When America's Economy Meets AI
Washington DC, June 11, 2026 — The Brookings Center on Regulation and Markets, The Foundation for American Innovation, and the Peterson Institute for International Economics
Our Scenarios Program convenes economists, AI researchers, policymakers and civil society leaders to explore the implications of transformative AI, and to identify the policy responses each future demands. The workshops are complemented by the Policy Atlas—a continuously updated guide to policy options.
We believe that countries should develop Economic Preparedness Plans. Our workshops and the Atlas are designed to provide the foundation for doing so.
On June 11, 2026 around 50 senior leaders gathered in DC to explore the “Paper Prosperity” scenario. The exercise dropped everyone into a world where AI diffuses rapidly through white-collar work, but the gains concentrate in capital rather than wages. Frontier systems handle complex cognitive work on their own. Unemployment barely moved, yet underemployment jumped from 8% to 14%. Millions of college-educated workers were pushed into gig work or roles beneath their training. Growth runs along a K-shaped curve.
The money pouring into more capable AI dwarfs the effort spent working out what its arrival does to workers, public finances, and the bargain that holds a society together. Closing that imbalance was the point of the Washington convening, which took transformative AI's arrival as a given and tested something narrower and more practical: how the economy and society would cope when it did.
By 2030, America looks better than ever—on paper. GDP growth is strong, equity markets are soaring, and productivity is the envy of the developed world—and underneath it, a middle class is sliding quietly into precarity. That was the future economists, AI researchers, and senior policymakers were asked to inhabit when they gathered at the Peterson Institute for International Economics on 11 June 2026. The question for the room: if this is where current trajectories lead, has the country done anything to prepare?
Scene Setting
Two presentations set the terms before anyone broke into groups.
An opening briefing from Anthropic put a number on the pace of change: the U.S. is now spending a larger share of GDP on data centres than it spent on railroads at the height of the Industrial Revolution. Models that could do little useful work a few years ago can now complete tasks that would take a software engineer roughly three hours, and performance on biology benchmarks has gone from around 20% accuracy in 2024 to near-saturation today, with models moving swiftly towards recursive self-improvement.
Economist Anton Korinek offered a longer arc: across history, the binding constraint on growth has moved from land, to labour, with AI potentially making labour dispensable and opening scenarios of growth without wage growth.
Together, the two presentations framed the central tension the breakout sessions would explore: unprecedented productive capacity arriving faster than institutions, labour markets, or politics can absorb it.
What we saw
Three observations stood out:
The promise that work leads to advancement is breaking down. The scenario describes a world where AI commoditises expertise, and education stops reliably buying middle-class security. Participants worried less about headline unemployment than about the fraying of the social contract as effort and education are no longer levers to upward mobility. One group pointed to recent electoral results in U.S. cities as early instances of the politics this produces. They expected generational divides to widen and birth rates to fall as young people lose confidence in the future—a dynamic that, unlike tax or institutional reform, plays out on a far shorter timeframe.
The country's strongest sectors become its fiscal liabilities. Professional services, finance, and high-end knowledge work earn America the most today, and the scenario hits them hardest. As the value AI creates pools among a few firms and the people who own them, the revenue that used to come from wages starts to drain away—shrinking the tax base just as government needs more capacity to respond, not less.
Government capacity is the binding constraint. A recurring worry was whether the state could deliver policy change at the scale and pace required. Polarisation, a narrow Overton window, and public distrust of institutions kept surfacing as the real obstacles—alongside a blunt admission: there isn't yet real alignment across stakeholders on what to do, even if more convergence is forming on adjacent issues like energy and jobs.
Policy Responses Worth Exploring
The group also explored what we can do about it, where ideas such as reskilling, universal basic income, sovereign wealth funds, and tax reform were all discussed as potential solutions to explore. However, the breakout discussions pushed groups beyond those defaults and surfaced other ideas worth exploring, including:
Universal capital ownership. If wages permanently capture less of national income, tax-and-transfer alone won't keep pace. The fix several groups reached for was to give people a piece of the capital doing the work: public equity stakes in AI firms, investment accounts open to everyone, or employee ownership schemes—currently rare enough (an estimated 2% of U.S. firms are ESOPs) that scaling them would be its own undertaking. One group raised SAG-AFTRA as a model for what knowledge workers might need next: collective-bargaining infrastructure built for an industry that didn't used to think it needed one. The reasoning was as much political as financial—owning a share of the upside changes how people live through a transition in a way a transfer cheque will not.
Revaluing human-centred work. Borrowing from the pandemic's “key workers,” groups argued for paying care, health, and education closer to their social worth than their market price, and for subsidising scarce human sectors directly. One participant coined the phrase “new paternalism”: the state subsidising pro-social behaviour to counter the isolation and falling fertility the scenario implies.
Fiscal reimagination. With labour income shrinking, groups looked to shift tax toward capital and immovable assets, and floated novel mechanisms: governments owning and renting out compute, a per-capita corporate tax that rewards firms for employing people, a displacement tax on firms with unusually high revenue-per-employee ratios, and an “automation-adjusted” insurance product modelled on existing wage insurance. A GI Bill-style program—funding mid-career retraining directly through the tax system, rather than as a separate ask—came up as the version of reskilling this room found more credible than the standard pitch. The common thread was proposals to tax non-productive accumulation instead of workers' dwindling pay.
Building the state's capacity to act. For one table, the first priority was “policy plumbing:” improving real-time data flows from private firms to government, a CFIUS-style interagency hub to break departmental silos for AI economic preparedness, and recruitment of technical talent from industry. Two ideas aimed at the same data gap: an AI Preparedness Index that would press companies with a stake in an “AI is good” narrative to publish real usage data, tracked state by state; and “circuit breakers”—unemployment thresholds that trigger policy responses automatically, so support doesn't wait on fresh legislation.
Levelling with the public. Rebuilding public trust—through clear communication and reforms to cut waste and money in politics—was treated as a precondition for everything else, echoing the scenario's demand for good jobs over handouts.
Unresolved Tensions
The day also exposed questions that an effective policy response would need to settle:
Do the decision-makers actually believe any of this? This question sat beneath many discussions. Most governments behave as though transformative AI is not close or not truly transformative, yet every response on the table assumes action well before the shock lands. Even leaders who accept the premise have to weigh it against live conflicts and a cost-of-living squeeze that favour short-term decision making.
What is strong growth worth if daily costs keep climbing? The scenario assumes brisk GDP growth, but participants doubted it would reach the public essentials that decide whether people feel better off—housing, healthcare, or education. This was noted as a particular challenge in the U.S. where public services are thin compared to European economies.
Is the future only disruption? Several pushed back on the framing. Some argued it underweighted the upside, from health breakthroughs to falling costs in some sectors; others doubted its realism, since AI's measured economic footprint still lagged what a genuine take-off would imply. The shared instinct was that policy needs to actively shape outcomes, rather than wait and respond after the fact.
Where This Leaves Us
Despite the diversity of perspectives in the room, there was surprising agreement on the core dangers, the most promising interventions, and the structural reforms required. As one participant observed, the economic challenges may be more tractable than the political ones. The room found it easier to agree on what to do than on whether the political system can deliver it.
A few participants compared the moment to the eve of the Industrial Revolution: a brief window to steer the outcome rather than inherit it. With the scenario set in 2030 and serious legislation taking years to draft, the case for starting now—by design, not emergency—was the one idea few contested. Whether the government can move as fast as this room believes it must is the question that remains open.
If you are working on these questions and want to be part of the conversation, contact us at contact@windfalltrust.org, or follow the work through our newsletter, the AGI Social Contract, at windfalltrust.substack.com.
