Economic Policy for AGI
3 hours ago
- AGI could bring huge economic benefits and disruptions, but past technological transitions have often hurt many people in the short term.
- AGI is different from prior technologies because it automates intelligence itself, making economic impacts uncertain and potentially more comprehensive.
- Policy should be flexible and tied to empirical triggers to avoid acting too late or overreacting.
- Three challenges: lack of granular data, determining policies that distribute benefits widely, and lack of a common rubric to evaluate interventions.
- A unified framework evaluates policies across welfare/resilience, agency/voice, feasibility/efficiency, and durability across scenarios.
- Using literature reviews, surveys, and 51 AI economist personas, 11 policies were assessed.
- Least-regret policies: expanded UI, EITC, and employer-led retraining for mild disruption.
- For moderate displacement, transform EITC into a Negative Income Tax (NIT).
- For structural labor-capital decoupling, use Universal Basic Capital (UBC) as a backstop.
- UBI is expensive, blunt, may not concentrate relief, and may fail to support economic agency.
- UBC grants direct ownership stakes so wealth compounds with the economy, unlike sovereign wealth funds or UBI.
- Feasibility varies: retraining and UI are popular; UBC faces political headwinds and institutional challenges.
- Universal Basic Services score highest on durability, while NIT and UBC also perform well across scenarios.
- Policies should be sequenced: expand UI and EITC now, transition to NIT if displacement deepens, and pre-design UBC for extreme scenarios.
- Improving data collection on AI's economic impacts is critical for timely and effective policy responses.
- Society has the tools to shape the AGI economic trajectory and ensure wide access to growth benefits.