Anthropic Institute Models AI Economic Scenarios Through 2030

Researchers Anton Korinek, Charles I. Jones, Szymon Sacher, Tess Cotter, and Peter McCrory published The Anthropic Institute working paper outlining economic models for AI between 2026 and 2030. The paper, designated Working Paper No. 2026-02, was published in September 2026.

The framework converts task-based parameters into paths for GDP, wages, labor share, labor reallocation, and unemployment. It evaluates three primary cases: modest, substantial, and extreme change.

In the modest change scenario, AI adds less than half a point to GDP growth by 2030 and raises unemployment by a tenth of a point. By 2030, GDP is 1.6 percent higher than it would be without AI.

The substantial change scenario maps a path where GDP rises by 8 percent and cognitive employment declines by 4 percent by 2030. In this scenario, GDP reaches a level 8.3 percent higher than the baseline without AI by 2030.

The extreme change scenario projects AI performing almost half of today's cognitive work by 2030. Annual GDP growth rises to 15 percent per year, making total GDP 32 percent higher by 2030. Under this scenario, nearly one in five cognitive workers is unemployed, and the labor share of income falls from 60 percent today to 45 percent in 2030. Cognitive wages drop 11.5 percent below their no-AI path, while wages in all other occupations end up 34 percent above it. Across all three scenarios, almost all of the divergence occurs after 2027.

To complement these scenarios, the authors surveyed 10,980 US adults regarding eight cognitive tasks of increasing difficulty, asking about usage rates, productivity gains, automation versus augmentation, and re-employment timelines. The median survey responses generated economic outputs close to the substantial change scenario.

The authors stated that Claude was used as a research and writing assistant for the paper.