The Question We're Not Asking
Something unprecedented is happening to the global economy, and we are not preparing for it. Artificial intelligence is advancing at a pace that outstrips every prior technological revolution — not by degree, but by kind. The steam engine displaced muscle. The computer displaced routine calculation. AI is poised to displace cognition itself — the one capability that, until now, kept human labor indispensable.
We don't know the timeline. Reasonable forecasts range from a decade to a generation. We don't know the extent — whether AI will automate 20% of jobs or 80%. But we know the direction, and we know that the economic consequences of large-scale displacement, if unmanaged, would be catastrophic: collapsing tax revenues, mass unemployment, social instability, and political extremism. Every one of these outcomes is preventable — if we build the policy infrastructure now, before the crisis arrives.
This article introduces a simple economic model and a legislative framework called the FAIR AI Act — the Federal Automatic Income Replacement for AI Displacement Act. Its purpose is not to predict the future, but to make the future manageable regardless of how it unfolds.
The Model: What It Does and What It Assumes
To design policy, you need to understand the fiscal math. We built an interactive model that takes a small number of inputs and produces the tax rates required to maintain economic stability as AI displacement grows. The inputs are deliberately simple — this is not a prediction engine but a scenario explorer.
On the displacement side, the model takes white-collar and blue-collar displacement as separate percentages of the U.S. labor force (approximately 160 million workers, split roughly 60/40). Displaced workers receive a Universal Basic Income payment — defaulting to $40,000 per year, roughly the median individual income — sufficient to maintain a basic standard of living and prevent the kind of mass economic distress that historically fuels social unrest.
On the revenue side, the model takes total AI company revenue and a “value capture ratio” — the percentage of total AI-generated economic value that flows to AI companies as taxable revenue. This distinction matters enormously. When an AI tool saves a bank $500 million in labor costs but the AI vendor only receives $50 million in licensing fees, $450 million in value has been created outside the taxable base. The capture ratio, defaulting to 25%, makes this gap visible.
On the cost side, the model accounts for something often overlooked: as workers are displaced, they stop paying income and payroll taxes. Federal income and payroll taxes currently generate approximately $2.5 trillion per year. As displacement rises, this revenue base erodes — meaning the AI tax must fund not only UBI payments but also replace the vanishing tax receipts that fund the rest of government. The model also allows users to specify what share of the existing $1.8 trillion annual deficit the AI tax should help address.
The model reveals a striking nonlinearity: at low displacement, the required tax rate is modest. At high displacement, it becomes untenable on AI company revenue alone — signaling that the tax base must broaden.
The model produces four layered outputs: the tax rate on AI company revenue needed to cover (1) UBI costs alone, (2) UBI plus lost tax revenue, (3) UBI plus lost tax revenue plus deficit coverage, and (4) the implied rate if you could tax all AI-generated economic value, not just AI company revenue. Built-in sanity checks flag when rates exceed plausible profit margins or when the scenario simply cannot be funded through AI company taxation alone.
What the Model Reveals
Run a few scenarios and three findings emerge consistently.
First, the early stages are manageable. At 10–15% displacement with $2–3 trillion in AI revenue, the required tax rates are in the range of existing corporate tax burdens. This is the window in which to act — when the policy is inexpensive and the political resistance is low.
Second, the lost tax revenue problem is bigger than the UBI problem. Most people intuitively focus on the cost of supporting displaced workers. But the collapse of the income and payroll tax base — which funds Social Security, Medicare, defense, and everything else — is often the larger fiscal hole. At 40% displacement, roughly $625 billion in annual tax revenue disappears. The AI tax isn't just funding a new program; it's replacing the fiscal foundation of the federal government.
Third, at high displacement, taxing AI companies alone is not enough. When the model's sanity checks start firing — typically around 40–50% displacement with moderate AI revenues — it signals that the tax base must expand beyond AI companies to include all businesses benefiting from AI-driven productivity gains. This is the single most important policy insight the model produces: we need the legislative framework to scale with the problem.
Why the Response Must Be Dynamic
Here is the core policy argument: the pace of AI-driven economic change will almost certainly outrun the pace of the legislative process. Congress takes months to years to pass major legislation. AI capabilities are advancing on a timeline measured in quarters. A static tax rate set by Congress in 2027 could be wildly insufficient by 2029 or punitively excessive if adoption slows.
The solution is a formula-based, self-adjusting tax mechanism — enacted once, through legislation, and then allowed to recalculate automatically based on published economic indicators.
This is not a radical concept. Tax brackets already adjust annually for inflation without legislative action. Social Security benefits adjust through cost-of-living formulas. The Federal Reserve sets monetary policy through institutional mechanisms precisely because democratic deliberation is too slow for real-time economic management.
The FAIR AI Act would work the same way. Congress would define the formula, designate the inputs (Bureau of Labor Statistics employment data, SEC-reported AI company revenues, published productivity metrics), establish the responsible agency, and set ceiling and floor rates. The tax rate would then adjust on a quarterly or annual cycle — rising as displacement increases, falling if the economy absorbs the transition more smoothly than expected.
This approach has three critical advantages. It eliminates legislative lag. It removes the opportunity for political paralysis at the moment of maximum crisis. And it creates predictability for AI companies, who can model their future tax exposure based on transparent, published formulas rather than waiting for the next unpredictable act of Congress.
What the FAIR AI Act Would Include
The legislation would need to establish several components:
A defined tax base. The primary tax would apply to the gross revenues of companies deriving a significant share of income from AI products and services, with clear definitional criteria updated by the administering agency. A secondary, broader levy could apply to all companies whose productivity gains (measured by revenue-per-employee growth) exceed a threshold, capturing AI-generated value wherever it flows.
The adjustment formula. Codified in the statute with explicit mathematical terms. The formula would take as inputs: the national displacement rate (measured by BLS), AI-sector revenues (measured by SEC filings and IRS data), lost federal tax receipts from displaced workers, and the target UBI amount (indexed to median income). The output: the applicable tax rate for the upcoming period.
An independent administering body. A new office or board — comparable in independence to the Federal Reserve or the Congressional Budget Office — charged with certifying the input data, running the formula, and publishing the resulting rate. This body would have no discretion over the formula itself, only the data inputs, minimizing concerns about delegation of legislative authority.
Rate ceilings and floors. A statutory maximum rate (perhaps 60–70% of revenue) to prevent confiscatory outcomes that would destroy the AI industry, and a minimum rate (perhaps 2–5%) to maintain the administrative infrastructure even in low-displacement scenarios.
A UBI disbursement mechanism. Direct payments to displaced workers, administered through existing IRS and Social Security infrastructure. Eligibility would be determined by employment status and prior work history, with payments indexed to a target standard of living.
Mandatory congressional review. Every five years, Congress would be required to review and reauthorize the formula parameters — not the mechanism itself, but the specific coefficients and thresholds. This provides a democratic check without requiring Congress to act in real time.
This Is Not a Left or Right Issue
The instinct to categorize this proposal will be strong. Progressives will see a tax on corporations and a government payment to individuals and assume it belongs to them. Conservatives will see a massive new entitlement and recoil. Both reflexes are wrong, because both assume the economy of the future will look like the economy of the past.
The traditional conservative argument against redistribution rests on the premise that markets, left alone, will create sufficient employment and wages to sustain broad prosperity. That premise has been roughly correct for two centuries. It may not survive the next two decades. When AI can perform cognitive work at a fraction of the cost of human labor — not just routine tasks, but complex analysis, creative work, strategic decision-making — the labor market does not “adjust.” It transforms. And if the transformation is fast enough and broad enough, the market alone cannot provide the bridge.
The traditional progressive argument for redistribution often rests on claims about inequality and fairness. Those claims are valid, but they're not what drives this proposal. The FAIR AI Act is driven by math. If 50 million Americans lose their incomes over a ten-year period and there is no mechanism to sustain their purchasing power, the economy collapses — not because of ideology, but because consumer spending is two-thirds of GDP. The companies that built the AI will see their own revenues crater as their customers disappear. Everyone loses.
This is not redistribution. It is economic stabilization.
The same logic that justifies deposit insurance, the Federal Reserve, and unemployment insurance justifies a dynamic fiscal response to the largest labor market disruption in human history.
The FAIR AI Act is pro-growth. It allows AI development to proceed at full speed by removing the political pressure for outright bans or moratoriums that would otherwise build as displacement mounts. It is pro-business. AI companies gain predictability and social license to operate, rather than facing an unpredictable patchwork of state-level restrictions born of desperation. It is pro-stability. It prevents the kind of mass economic dislocation that, historically, produces not thoughtful policy but radicalism, demagoguery, and institutional collapse.
Whether you believe government should be large or small, you should believe it should be solvent. The current federal revenue structure depends on 160 million people earning taxable incomes. AI threatens to erode that base in a way no prior technology has. A dynamic tax that replaces lost revenue while funding the transition is not an expansion of government. It is the preservation of government's ability to function at all.
What If the Optimists Are Right?
There is a natural objection to all of this: what if AI displacement turns out to be modest? What if new job categories emerge, as they have after every prior technological revolution, and the labor market adjusts? What if the techno-optimists are vindicated?
This is precisely the scenario in which a dynamic mechanism proves its worth. If displacement stays low — say 5–10% of the workforce — the formula produces a tax rate of perhaps 3–5% on AI company revenue. At that level, it is a rounding error for an industry generating trillions in annual sales. It builds a modest stabilization fund that sits quietly as insurance, costing the economy essentially nothing. The administrative infrastructure hums along in the background, dormant but ready.
If Sam Altman's vision materializes — AI makes everything so cheap that $15,000 a year provides a higher standard of living than $40,000 does today — the UBI target drops, the displacement inputs fall, and the rate shrinks further. The mechanism celebrates good news by getting out of the way. A formula-based tax does not presume catastrophe. It prepares for it while self-correcting if catastrophe does not arrive.
Now consider the alternative: no mechanism in place, and the pessimistic scenario unfolds. Displacement hits 30, 40, 50 percent over a decade. Tax revenue collapses. Congress, gridlocked in the best of times, faces simultaneous demands to fund a massive safety net and replace hundreds of billions in lost revenue — all while navigating the most intense corporate lobbying campaign in history from an AI industry that is, by then, the most powerful economic force on Earth. The legislation that emerges from that crisis will be written in panic, shaped by populist rage, and almost certainly worse for everyone — including the industry — than a formula calmly enacted years earlier.
The cost of being wrong about enacting this is near zero — a small, quiet tax that phases itself down. The cost of being wrong about not enacting it is catastrophic. That asymmetry is the entire argument for acting now.
The Window Is Now
There is a narrow window in which this policy can be enacted wisely, calmly, and with broad support. That window exists before mass displacement begins — when the tax rates the formula would produce are low, when AI companies are not yet threatened by the mechanism, and when the political environment has not yet been poisoned by economic desperation.
Wait too long, and the policy becomes reactive rather than preventive. It will be drafted in crisis, shaped by anger rather than analysis, and almost certainly worse for everyone — including the AI industry. The history of financial regulation teaches this lesson clearly: the regulations written after the 2008 crisis were harsher and less efficient than the ones that could have been written before it.
The interactive model accompanying this article is deliberately simple. It is not a forecast. It is a tool for exploring the fiscal geometry of a problem that is coming whether we prepare for it or not. Move the sliders. See where the math breaks. Understand what those break points mean for your community, your industry, and your country.
A Call to Candidates
We are calling on every candidate for federal office — in 2026 and beyond, in every party and in every district — to take the FAIR AI Pledge:
I pledge to support the development and passage of the FAIR AI Act — a Federal Automatic Income Replacement framework that dynamically adjusts to protect American workers, preserve federal solvency, and ensure that the economic benefits of artificial intelligence are shared broadly across society. I commit to working across party lines to enact this legislation before displacement demands it, not after.
This is not about being for or against AI. It is about being prepared. The technology is coming. The displacement will follow. The only question is whether we build the bridge before the flood or after.
The FAIR AI Act is that bridge.