Fiscal policy will need new tools for uncertain AI future | Harvard Kennedy School

TEXT START: Artificial intelligence could reshape the U.S. economy in profound ways, but scholars, business leaders, and the public aren’t yet sure how.

The Dissection

This is a fiscal risk-management memo disguised as open-ended foresight. It acknowledges productivity gains, job displacement, capital concentration, unemployment, mortality, mental-health damage, and loss of purpose—but translates the rupture into a menu of taxes, benefits, retraining, and public investment.

The text’s real function is to make systemic discontinuity administratively legible without accepting its terminal implication. It imagines the state managing AI’s consequences while the underlying economic order remains intact.

The Core Fallacy

The article treats AI disruption as a distribution problem. Under the Discontinuity Thesis, it is a participation problem first.

Transfers can preserve consumption. They cannot restore the mass employment → wage → consumption circuit once AI makes human cognitive labor economically unnecessary. Retraining assumes displaced workers can be reabsorbed into new productive roles; P1–P3 imply that automation and competitive pressure progressively destroy that assumption.

The fourth scenario is therefore not sufficiently disruptive. It describes high productivity, permanent unemployment, labor exit, and capital capture, but still frames the outcome as a difficult fiscal environment. The deeper event is the collapse of productive necessity for the majority and the conversion of citizenship into dependency on whoever controls AI capital.

A government investment fund or individual equity accounts may distribute claims on output. Unless they confer actual control over AI systems and their bottlenecks, they create beneficiaries, not Sovereigns. Ownership theater is not sovereignty.

Hidden Assumptions

• Productivity gains remain large enough, and taxable enough, to finance expanded transfers.
• The state can act before capital owners become entrenched political power.
• Displacement is mainly a transition problem rather than permanent exclusion.
• Retraining produces economically necessary work after cognitive automation dominates.
• Fiscal policy can preserve social stability even when work no longer supplies status, purpose, or bargaining power.
• Public equity purchases produce meaningful control rather than passive financial exposure.
• National governments can tax mobile capital and govern AI infrastructure within national borders.
• Policy institutions will coordinate effectively despite the article’s own warning that policy lags behind understanding.
• Energy, logistics, maintenance, and other physical bottlenecks remain politically manageable rather than becoming the new ownership battlefield.

Social Function

Primary classification: transition management.

Secondary classifications: partial truth and ideological anesthetic.

The article is not pure propaganda. Its recognition of inequality, joblessness, policy lag, and nonfinancial damage is materially correct. But its policy menu domesticates the threat. Retraining, modest assistance, and “insurance” policies let institutions rehearse administering the fallout while avoiding the question of who owns and controls the productive machine.

It is a lullaby for policymakers who understand that the labor market may fail but still want to believe the budget can absorb the failure. Fiscal policy is being asked to bandage a severed circuit.

The Verdict

The article correctly identifies the symptoms and misidentifies the disease. Its fourth scenario is the nearest approach to the Discontinuity Thesis, but it stops at unemployment, inequality, and debt management. The actual terminal condition is the collapse of mass productive participation and the death of post-WWII capitalism’s organizing mechanism.

Fiscal tools can delay social death, redistribute the carcass, and manage the transition. They cannot preserve the old system. Viability migrates to Sovereigns who control AI capital, Servitors who remain indispensable to its bottlenecks, and operators of energy, logistics, maintenance, verification, and transition intermediation. Everyone else is being offered consumption without power.