The fear that AI will bring mass unemployment is, according to Philippe Aghion (Collège de France, LSE), Antonin Bergeaud (HEC Paris), Simon Bunel (ENS) and Alexandra Roulet (INSEAD), built on a flawed method: studies that predict job losses typically break existing occupations into tasks and count how many AI could perform, ignoring that workers reallocate their time to the tasks that remain, that jobs are bundles which resist neat unbundling, and that productivity gains let firms expand and hire rather than simply cut headcount.
We show the main point publicly. Create a free account to continue reading the full article, save it, discuss it, and connect it with market and OSINT context.