AI Is Not Improving Productivity: Nobel Laureate Daron Acemoglu

The economist who put a number on AI’s productivity gain explains how he got it, and why it is so much smaller than the industry’s. His method is to count which tasks the technology can actually take over and what each saves, rather than reasoning down from how capable the models look.

The distinction he keeps returning to is between automation and complementarity. Automation removes tasks from workers, which is good for whoever owns the capital and not by itself good for anyone else; complementary technology creates new tasks people can do. He argues AGI as an ambition sits tightly on the automation side, and that which of the two a technology becomes is decided by the people building it rather than by the technology.