AI's Labor Crisis: Preparedness Theater or Real Policy?

Overview

In the last six months, leading artificial intelligence laboratories have shifted from building models to publishing economic policy papers. They are grappling with a singular, daunting question: what happens if their technology sharply reduces the demand for human labor?

Anthropic recently outlined a framework suggesting that AI could create unprecedented abundance while acting as a general substitute for labor. Its June paper sorts the future into three tiers of rising unemployment, with the final tier exceeding anything in the historical record. To support this analysis, the company announced a $200 million Economic Futures Research Fund and a $150 million national fellowship program.

However, critics argue that this elaborate apparatus amounts to preparedness theater. While the labs publicly build institutions to demonstrate they take the problem seriously, they largely refuse to answer the most fundamental accounting question: how large should the public’s share of AI-generated wealth actually be?

The Architecture of Preparedness

Anthropic’s framework pairs specific unemployment thresholds with tailored policy responses, creating a escalating ladder of intervention. At roughly 5 percent unemployment, the proposed measures are designed to keep people employed through wage insurance, training grants, job matching, and licensing reform.

If unemployment rises to approximately 10 percent, the response extends unemployment insurance and adds monthly payments for those whose benefits have expired. It is only at the third tier, representing catastrophic job loss, that the framework turns to structural economic changes such as new taxes, a possible basic income, sovereign wealth funds, and equity sharing.

"At the last tier, Anthropic says it has moved beyond what economists know how to map, and that it is 'not yet ready to advocate specific policies for this scenario.'"

This hesitation is significant. Each step up the ladder makes the question of who receives the gains more urgent, yet the policy proposals become less concrete. A natural follow-up question is how large the public stake should be in the wealth these systems create. This is not an exotic inquiry, but Anthropic never answers it, nor do most other institutions engaging in this debate.

An analysis of seventeen AI economic-policy papers and statements that address distribution found that fourteen never specify how large the public’s share should be. Only three attempt to put a number on it. This suggests that while the institutional machinery is being built, the core fiscal settlement remains deliberately vague.

The Simplicity of Distribution

The complexity surrounding AI’s impact on labor may be manufactured. Roughly half the population does not work for pay at any given time. Most nonworkers are children, students, unpaid caregivers, the unemployed, disabled people, or retirees. These are stages nearly everyone passes through, yet they still require income to buy food, pay for housing, and live.

This is the fundamental problem the welfare state was designed to solve. National income is divided between labor and capital, and a portion of that income is routed to people who are not currently earning a wage. Matt Bruenig has long argued that technological displacement does not create a new economic category. The immediate problem remains lost labor income, and his conclusion is aggressively unoriginal: "The policy solution to unemployment remains the same."

If AI leaves many more people without labor income for much longer, the problem simply gets larger. The division between labor and capital is not fixed. For much of the postwar period, labor’s share of nonfarm business output sat between 60 and 66 percent. By the second quarter of 2026, that figure had dropped to 52.8 percent, the lowest reading in the Bureau of Labor Statistics series going back to 1947.

AI did not cause this decline, but the trend demonstrates that the balance shifts over time. If AI substitutes for labor while producing the growth the labs forecast, less national income will be paid to workers and more will be claimed by owners of capital. The resource problem may get easier, but the fight over who receives the gains gets bigger.

Who Owns the Gains?

If labor’s share shrinks and capital’s grows, the transfers that once drew on wages must draw on capital income. The policy literature often treats this shift as background noise rather than the central question, leaving the public claim on capital income for later discussion.

Proposals such as Bruenig’s social-wealth-fund suggest a direct answer: public ownership of capital and a universal dividend from its returns. This would give the public a durable claim on capital income, ensuring that the gains from automation are not entirely concentrated among shareholders.

Anthropic has been specific about the institutional response it wants to build. Its $200 million research fund states a desire for interventions that can "spread the gains broadly," noting that "the most promising solutions are ones nobody has tried yet." A national fellowship program sits beside this fund, alongside the Anthropic Institute, an Economic Index, and an expanding network of advisers.

Since July, former Federal Reserve Chair Ben Bernanke has joined the company’s Long-Term Benefit Trust. While trust members hold no equity in the company, they will contribute to its economic research. Bernanke described the potential of AI as enormous and noted that the range of outcomes depends "in part, on the institutions we build around it."

There is plenty for this apparatus to learn: how to capitalize a public fund, how to prevent capital tax avoidance, and how a public fund should vote its shares. However, the essential question remains unanswered. If the public is to benefit from AI-driven abundance, the policy community must move beyond theatrical preparedness and define exactly what that share looks like.