Strategic dossier · 2026
AI, productivity and power
AI can substantially increase productivity in some tasks. The decisive economic and social question is how gains are distributed among workers, firms, platforms and institutions.
Who captures value when intelligence becomes economic infrastructure?
METHOD
- Separate task automation from job disappearance.
- Study complementarity between people and tools.
- Examine ownership of models, data, capital and market power.
FINDINGS
Tasks
Jobs are bundles of tasks: some will be automated while others are augmented.
Capital
Gains may concentrate among actors controlling compute, data, distribution or intellectual property.
Skills
Value shifts toward judgment, orchestration, responsibility and contextual understanding.
Institutions
Taxation, labor law, competition and training influence the final distribution of gains.
STARTING BIBLIOGRAPHY
— Daron Acemoglu & Simon Johnson — Power and Progress (2023)
— Erik Brynjolfsson & Andrew McAfee — The Second Machine Age (2014)
— ILO — Generative AI and Jobs, analytical reports