Sayash Kapoor and Arvind Narayanan respond to Derek Thompson's argument that AI's emergent risks justify extraordinary government intervention. They define extraordinary interventions as precautionary, liberty-restricting, and bypassing normal governance processes, then argue these carry high costs and limited effectiveness — especially since AI lacks a physical bottleneck like nuclear weapons. Instead, they advocate for building societal resilience: improving cybersecurity infrastructure, biosecurity screening, bug bounties, and red-teaming across vulnerable sectors. They acknowledge that resilience requires competent, polycentric governance that the U.S. federal government currently struggles to deliver, but argue that relying on nonproliferation as the primary defense creates fragility. The piece calls for investing in normal policymaking processes rather than escalating unilateral executive actions.
Table of contents
A Recap of AINT’s Arguments and Thompson’s Areas of Agreement and DisagreementExtraordinary Government Action is Extraordinarily CostlyResilience is an Under-Emphasized Defense Against AI RisksIf Resilience is So Helpful, Why Haven’t We Prioritized it Already?287 Impressions1 Comment