DeepMind Essay: AGI Economic Impact Depends on Organizational Coordination, Not Individual Capability

Details

Economic transformation from AGI is determined by the competition between AI organizations and human organizations, not by individual agent benchmarks.
Human teams achieve 'mosaic intelligence' through complementary experts, yielding 32% higher productivity than random pairs in experimental domains.
Four regimes define the transition: Minimal AGI (wage convergence to compute costs), Supercompetent AI (lean firms, abundance without concentration), Supercooperative AI (phase transition to large AI firms), and ASI (commoditization of goods, shift to relational sectors).
Current AI exhibits exponential competence growth but lags in coordination, with multi-agent contributions estimated at less than 10% of single-agent value.
The 'Supercooperative' threshold is the critical tipping point where adding agents increases team productivity, collapsing the Coasean ceiling and allowing AI firms to dominate deep markets.
Wage effects precede structural reorganization, meaning current employment statistics underestimate AI's economic impact.
Future labor markets will see a shift toward the 'relational sector' (healthcare, education) where human interaction remains valuable, while routine expertise becomes commoditized.

Key point