder Larry Ellison said October 27 that governments could tax value generated by robots and artificial intelligence to fund what he called “universal high income”, a proposal he said would take more than a decade to develop. His remarks come as South Korea runs a rural cash-payment pilot, offering a current test of regular public payments—but not of Ellison’s proposed tax or an AI-funded national income.
Ellison made the comments during a question-and-answer session at Oracle AI World in Las Vegas, responding to a question about economic models for a world with more robots and AI. Accounts by TechRadar, Diginomica and Business Standard report his central proposal: shift taxation away from human labor and toward value added by automation. Oracle’s event replay is behind an account login, according to the source material, so the quotations are based on those media reports rather than an independently accessible recording.
Ellison argued that predictions of AI replacing everyone were “kind of nonsensical,” while acknowledging that some occupations, including taxi and ride-share driving, could be displaced. He used a robot-built house as an example of how lower labor costs might reduce prices. He also invoked economist Milton Friedman’s negative income tax, though that policy—which sends payments to people below an income threshold that decline as earnings rise—is distinct from a broad payment to everyone.
South Korea’s rural basic income pilot began in February 2026 and is scheduled to run through 2027. It pays eligible residents of 17 counties identified as losing population 150,000 won a month, about $112 according to Yonhap’s conversion, in local gift vouchers. The source says residents must be registered and actually live in a pilot county, with a 30-day residency requirement before applying; there is no means test or work requirement. The available source text ends before giving all eligibility details, so its rules for newcomers and people working or studying elsewhere cannot be fully summarized here.
Funding is split among levels of government: in 2026, the national government covers 40% of the cost, the province 30% and the county 30%. For 2027, the agriculture ministry has proposed a 50/25/25 split and adding about 18 counties, but that expansion depends on a parliamentary budget vote. The proposed change is not yet a confirmed outcome.
Infographic 01 · Two ideas of a basic income
A promise for the future, a test running now
What Oracle’s co-founder proposed on stage, and how South Korea’s rural basic income actually works and who funds it.
The promise
Larry Ellison, Oracle AI World, Las Vegas, 27 Oct 2026 (as reported)
- Robots build a house: if labor was 80% of the cost, the house gets 80% cheaper
- Change from taxing human labor to taxing value added
- Tax what robots add, fund a “universal high income”
- AI replacing everyone is “kind of nonsensical”
- No amounts, no costing, “more than a decade”
“We need to change from taxing human labor to taxing value added.”Larry Ellison, as quoted by TechRadar, Diginomica and Business Standard
The test
South Korea’s rural basic income pilot, 2026–2027
- Every citizen registered and living in a pilot county for 30+ days
- No means test, no job requirement; newcomers wait 90 days
- Paid as a local gift voucher (card or phone), not cash
- Expires after 3 months (towns) or 6 months (townships)
- Max ₩50,000 at gas stations, convenience stores, co-op supermarkets
- 10 counties from February, 7 more from 28 August; about 18 more planned for 2027
Who pays in 2026
Share of the ₩150,000 base payment
Proposed for 2027
Ministry’s budget proposal; parliament decides
Some counties add their own money: Gurye, Boseong and Cheongsong pay ₩200,000, Boeun ₩160,000 (ministry, 25 Aug). The ministry puts the total cost of the 2026–27 pilot at about ₩3.09 trillion.
Sources: TechRadar, 30 Oct · Diginomica, 27 Oct · Business Standard, 28 Oct · Ministry of Agriculture, Food and Rural Affairs: MAFRA, 10 Feb, 11 Jun, 25 Aug, 8 Oct 2026.
Powered by Thorsten Meyer AI
What Cash Trials Say About Work
Past basic-income and cash-transfer studies offer evidence about how recipients respond to payments, but they do not settle the larger question Ellison raised: who should pay when automation creates value, and how much revenue that would raise. The findings described in the source vary by program and location. In the US OpenResearch study, recipients of $1,000 a month earned about $1,900 less per year and worked one to two fewer hours a week. Finland’s trial produced little change in employment, while Kenya’s study found some recipients moved from wage work toward running their own businesses.
Those results complicate claims that regular payments would either end work or have no effect on it. They also cannot be treated as direct forecasts for a universal, permanent program: the studies differ in payment amounts, duration, populations and economic settings. The source cites an estimate that $1,000 a month for every US adult would cost about $3.1 trillion annually, compared with $5.40 trillion in federal receipts in fiscal 2026. That is the source author’s arithmetic, not a government cost estimate, and it does not specify a funding plan.
Price effects are another open issue. An International Labour Organization research brief, as described in the source, warns that extra spending could raise prices in sectors where supply does not grow with demand, including housing and care. A payment funded by taxing automated production would also depend on the size of the tax base, how the tax is designed and whether it changes companies’ investment decisions. None of these questions is answered by the Korean pilot.
Popular right now
Vitamix Propel Series 750 Professional-Grade Blender, 4 Programs, Black
Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
De'Longhi 15-Bar Pump Espresso Machine with Milk Frother & Temperature Control
As an affiliate, we earn on qualifying purchases.
A Proposal Beside a Live Pilot
Ellison’s remarks are a proposal, not an announced Oracle policy. Diginomica reported that robot-tax ideas were not an Oracle strategy and disclosed that Oracle is a Diginomica partner. The idea also predates this speech: the source notes that other technology figures have proposed similar approaches, and that Sam Altman’s 2021 essay “Moore’s Law for Everything” argued for taxing capital and land rather than labor.
The tax language needs care. A value-added tax is generally collected as goods and services gain value through production and sale. The source reports that the United States uses retail sales taxes rather than a federal VAT as its main consumption tax, unlike other OECD countries. Ellison’s phrase, however, does not specify a legislative design, tax rate or mechanism for directing revenue to households. A negative income tax would be a separate approach targeted by income level.
South Korea’s program is a limited, geographically targeted test, not a universal basic income and not a robot-tax experiment. Payments are made in local gift vouchers, and the program’s stated focus is residents in counties losing population. Its value for the AI debate is that it provides a current example of public cash support in practice, while its scale and funding arrangement differ sharply from Ellison’s hypothetical national system.
„“We need to change from taxing human labor to taxing value added.”“
— Larry Ellison, as quoted by TechRadar and Diginomica
Funding and Policy Design Remain Open
Ellison did not set out a tax proposal with a rate, revenue estimate, definition of taxable value or process for distributing payments. It is also unclear whether “universal high income” means a payment to every person, a level of income guaranteed through a tax system, or another model. His comparison with a negative income tax points to a separate policy concept rather than resolving that ambiguity.
It remains uncertain how much revenue automation could generate, who would bear the tax in practice, and whether the proceeds could support payments at a scale comparable to the US estimate cited in the source. The basic-income studies discussed here do not test that funding mechanism. South Korea’s planned 2027 expansion and proposed funding split are also subject to a parliamentary budget decision. Further details on the pilot’s eligibility rules are not present in the supplied source excerpt.
Korea’s 2027 Budget Decision
The next stated decision for the Korean pilot is the parliamentary vote on the 2027 budget, which will determine whether the proposed 50/25/25 funding split and expansion to about 18 additional counties proceed. The pilot is scheduled to continue through 2027; the supplied material does not give a date for a final evaluation or report on its effects.
For Ellison’s proposal, the source identifies no legislation, timetable beyond his “more than a decade” estimate, or formal Oracle initiative. The immediate question is therefore whether public officials or researchers develop a concrete design that can be costed and compared with existing tax and income-support policies. Until then, the Korean program and earlier trials can inform discussion of payment delivery and recipient responses, but they cannot verify the feasibility of an AI-funded universal income.
Key Questions
What did Larry Ellison propose?
Ellison said governments could shift taxation from human labor toward value created by robots and use the revenue for what he called “universal high income.” He described the idea as more than a decade away and did not provide a detailed tax or payment plan.
Is Oracle launching a robot tax or income program?
No such Oracle program is identified in the source. Diginomica reported that robot-tax ideas were not an Oracle strategy; Ellison’s comments were made during an event discussion.
How does South Korea’s rural pilot work?
The program pays eligible residents in 17 rural counties 150,000 won a month in local gift vouchers. It began in February 2026 and is scheduled to run through 2027. A proposed expansion and revised funding split for 2027 still require a parliamentary budget decision.
Do basic-income trials show that people stop working?
The results cited here do not show one uniform response. The US OpenResearch study found lower earnings and a small reduction in weekly work hours among recipients; Finland’s trial saw little employment change, while Kenya’s study found some movement from wage work to self-employment. These findings are specific to their studies and do not establish what a permanent universal program would do.
Could robot-generated taxes pay for a universal income?
The source material does not establish that. Ellison did not provide a tax design or revenue estimate, and the cited trials did not test an AI-funded tax system. The amount of taxable value, the cost of payments and possible price effects remain open questions.
Source: ThorstenMeyerAI.com
