The lump-of-labour fallacy is the assumption that the total amount of work available in an economy is a fixed, finite quantity — like a pie that gets divided into slices. If a machine takes one slice, the thinking goes, a human must go without. It sounds intuitive. It is also consistently, historically wrong. Economists have a name for it precisely because it keeps reappearing, wearing new clothes, with every wave of technological change.
The fallacy breaks down because it treats the economy as static. In reality, technology does not merely redistribute existing tasks — it creates entirely new ones. The printing press did not put scribes out of work permanently; it created an enormous new industry of publishing, editing, and distribution that had never existed before. Spreadsheet software did not eliminate accountants; it made financial analysis so much faster and cheaper that demand for it exploded, and the profession grew. New capability does not consume demand. It generates it.
In AI discussions, the lump-of-labour fallacy underlies most doomsday predictions about mass unemployment. The argument typically runs: AI can now do cognitive tasks that humans used to do, therefore humans will no longer be needed for cognitive work. This treats the set of cognitive tasks that need doing in the world as fixed and fully enumerated — as if we have already thought of everything worth thinking, solved every problem worth solving, and invented every product worth building. That assumption does not survive contact with the actual state of medicine, climate science, education, mental health, or the roughly four billion people on earth still without access to basic financial services.
The practical consequence of recognising the fallacy is not complacency — transitions are real, and they involve genuine disruption for real people. But it reframes the question. Rather than asking how to protect existing jobs from AI, the more useful question is how to ensure that the new jobs AI creates are accessible to people being displaced by it. That is a question of education, policy, and support systems. It is a harder and more interesting question than simply sounding the alarm.
