Open Output (the explorer) · Open Analysis: Report and Slides · Open Materials · Overview

Coverage tracker: "Economic Scenarios for Transformative AI"

Reception of the paper: press, commentary, citations, and any revision of the artifact itself. This is not a tracker of whether the scenarios come true, and not a reading list of my own understanding, which lives in a separate tracker kept out of this repository.

Updated by hand when new coverage appears. coverage.html, the version published with the site, is regenerated from coverage.md.

The artifact being tracked
Paper Korinek, Jones, Sacher, Cotter and McCrory (2026), "Economic Scenarios for Transformative AI"
Venue The Anthropic Institute Working Paper No. 2026-02, September 2026
Released 2026-09-10 (first commentary appeared the same day)
Landing page anthropic.com/institute/econ-scenarios, labelled Version 1.0
PDF www-cdn.anthropic.com/files/4zrzovbb/...pdf, 57 pages, local copy paper.pdf
Companion the interactive scenario explorer on the landing page
Replication package none, checked 2026-09-10 and 2026-09-17

Conventions for an entry

Official

Date Item Read? Engagement Note
2026-09 Landing page and scenario explorer yes model Gives the three scenarios as +1.6% / +8.3% / +32.4% with 2030 GDP of $34.1T / $36.3T / $44.4T. States "External reviewers were not asked to endorse our conclusions", flags itself as Version 1.0, and concedes the model "isolates a few key forces and omits many others"
2026-09-10 Anton Korinek, thread on X no none Author's own launch thread, "My first working paper at Anthropic"

Commentary worth reading

Date Item Read? Engagement Note
2026-09-10 Tyler Cowen, Marginal Revolution yes numbers "sober, reasoned, scientific, and (for the most part) dynamically consistent"; calls it "a strong line-up from Anthropic". Picks his own path: "I will opt for something a modest bit more than modest", i.e. between modest and substantial. Does not touch the equations or the calibration. Points readers to commentary by Alex Imas
2026-09 Alex Imas commentary, linked by Cowen no none Lead only, URL not yet found. Worth chasing since it is the one piece Cowen singled out
2026-09 AI, Redistribution, and the Size of the Pie, Marginal Revolution no none Appears to use the paper as a jumping-off point for the distributional question
2026-09 Anthropic Modeled the End of the Knowledge Worker. Watch the Capital Share. no none Title suggests it keys on the labor share, the 60 to 45 percent fall, which is the paper's real distributional claim
2026-09 Anthropic's Transformative AI Economic Scenarios: Model, Evidence, and Civilizational Stakes no none Longest independent piece found. A search summary attributes to it the claim that critics think the paper understates AI by giving ideas only a 0.6 percent boost by 2030, contrasting Tom Davidson and Leopold Aschenbrenner. The attribution is unverified; the 0.6 percent itself matches Table 3's ideas stock in the extreme scenario (0.61; this reproduction gives 0.607)
2026-09 Three Paths for Work and Income under AI, 2026 to 2030 no none

Press

Date Item Read? Engagement Note
2026-09-11 Euronews, Anthropic says AI could bring both 15% growth and mass unemployment by 2030 no numbers The "15%" is the extreme scenario's annual growth rate in 2030, a different object from the 32.4 percent level gap. See the conflation note below
2026-09 The Next Web, Anthropic's three AI economic scenarios for 2030, explained no none
2026-09 unite.ai, Anthropic Releases Interactive Model of AI's Possible Economic Futures no none Leads on the explorer rather than the paper
2026-09 XenoSpectrum, Growth and Worker Wages Can Diverge no none Framing matches the paper's actual point
2026-09-15 Disruption Banking, Anthropic is Fine With You Losing Your Job, So Long as It's Done Safely no framing Adversarial framing aimed at Anthropic, likely not at the model
2026-09 Startup Fortune, Anthropic Says AI Could Grow The Economy Or Wreck The Job Market By 2030 no none
2026-09 Anthropic Maps Three AI Economic Scenarios to 2030 no none Asset-management trade press

Aggregators and career-advice derivatives

Logged for completeness, low signal, none of them read: Workopia, The AI Career Lab, Flowtivity, a dev.to repost, and a WordPress repost.

Claims circulating in coverage that the paper does not support

Checked against the paper, since this is where a reproduction has something to say.

  1. "Peer reviewers" or "external reviewers vetted it." The paper's first-page footnote thanks 22 named economists, among them Acemoglu, Autor, Restrepo, Romer, Nakamura, Klenow, Moll and Steinsson, "for insightful comments on an earlier version". That is an acknowledgement, not review, and the landing page says outright that reviewers were not asked to endorse the conclusions. Coverage that presents the list as vetting overstates it. Verified against paper.pdf p. 1.
  2. Level gaps reported as growth rates. The headline 1.6 / 8.3 / 32.4 percent are 2030 GDP levels relative to a no-AI path, not growth rates. Several outlets, and the landing page's own "+32.4% GDP growth" label, blur this; Euronews's "15% growth" is a third quantity again, the extreme scenario's 2030 annual rate. The deck's backup slide on growth rates against level gaps exists precisely for this.

What to watch

Log