AI Drives Early Productivity Gains Across Industries

A working paper released August 10, 2026 by the National Bureau of Economic Research reports that advances in artificial intelligence have already produced early productivity gains across various industries in the United States. The authors find that these initial benefits are concentrated among capital owners before diffusing throughout society over several generations. ¹

The findings echo patterns observed during previous technological revolutions—such as the introduction of electricity and internal combustion engines—whereby early productivity increases were largely confined to those who owned the technology or capital assets that could exploit it most effectively. This historical parallel suggests that AI’s current trajectory will similarly see its first gains accrue to firms and investors before trickling down to workers and consumers. ¹

To manage this diffusion process, the paper calls for institutional changes, including new regulatory frameworks around AI deployment standards that mirror how we currently regulate autonomous vehicles today. Such oversight would help ensure that AI is deployed responsibly while also creating mechanisms to monitor and adjust the distribution of its economic benefits. ¹

A key challenge highlighted by the researchers is preventing early AI‑driven productivity from exacerbating existing wealth inequalities. They propose proactive measures toward equitable distribution through policies supporting worker retraining programs funded via taxes on increased corporate profits directly attributable to AI adoption rates exceeding 5%. By channeling a portion of the additional earnings generated by AI into reskilling initiatives, the study argues that societies can mitigate inequality risks while still reaping the efficiency gains of the technology. ¹