A new paper from the Brookings Institution suggests that the artificial intelligence revolution could play out like a classic boom-bust cycle, even in wages rather than markets.
First, automation could increase wages as workers become more productive along with smarter tools.
But as AI systems master more tasks, the demand for human workers in those areas may decline, pushing people into lower-value or slower-growing jobs, which could erode initial gains.
Conrad Cording, professor of neuroscience at the University of Pennsylvania’s Institute for Integrative Knowledge (PIK), and Ioana Marinescu, associate professor at the University of Pennsylvania’s School of Social Policy and Practice, found in simulations that “automation in the information sector first increases wages and then decreases them.”
After the initial productivity spike, they write, “the negative effects become predominant as most workers are frozen out of intelligence work.”
Wage boom peaks before bursting
To explain how the AI era will unfold, researchers have developed an interactive model that tracks the transition from human-driven to machine-driven intelligence.
This suggests that wages will initially rise sharply as AI increases productivity, then plateau, and then decline as automation becomes more widespread.
Even the pay slips reveal that output continues to rise, with profits flowing more and more to capital rather than to labor.
As cognitive tasks become increasingly automated, humans are shifted to slower-growing, more physical jobs, from construction to nursing care, and wages are falling.
The result is a humped curve. Wages will experience a temporary boom, followed by a correction as the digital economy outpaces the physical world.
“While automation may initially increase wages, it may ultimately lead to significant wage declines,” the authors write.
The limits of intelligence — and the case for delayed deployment.
This paper rejects the two extremes of the current AI debate: the techno-utopian dream of infinite abundance and the apocalyptic fear of complete job loss.
Instead, Cording and Marinescu propose a middle ground they call “intellectual saturation.” While AI can make economies smarter and more productive, progress will ultimately be slowed by the continued reliance on humans, physical tools, and equipment to get the job done.
To prevent this curve from tipping against workers, the authors suggest slowing the pace of automation and investing in physical capital such as machinery, equipment, and tools so that the human workforce can remain productive even as digital tasks disappear.
They also propose taxing virtual substitutes for in-person services to prevent AI from hollowing out entire industries, a proposal similar to Sen. Bernie Sanders’ call for a “robot tax” on companies that use AI to replace jobs.
