The Post-AGI Economy
If cognitive labor becomes cheap and abundant, the foundations of modern economies — wages, property, comparative advantage, the social contract — strain in ways economists are just beginning to model.
Key facts
- Studies (Eloundou et al. 2023; OECD 2024) estimate 30–70% of US/OECD tasks have meaningful AI exposure.
- Compute spending is now the dominant driver of global data-center capex.
- UBI pilots in multiple countries show small or no labor-supply effects.
- The CHIPS Act commits $52B+ to US semiconductor capacity.
- Acemoglu (2024) projects 0.5–1.0% TFP gains from AI over 10 years — far below the most optimistic estimates.
Wages in a World of Abundant Cognition
When AI substitutes for cognitive labor, the marginal product of human cognition falls. Wages either compress, polarize, or are sustained by tasks AI cannot do — physical, relational, accountable, embodied, or status-laden work.
Daron Acemoglu and Pascual Restrepo's task framework, and David Autor's work on the polarization of US wages, provide the dominant theoretical lens.
Capital, Compute, and Power
Compute, data, and energy become the bottleneck inputs of advanced economies. Concentration of these inputs concentrates economic power, with serious implications for competition policy, antitrust, and industrial policy.
The CHIPS Act, EU Chips Act, and UK / Japanese sovereign-compute programs reflect a global shift toward treating compute as strategic infrastructure.
Universal Basic Income and Distribution
UBI, social wealth funds, negative income taxes, and 'AI dividend' proposals address distribution but not meaning. Economists differ sharply on transition design.
Pilots in Finland, Kenya (GiveDirectly), Stockton, and Sam Altman / OpenResearch's US study (2024) show mixed but generally positive effects on wellbeing without large labor-supply collapses.
Growth, Stagnation, and Productivity
Optimistic models (Davidson, Open Philanthropy; Tom Cotton & Tyler Cowen debate) suggest sustained AI capability could produce growth rates not seen since industrialization.
Skeptics (Robert Gordon, Daron Acemoglu) note that headline productivity gains from prior IT waves were narrower than predicted, and that diffusion takes decades.
Comparative Advantage and Globalization
Ricardian comparative advantage assumes humans are the producers. AI alters which countries hold advantage in which sectors, with major implications for trade, development, and offshoring.
Service offshoring — previously a development lever for India, the Philippines, and others — is the category most directly exposed.
Frequently asked
Will AI cause mass unemployment?
+
Possible but not assured. Historically, automation reshapes work more than it eliminates it — but the speed of AI may overwhelm normal adjustment mechanisms in specific sectors.
Is UBI a serious policy option?
+
Yes — pilots have produced enough evidence to take it seriously, though full implementation faces fiscal and political design questions.
Will inequality grow?
+
Default expectation: yes, without active redistribution policy. AI accrues gains disproportionately to capital owners and high-skill complements.
Sources & further reading
Continue in this series
Biology
AI, Biotech, and Longevity
Philosophy
Transhumanism and Human Enhancement
Long-termism
AI, Space, and the Long-Term Future
Human Condition
Meaning, Purpose, and Identity in an AI Era
Scenarios
Scenarios for the Next Century