New evidence · September 2026

Three economic futures. One practical stress test.

Anthropic's new model asks how AI capability, adoption, automation, productivity, and new-task creation could reshape the US economy by 2030. These are conditional scenarios, not forecasts. We crosswalk its published SOC grouping to all 620 roles so you can see which broad results apply to your occupation.

Published model outputs

What changes by 2030

Every value below comes from the paper's baseline calibration and is measured against its no-AI path unless stated otherwise.

Modest change

+1.6% GDP

AI has an internet-scale macroeconomic effect that arrives gradually.

Tasks affected and used
4%
2030 GDP growth
2.4%
Cognitive employment
-0.5%
Overall unemployment
3.9%
Labor share of income
59.4%
Substantial change

+8.3% GDP

AI affects roughly one fifth of cognitive task instances and the economy grows at about twice its normal rate.

Tasks affected and used
12%
2030 GDP growth
5.4%
Cognitive employment
-3.9%
Overall unemployment
4.6%
Labor share of income
56.1%
Extreme change

+32.4% GDP

AI reaches nearly half of cognitive work, diffuses quickly, and creates no replacement cognitive tasks in the model.

Tasks affected and used
30%
2030 GDP growth
15.4%
Cognitive employment
-21.5%
Overall unemployment
11.9%
Labor share of income
45.2%

Occupation crosswalk

Stress-test your role

The paper models cognitive occupations (SOC major groups 11-29, 41, and 43) directly and combines every other occupation into one comparison group. This tool reproduces that boundary exactly.

Cognitive occupation group

Customer Service Representatives

Existing displacement.ai pressure: 73/100 · High. This score is unchanged by the new source.

ScenarioGroup wage vs. no-AI pathGroup employment vs. mid-2026Group unemploymentOverall unemployment
Modest change+0.4%-0.5%2.9%3.9%
Substantial change-0.3%-3.9%4.5%4.6%
Extreme change-11.5%-21.5%17.9%11.9%

These figures apply to the paper's broad worker group, not to the selected occupation. They do not estimate your chance of unemployment, your future wage, or local hiring conditions.

Our extrapolation

What the crosswalk adds

We applied Anthropic's published SOC boundary to the current displacement.ai catalog, without changing any role score.

Directly modeled cohort

398 roles fall in the paper's cognitive group; 222 fall in its all-other group.

Existing pressure gap

The cognitive cohort averages 39/100 in the current seed model, compared with 25/100 for all other roles.

Elevated planning pressure

85 cognitive roles score high or very high, compared with 7 roles outside that cohort.

The decision implication

The new evidence strengthens the case for scenario-based planning around knowledge work. The practical trigger is not a single 2030 forecast: it is whether capability, adoption, autonomous use, and worker re-employment time begin moving together toward the substantial path. Workers in high-pressure cognitive roles should build proof of adjacent skills early; employers should track task redesign and internal mobility; policymakers should watch occupation switching and the labor share alongside headline GDP.

Read before using

What this model cannot tell you

  • The scenarios have no assigned probabilities and diverge mainly after 2027.
  • The model has two worker groups; it does not follow individual workers or estimate occupation-specific layoffs.
  • It omits policy responses, business cycles, aggregate-demand and financial-market disruption, detailed worker differences, and rapid progress in physical robotics.
  • Its survey included 10,980 US adults; the modeled outcome distribution uses the 3,259 respondents who answered all five model inputs.
  • Capital supply and wage rigidity materially change who receives the gains and whether disruption appears through lower wages or higher unemployment.

Source: Anthropic Econ Scenario Explorer v1.0 and The Anthropic Institute Working Paper No. 2026-02.