AI Designed to Aid Human Workers

A coalition of more than 200 economists and AI researchers, including 16 Nobel laureates, has warned that AI should complement humans rather than imitate them, arguing that the technology’s greatest economic benefit comes from expanding the set of activities people can pursue.

AI can reshape work.

Why augmentation matters more than substitution

The signed “We Must Act Now” statement stresses that AI that merely replaces human labor narrows the economic pie, while tools that broaden access and lower costs can grow it. When AI enlarges the set of things worth doing, it creates new markets; when it simply transfers existing tasks to models, it merely reshuffles value.

The document cites historical examples such as the loom, tractor and personal computer, which were all expected to cause mass unemployment but instead lowered production costs and spurred job growth.

Erik Brynjolfsson, a co‑organizer of the statement, labels the mistaken belief that a fixed amount of work exists the “Turing trap.” He argues that designing AI to imitate humans steers development toward substitution, concentrating power among owners of the technology and limiting opportunities for ordinary workers. By contrast, augmenting human capabilities can make the pie larger, even if it requires more complex business cases.

Historical evidence of complementarity

James Bessen’s research on automated teller machines illustrates how automation can expand employment. ATMs cut the number of tellers per branch from about 20 to 13, but the lower operating cost led banks to open more branches, ultimately increasing the total number of tellers for decades. The job shifted from cash handling to customer advising, showing that a technology once thought to eliminate a profession instead broadened it.

A 2019 paper by Daron Acemoglu and Pascual Restrepo provides a framework for this dynamic. They note that while automation moves some tasks from workers to machines, higher productivity and the creation of entirely new tasks can offset those losses. Many occupations that have driven employment growth in recent decades were virtually nonexistent a generation earlier. Their analysis also warns of “so‑so automation,” where technologies displace workers without generating enough demand to offset the loss, shrinking labor’s share of the economy.

When AI lowers the cost of cognition itself, the potential for expansion is even larger. Reducing the price of legal, medical, pedagogical and scientific expertise could allow specialists to serve far more people, increasing the value of judgment, taste and accountability—areas where AI still falls short. The argument is that as raw computational power becomes abundant, uniquely human contributions become more valuable.