Research

Finds that approximately 80% of the U.S. workforce could have at least 10% of their tasks affected by LLMs, with around 19% of workers seeing at least 50% of their tasks impacted.

Examines why AI hasn't yet shown up in aggregate productivity statistics despite rapid advances, and argues the lag is consistent with historical patterns of general-purpose technology adoption.

A framework for thinking about how AI will affect output, wages, and inequality. Argues that the economic effects of AI will be more modest than many predict unless the technology moves beyond task automation.

Develops a framework showing that automation displaces workers from existing tasks but can be counterbalanced by the creation of new tasks. The balance between these forces determines labor's share of income.

The landmark study estimating that 47% of U.S. jobs are at high risk of automation. While the timeline has been debated, the framework for assessing job vulnerability remains influential.

Documents the significant decline in labor's share of national income since 2000, exploring globalization, technological change, and institutional factors as potential explanations.

Argues that the decline in labor's share is driven by industry concentration, as "superstar firms" with high productivity and low labor shares capture increasing market share.