Economic policy for AGI

AI capabilities have increased tremendously in recent years. A future economy infused with artificial general intelligence—systems capable of performing cognitive tasks at least at the level of the average person—could bring remarkable gains in health, leisure, and living standards. It could also outpace society’s ability to adapt.

The question is not whether progress should continue. It is whether public institutions can prepare flexible, evidence-led responses before disruption becomes acute. This framework compares common interventions across welfare, agency, feasibility, and durability in very different economic futures.

11 policies we evaluated

Taxonomy and operational scope of household-facing interventions considered in this paper.

Policy intervention	Definition & operational scope
Active labour-market policies	Reskill displaced workers and align human capital with AI-complementary work.
Wage insurance	Temporarily replace a portion of income lost after re-employment at a lower wage.
Earned Income Tax Credit	A refundable, means-tested wage subsidy for low- and moderate-income workers.
Federal jobs guarantee	Provide a publicly funded role to anyone ready and willing to work.
Unemployment insurance	Offer temporary partial wage replacement during a period of displacement.
Negative income tax	Guarantee a minimum income through transfers that taper as earnings rise.
Universal basic income	Make an unconditional cash payment to every citizen.
Sovereign AI dividend	Distribute returns from public ownership or taxation of AI-related assets.
Universal basic capital	Give people a direct, compounding stake in publicly managed productive assets.
Universal basic services	Provide essential services such as healthcare, housing, and transport.
Industrial policy	Target investment toward public goods and AI-complemented employment.

Welfare and agency

Policy should do more than maintain consumption. It should protect people’s ability to shape their lives, participate in their communities, and share in the returns from a more productive economy. Work has social and psychological value, so the strongest options maintain room for contribution alongside a resilient floor.

Scenario 1

Mild disruption

Expand unemployment insurance, EITC, and employer-led retraining.

Scenario 2

Broad displacement

Evolve EITC toward a negative income tax when wage losses persist.

Scenario 3

Structural change

Prepare universal basic capital as a durable ownership backstop.

Agency and democratic empowerment

Direct ownership matters when the gains from automation compound in capital rather than wages. Universal basic capital scores strongly on ownership of gains; wage-linked supports and retraining preserve meaningful routes into economic participation. A credible approach should make these trade-offs explicit rather than hiding them behind a single average score.

Welfare and resilience

Negative income tax, EITC, and unemployment insurance offer a practical sequence: they stabilise household income while keeping support responsive to changing labour-market conditions. This makes them more useful than a static intervention deployed before the scale of disruption is clear.

Some interventions are more feasible than others

A policy is not useful if it cannot be administered at speed. Existing programs—especially wage subsidies and unemployment insurance—have institutional scaffolding already in place. More ambitious ownership models deserve design work now, even if their deployment should wait for clearer economic signals.

Durability across scenarios

Durable policy performs reasonably well under mild automation, broad displacement, and profound transformation. Universal basic services and basic capital are most robust across the full range, while wage insurance and jobs guarantees are better suited to narrower transition periods.

Building positive AGI economic futures

The least-regret strategy is sequenced and measurable: build better labour-market data, pre-design automatic triggers, strengthen stabilisers for early disruption, and prepare ownership mechanisms for a future in which labour income no longer captures a fair share of growth.

The central question is ultimately democratic: if AI creates extraordinary wealth, how do people retain security, agency, and a real stake in what comes next?