The AI job-apocalypse panic is the lump-of-labor fallacy with updated branding
The "permanent underclass" narrative rests on a single hidden premise: that there is a fixed amount of useful work to be done. If work is fixed, then AI doing more means humans do less — a zero-sum race to the bottom. That premise is the lump-of-labor fallacy, and it has been falsified every time a major productivity technology arrived.
Keynes predicted almost a century ago that automation would produce a 15-hour workweek. He was right that it created a labor surplus. He was wrong that we'd sit back and enjoy it. Instead, we found new productive endeavors to fill our time, as humans always do. The shape of the labor market changed; its size grew.
The claim that AI will produce economy-wide, permanent unemployment isn't new thinking — it's old thinking dressed in ChatGPT branding. Saying "this time it's different" requires much more than frantic handwaving about the collapse of human cognitive advantage. Bad economics and worse history, nothing more.
The lump-of-labor fallacy is seductive because task-substitution is real and visible. The demand expansion it unleashes is diffuse, delayed, and distributed across industries that don't exist yet — which makes it easy to ignore.
Source claim: The AI job-apocalypse narrative is built on the lump-of-labor fallacy — the false premise that the total amount of useful work is fixed, making automation zero-sum.