---
id: "lib-2026-09-00-econ-scenario-explorer"
type: "library"
title: "Econ Scenario Explorer"
lang: "EN"
org: "Anthropic Institute"
source_type: "Article"
date: "2026-09-01"
week: "2026-W36"
url: "https://nosov.ai/library/lib-2026-09-00-econ-scenario-explorer/"
original_url: "https://www.anthropic.com/institute/econ-scenarios"
tier: "B"
tier_name: "Useful, limited"
tier_reason: "Interactive companion to the Korinek-Jones scenario model; reliable on the model's figures, but a public-facing tool, with the analysis in the technical report itself."
summary_basis: "full-text"
original_minutes: "15"
tags: ["scenarios"]
---

# Econ Scenario Explorer

Tier B (Useful, limited): Interactive companion to the Korinek-Jones scenario model; reliable on the model's figures, but a public-facing tool, with the analysis in the technical report itself.

Summary based on the full text.

## In short

Version 1.0 of Anthropic's Econ Scenario Explorer (September 2026) is an interactive web page built on the technical report "Economic Scenarios for Transformative AI" (Korinek et al., 2026). It explains the economy as bundles of tasks, using a nurse's job as the example, then lets readers enter their own expectations for AI capability, adoption, autonomy, productivity and re-employment time and see the implied US economy in 2030. Its headline: up to about twice normal growth, unemployment stays in the historical range; beyond that, knowledge workers face falling wages and job loss while the economy gets much richer.

## Key numbers

- $34.1T / $36.3T / $44.4T: US GDP in 2030 at 2025 prices in the modest, substantial and extreme scenarios
- 54.8%: capital share of GDP in the extreme scenario, up 14.8 points from 40%
- 10,980: US adults surveyed in August 2026; the typical answer implies the substantial scenario
- ~10%: share of respondents whose views match the extreme scenario

## What it reports

- Three scenarios: modest (impact like the internet), substantial (bigger than the internet or the railroad), extreme (likely requires recursively self-improving AI adopted quickly).
- 2030 GDP: +1.6% ($34.1T), +8.3% ($36.3T), +32.4% ($44.4T), at 2025 prices.
- Labor share: 59.4%, 56.1% and 45.2%; capital share up 0.6, 3.9 and 14.8 points from today's 40 cents per dollar.
- Extreme: growth reaches 15% a year, doubling the economy every 4.5 years; knowledge-worker wages fall by more than 10% by 2030; total labor income barely changes.
- Substantial: knowledge-worker wages essentially flat, other workers gain; the page states AI can do half of all knowledge work by 2030 but most of it is still done without AI.
- Survey of more than 10,000 Americans in August (n = 10,980) plus 29,578 site visitors: the typical respondent implies GDP 10% higher by 2030 and unemployment around 5%.
- Tasks are drawn from the US Department of Labor's O*NET taxonomy.

## How they got it

The page runs the paper's model in the browser. Inputs are five questions mapped to model parameters; outputs are GDP, job reallocation, wages by group and the labor-capital split. Eighteen named economists, including Daron Acemoglu, David Autor and Pascual Restrepo, commented on the draft report; they were not asked to endorse it.

## Limits

- The page's rounding differs from the paper: it gives about 5% unemployment and +10% GDP for the typical respondent, where the paper reports 4.6% and +8.6%.
- Its description of the substantial scenario (AI capable of half of knowledge work) is stronger than the paper's parameter (about 20% of cognitive tasks affected by 2030).
- Reviewers' open criticisms are listed: no individual workers, no data-center demand effects, the extreme case may be a thought experiment, the modest case may understate current data.

## Why it matters for companies

It is the quickest way to show a leadership team how different beliefs about AI translate into wages, headcount and the labor share, using the same model. The task-bundle framing (unchanged, augmented, automated, new) maps directly onto job architecture work. Use it for scenario discussion, and quote the technical report rather than the page when numbers matter.

Original: https://www.anthropic.com/institute/econ-scenarios