The Economic Scenarios of Transformative AI: A Wide Range of Possibilities for Jobs and GDP by 2030

​According to new research and economic modeling from one of the world's leading frontier AI developers, the scale of AI's impact on GDP and jobs in just the next three to four years is still very uncertain.

The new working paper from The Anthropic Institute—"Economic Scenarios of Transformative AI" authored by University of Virginia Darden School of Business Professor Anton Korinek, Stanford University and The Anthropic Institute economist Charles "Chad" Jones, Anthropic Chief Economist Peter McCrory and Anthropic economist Szymon Sacher—presents a framework for assessing the economic consequences of AI between 2026 and 2030. It models three potential paths—modest, substantial and extreme change scenarios—of AI capabilities into implied paths for GDP, the labor share, wages, labor reallocation and unemployment.

The extreme change scenario includes some eye-watering predictions. Transformative AI would send one in five knowledge workers into unemployment by the end of the decade while GDP growth reaches an astounding 15%.

However, the impact of the modest change scenario? Not so sensational. The researchers found AI would drive a 0.1% increase in overall unemployment and lift GDP by 0.5% annually by 2030.

In the substantial change scenario, AI transformation leads to a 4% increase in knowledge worker unemployment and 8% annual GDP growth by 2030.

The Anthropic Institute released a scenario explorer to visualize findings across the economic model's full range of predicted outcomes.

"AI may make the economy far more productive while changing much of today's work, unevenly across occupations," Korinek said. "Yet estimates span orders of magnitude and are hard to compare."

In a thread on X describing the work and why the findings matter, Korinek highlighted the wide range of uncertainty that remains about how AI could transform the economy in the coming years, as seen in GDP predictions under the three scenarios.

Knowledge Workers Face the Reality of Finding New Work

The researchers found that GDP growth and labor disruption rise together in correlation with the scale of AI transformation.

Unemployment in the extreme change scenario is expected to surge to 11.9% in 2030, compared to an expected 3.8% in the no-AI base case. Knowledge workers experience the brunt of the pain, with 17.9% unemployment by 2030 in the extreme scenario—reflecting a 22% decrease in employment from 2026.

In the modest change scenario, though, unemployment only ticks up 0.1% to 3.9% for all workers and remains unchanged at 2.9% from the no-AI scenario for knowledge workers.

Korinek predicted it will take time for knowledge workers to find work in other occupations, but that the value created by AI transformations could be used to help those workers through the transition.

"The hard questions concern the transition, not growth," he said. "In the extreme scenario, the aggregate gain is around three times what cognitive workers lose. A transfer could compensate cognitive workers for their income losses. But growth alone won't get it there; policy is needed."

Interestingly, Korinek noted that not all labor disruptions are negative. While knowledge-worker employment and wages fall in every AI transformation scenario, average wages rise overall.

In the extreme change scenario, the average worker's wage rises nearly 10% by 2030 above the no-AI base case—even as knowledge worker wages fall 11.5%. For non-knowledge workers, wages rise 34% above the no-AI path.

Even in the modest change case, average wages rise 0.7% above a no-AI path, while knowledge worker wages fall 0.4%.

More AI automation could dramatically change where income is created. Currently, about 60% of income is generated by labor. However, AI automation would increase the demand for capital, increasing the amount of income generated by capital returns. In the extreme change scenario, the share of income generated by labor falls to 45%.

What Do the Rest of Us Think Will Happen?

In addition to findings from their model, Korinek and his co-authors released findings from a survey of nearly 11,000 U.S. adults measuring their beliefs in how AI will transform the economy.

Much like the researchers' model, views varied widely. However, median outcomes from the survey mapped closely to the researchers' substantial change scenario. The median survey respondent predicted AI transformation would drive GDP 8.6% higher than a no-AI path and result in 4.6% unemployment by 2030.

Read the full technical report "Economic Scenarios for Transformative AI" (Korinek et al., 2026) or use The Anthropic Institute's tool to visualize and explore the findings of the economic model.