AI models alone rarely transform organizations; the scarce resource is the institutional capability to integrate models — data pipelines, workflow redesign, evaluation practices, and trust mechanisms. Without those complements, access to powerful models spreads widely but productive use remains concentrated.
— This reframes public and policy debates from model access or capability ceilings to building institutions and governance that shape whether AI produces broad economic gains or concentrated disruption.
Arnold Kling
2026.03.19
100% relevant
Sebastian Galiani’s observation about access vs. effective use and Hal Varian’s emphasis on complements at Google (two-sided auctions, platform economics) illustrate that institutions determine AI’s economic effects.
← Back to All Ideas