Organizing Intelligence
Imas et al.
A model of firms in which humans and AI sort endogenously finds that better AI coordination can break the limit on firm size and produce a phase transition towards concentrated, AI-heavy firms.
Source published · In the Daily Brief of 9 November 2026
Filed under Economy, AI industry
Alex Imas, Andrew Koh and Shane Legg model agents, human or AI, with multidimensional capabilities who sort endogenously into competing firms. Productivity depends on competence, which shifts its level, and coordination, which governs how it scales with firm size. AI expertise is reproducible and may not face human limits on coordination.
When AI is more capable than the average human, organisations of complementary humans can still outperform AI. Improvements in AI competence can shrink firms and change their composition. Improvements in AI coordination can break what the authors call the Coasean ceiling on firm size, and the market undergoes a phase transition towards large, concentrated, AI-heavy firms. Wage changes split into a technology effect and a reorganisation effect, and the second can reverse the first.