The Floor We Refuse to Build

The Floor We Refuse to Build

On May 13, 2026, the Federal Reserve released its annual report on the economic well-being of American households. The headline figure—73 percent of adults said they were "doing okay" or "living comfortably"—was dutifully cited by most outlets and promptly forgotten. What those same outlets largely ignored was the number sitting directly beside it: only 63 percent of adults said they could cover a $400 emergency using cash. In the same report, one in four workers said they had used generative AI on the job in the prior month.

I've seen a lot of economic data framed to obscure what it actually shows. This is a near-perfect example. The administration and the financial press want you to take away "73 percent doing okay." The honest takeaway is that roughly 100 million American adults are one broken transmission away from financial crisis, in the same economy that just handed the AI industry its most profitable quarter on record.

That gap is not an accident. It is a policy choice. And it is getting harder to ignore.

This Is Not a Scarcity Problem

Let me be direct about something that frustrates me in most basic income debates: the people arguing against a cash floor rarely engage with the actual structural argument. They point to the gross cost figure—a $1,000 monthly adult basic income runs to roughly $3.24 trillion annually before taxes, clawbacks, and program offsets—and treat that number as a self-evident refutation. The implicit assumption is that the United States is running a tight fiscal ship and simply cannot afford new expenditures.

This is a strange argument to make in a country that, according to the Federal Reserve's own Distributional Financial Accounts, allowed the top 1 percent to accumulate 31.7 percent of aggregate net worth by Q3 2025, while the bottom half of households share 2.5 percent. The scarcity is real for those households. It is not a feature of the overall economy.

The parallel I keep returning to is the Nullification Crisis of the 1830s. When Southern states pushed back against the "Tariff of Abominations," the surface dispute was about import duties, but the structural issue was whether federal economic policy could be designed to benefit one class of people at a sustained, deliberate cost to another. We resolved that crisis badly, in ways that echoed for generations. The question underneath the current basic income debate is structurally similar: who does the economy's distributional architecture actually serve, and who has the power to change it?

The current architecture, to be blunt, was built. It was built through tax policy, through corporate law, through labor regulation, through the design of the social safety net, through what gets subsidized and what gets means-tested into uselessness. None of it fell from the sky. Which means none of it is permanent.

The AI Economy Is Accelerating the Problem

I want to be precise about what artificial intelligence does and does not change here, because the discourse around this is genuinely muddled.

AI is not a moral agent. It does not have preferences about distribution. What it is—and this part matters—is a high-speed synthesis engine built almost entirely from the accumulated public output of human civilization: research funded by governments, language developed over centuries, knowledge produced by workers, teachers, engineers, and grandmothers who wrote nothing down but trained the people who did. The profits from that synthesis are currently flowing with remarkable efficiency toward a fairly narrow set of capital owners.

The IMF, which is not an institution given to radical politics, flagged this directly in its 2024 analysis of generative AI's fiscal implications. Its researchers warned that AI-driven productivity gains risk producing substantial job displacement and greater inequality, and argued that governments would need stronger social protection and heavier taxation of capital income to prevent technological abundance from becoming mass precarity. That is the IMF. When the IMF is telling you to tax capital and expand the safety net, the "fiscally responsible" case for doing nothing deserves some skepticism.

The self-proclaimed abundance advocates who claim to want AI-powered prosperity but resist any redistribution mechanism remind me of the people who say they want energy independence but oppose the permitting reforms that would actually make domestic production viable. You cannot say you want the upside of a structural transformation while refusing to engage with what that transformation structurally requires.

Cash Works. The Evidence Is Not Ambiguous.

There is a version of the anti-UBI argument that engages with the research and says the effects are too modest to justify the cost. I disagree with that conclusion, but at least it's an honest argument. The version I encounter more often simply asserts, without evidence, that giving people cash will cause them to stop working, stop contributing, and generally revert to some baseline of human laziness.

The research does not support this. Alaska has run a universal cash dividend since 1982. A peer-reviewed study in the American Economic Review found no negative employment effect and an increase in part-time work. Finland's basic income experiment found small labor market effects and significant, measurable improvements in mental health, economic security, and confidence in the future. A large-scale cash transfer program in Kenya found positive economic spillovers to non-recipients, minimal inflation, and a local economic multiplier of 2.4—meaning each dollar distributed generated $2.40 in local economic activity.

If a government program demonstrated a 2.4x return in any other domain, the people currently arguing against basic income would be holding press conferences about it.

The Means-Testing Trap

Here is a frustration I want to name directly: the political coalition that most loudly opposes a universal income floor is often the same coalition that also opposes the robust means-tested programs that would serve as the alternative. The argument shifts depending on which is being debated. Against universal programs, the objection is cost and moral hazard. Against targeted programs, the objection is dependency and government overreach. What both objections have in common is their conclusion: do nothing.

Means-tested programs have a real structural problem that gets underreported. The more precisely you target a benefit, the more administrative infrastructure you need to define eligibility, verify compliance, process applications, and manage the inevitable errors. That infrastructure is expensive, it creates benefit cliffs that punish people for earning more, and it trains the public to view recipients as a separate and suspect class rather than as participants in a shared civic institution.

A universal floor avoids those traps by design. It is paid to everyone. It is unconditional. It does not require the government to predict your specific need category in advance or to monitor whether you remain desperate enough to qualify. It simply establishes that participation in this economy and this society entitles you to a minimum stake in its output.

That is not a radical premise. It is the logic behind Social Security, which polls at around 80 percent approval and has been politically untouchable for decades. The design principle is the same: universality creates shared ownership, and shared ownership creates political durability.

The Real Cost Question

Let me return to the cost, because serious people deserve a serious answer.

Yes, $3.24 trillion is the gross annual figure for a $1,000 monthly adult basic income at current population levels. No, that is not the net fiscal burden, because a substantial portion of that payment flows to households that will pay more in taxes than they receive in benefits. The check is the same; the tax bill is progressive. You can preserve the simplicity of a universal payment while maintaining a distributional structure that concentrates the net benefit on the households that need it most.

The more pointed question is what we are already paying to maintain the current system. Emergency room utilization for conditions that could have been treated with routine care. Homelessness infrastructure. Predatory lending markets that exist specifically to service households without financial cushions. Administrative bureaucracy for programs that spend enormous resources determining who is poor enough to qualify for help. Political extremism that research consistently links to economic insecurity and status anxiety.

The United States is not choosing between spending and not spending on economic precarity. It is choosing between spending on a floor and spending on the consequences of not having one. The current choice is expensive, cruel, and largely invisible in the federal budget because it is distributed across emergency rooms, county jails, food banks, payday lenders, and family court systems instead of appearing as a single legible line item.

Visible support for ordinary people is always asked to justify itself more aggressively than invisible support for institutional failure. That asymmetry is worth naming every time this debate comes up.

What AI Cannot Do

A language model can summarize every peer-reviewed study on cash transfers, map every benefit cliff in the federal safety net, draft model legislation, and generate the counterarguments to every objection I have raised in this piece. I know, because I have used them to do exactly that kind of work.

What AI cannot do is create the political will to act on any of it. That work is irreducibly human and irreducibly slow—organizing, coalition-building, electoral pressure, institutional patience, and the willingness to keep making the same argument until it becomes impossible to ignore.

The technological moment we are in makes the distributional question more urgent, not less. If productivity gains increasingly flow to capital while workers absorb the disruption of automation and restructuring, the basic income movement has a cleaner, more intuitive argument than it has ever had: the economy is producing more than enough to prevent mass insecurity, and it is a policy choice that it doesn't.

Making that argument stick requires more than a well-designed proposal. It requires the kind of political infrastructure that turns an elegant idea into an unavoidable one.

A Floor Is Where You Start

Humanism, as a practical matter, means using the intelligence and resources we have collectively built to solve problems we have collectively created. We built the rules that route productivity gains upward. We built the bureaucracies that make aid degrading. We built the technologies accelerating this transition. The distribution we live with is not natural. It is legible, it was designed, and it can be redesigned.

Basic income is not the end of that project. It is the precondition for it. Workers with a floor have more leverage to refuse bad bargains. Caregivers with a floor have recognition without surveillance. Families with a floor have room to plan rather than just react. Democracy functions better when the people it is supposed to represent are not operating in permanent survival mode.

The global brain we have built does not need to be worshiped or feared. It needs a policy environment that ensures its gains are broadly shared. It needs the floor we refuse to build.

The most obvious starting point is also the most direct one: a floor under every person.