Anthropic Models Three AI Economic Scenarios Through 2030 After Researcher's Exit Warning

Anthropic has released a model showing how artificial intelligence might transform the US economy by 2030, and its most rapid growth scenario projects knowledge worker wages falling over 10%.

The firm's economics unit published the simulation alongside a technical paper titled Economic Scenarios for Transformative AI, inviting users to input their own assumptions.

Three Scenarios With Contrasting Outcomes

The model considers three possibilities, released shortly after researcher Jacob Coxon said he had left Anthropic and criticized the sector for pursuing self-improving superintelligence.

That same capability drives the model's most pessimistic outcome:

• Modest scenario

AI expands on the scale of the internet, lifting gross domestic product (GDP) to $34.1 trillion — a 1.6% increase.

• Substantial scenario

Assumes AI performs half of all knowledge tasks by 2030. GDP reaches $36.3 trillion, roughly double the typical pace, with unemployment near 5%. Knowledge worker wages are unchanged.

• Extreme scenario

Requires AI that improves itself without human intervention. Annual growth hits 15%, GDP climbs to $44.4 trillion, and unemployment surges to record levels. Knowledge worker pay drops more than 10%.

Labor's portion of national income mirrors this pattern. It declines slightly in the modest case, falls to 56.1% in the substantial one, and reaches 45.2% in the extreme.

“In the extreme scenario, the gains from a rapidly expanding economy are unevenly distributed,” read an excerpt in the report.

Americans Favor the Middle Scenario

Anthropic polled 10,980 people in August. The median response implied GDP roughly 10% higher by 2030, aligning with the substantial scenario. About one in 10 expected the extreme.

That measured view matches earlier findings. Anthropic's June survey showed Americans cite job loss as their primary AI concern, while Goldman Sachs identified the steepest hiring reductions in entry-level tech positions.

Anthropic presents the outcome as a matter of choice, not a prediction.

The model makes the trade-off explicit: the scenario generating the most wealth also allocates the smallest share to workers. The harder question is who decides.