Who gets the dividend?

AI can perform fragments of many roles before it can replace an entire profession. That matters: employment can change through smaller teams, altered entry-level hiring, surveillance, and a slower wage ladder — even when official unemployment is steady. The IMF estimates that almost 40% of global employment is exposed to AI; exposure includes possible productivity gains, not only substitution.

A society that becomes more productive can choose shorter hours, better services, safer work, and new opportunities for people previously excluded by barriers of cost or disability. These outcomes are possible, not automatic. Firms, workers, and public institutions negotiate how productivity gains are distributed.

Look for changes in entry-level vacancy rates, wage growth within occupations, hours worked, job quality, business formation and the share of income going to labor. Avoid treating a company’s announced “AI-related” layoffs as proof of a net economy-wide employment effect.

Governments can fund portable retraining linked to real vacancies, modernize unemployment protection for people with irregular incomes, strengthen worker consultation over automated decisions, and measure who captures the gains. Publish disaggregated evidence rather than one headline number.