JULY 11, 2026
AI Is Boosting Individual Worker Productivity but Has Yet to Transform Company-Level Output
Research shows AI has measurably increased individual worker productivity in narrow domains such as call centers, software development, and legal work, but firm-level productivity gains remain rare and uneven. Most companies have deployed AI in at least one business function but the majority remain in pilot stages, with only a minority reporting measurable bottom-line impact. Economists have found the largest productivity lifts among the lowest-skilled workers, while evidence of economy-wide transformation has not yet materialized.
Artificial intelligence is improving individual worker performance in specific, narrow categories of work, but a broader economic transformation has not yet arrived, according to a commentary by Will Rinehart, a senior fellow at the American Enterprise Institute, published in the Washington Examiner. Drawing on historical analogies and recent economic research, Rinehart described AI as being in its "Wheat Belt stage" — a period of narrow, early adoption before the technology becomes general enough to reshape the wider economy.
The analogy references Harvard Business School economist Daniel Gross's research on tractor diffusion, which found that tractors first spread through the Wheat Belt in the late 1910s but took another two decades to reach the Corn Belt, where conditions and farm organization had to be restructured to accommodate the technology. Rinehart applied the same framework to AI, arguing the technology's current impact is real but limited in scope.
The most-cited research on AI's workplace effects comes from economists Erik Brynjolfsson, Danielle Li, and Lindsey Raymond, who studied a modern AI system rolled out to call center workers. Their study found average worker productivity rose 14% as measured by issues resolved per hour, but the lowest-skilled agents saw a 34% increase — far exceeding gains at the top. Similar skill-equalization effects have since been documented in software development, legal work, and writing, according to the piece.