Could AI reduce worker bargaining power and wages even without producing large increases in unemployment?
Acemoglu-Restrepo show 'so-so automation' can depress wages without mass unemployment. Autor's counter-case: AI could rebuild middle-skill work by extending expertise to more people. Both are live.
View on the map → · Open in Browse →
The wage question separated from the labor-share question: Autor and Kausik show automation can raise wages while cutting labor share, and Autor reframed the risk as devaluation of expertise. On the ground, NY Fed postings data find no distinct AI-driven demand decline — while Smith documents the skill premium compressing inside software.
When labor share exceeds this wage-maximizing level, further automation increases wages even while reducing labor's output share.
Model plus twelve-country data: automation can raise wages while reducing labor's share — the two questions come apart.
The concern is about devaluation of expertise
The main risk is falling returns to accumulated expertise, not job scarcity; proposes wage insurance and universal basic capital.
The evidence from job postings provides little indication of a distinct AI-driven decline in labor demand.
Lightcast postings against exposure scores: declines predate ChatGPT; no junior-senior divergence.
AI coding tools compress the software skill premium the way power looms devalued master weavers — wages and status, not employment.
Asset-manager macro case that scalability plus hyperscaler market power push labor share down without large unemployment.
Argues comparative-advantage defenses of permanent human wages fail once humans become net-negative contributors; the collaboration era lasts a decade or two.