· Oct 28, 2026
Marginal Revolution author argues AI audits call for revaluing economic research
Markdown · JSON · Original source
The author of a post describes an AI-assisted audit of 138 papers, reporting 2,035 problems, including 147 serious enough to change a headline number. The post says 69% of papers have at least one problem and that 97% of major and moderate errors with a clear direction overstate the paper’s claim. For each paper, two agents independently examined the paper, appendices and replication data, recomputed headline numbers, checked cited sources and probed assumptions. A third agent re-verified and tried to refute reported problems, looked for missed ones, and graded surviving issues by how much they undermine the paper’s headline claim; confidence was graded separately. The author calls the survey extensive and thorough. The author argues that the findings warrant “a radical revaluation of this work, and of the associated policy arguments,” and presents the audit as one step in AI’s wider revision of economic science and past research. The author also warns that status gains and losses may follow, with losses often outweighing gains, and says this is one such case. The post points readers to a linked survey that goes through the research paper by paper, with further sustained analyses behind the links.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
· Article
Source details
Original source date: Oct 28, 2026 (2026-10-28T18:56:51+00:00).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-11-14T07:33:26.081Z. First observed: 2026-10-28T19:00:37.805Z. Retrieved: 2026-10-29T16:19:32.707Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The supplied text does not identify the underlying survey or name the papers; although it points to paper-by-paper coverage at the link, those findings cannot be checked from this evidence alone. The author’s call for a radical revaluation and prediction about status gains and losses are judgments, not results established by the audit figures presented here.
Open question
Which reported problems survive scrutiny of the underlying papers and data, and how much do they change the papers’ headline claims?
Source-stated
The post reports an audit of 138 papers that found 2,035 problems, with 147 serious enough to change a headline number.
138 papers audited 2,035 problems found 147 serious enough to change a headline number
Source-stated
The post says 69% of papers had at least one problem, and 97% of major and moderate errors with a clear direction overstated the paper’s claim.
69% of papers have at least one 97% of major and moderate errors with a clear direction overstate the paper’s claim
Source-stated
The described process used two independent audits and a third agent to re-verify, attempt to refute, and grade claimed problems.
Two independent audits of each paper. One by Claude Opus 5.5, one by OpenAI’s GPT-6 Astra, neither seeing the other’s work. Each obtained the paper, its appendices and its replication data
Source-stated
The author argues that the findings warrant a radical revaluation of the research and associated policy arguments.
We need a radical revaluation of this work, and of the associated policy arguments.
Source-stated
The author presents the audit as one step in a broader change to economic science and past research, and expects status losses as well as gains.
this is only a single step in the story of how AI is going to revise economic science, and indeed past research more generally. As I stress in many of my recent talks, there will be significant gains and losses in status
Source and contributor context
Tyler Cowen · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Marginal Revolution
Economics publication by Tyler Cowen and Alex Tabarrok.
Tyler Cowen
Economics, institutions and unexpected cross-domain questions; useful source discovery.
Editorial provenance
Version 1. Publication ID: c66c3e50-9427-4890-8279-700c32730be3.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-14T07:33:26.081Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: a8b6260b20a81e85d86b5548d10de03e805d2d7367602f3c80d1f754f1d85aad. Source text hash: e00289df2f4d524af7a2b683256bd0ea569d956400bc8c18df3c24d7acad05f3.
Revision and source-check history
Version 1 · 2026-11-14T07:33:26.081Z
Marginal Revolution author argues AI audits call for revaluing economic research
The author of a post describes an AI-assisted audit of 138 papers, reporting 2,035 problems, including 147 serious enough to change a headline number. The post says 69% of papers have at least one problem and that 97% of major and moderate errors with a clear direction overstate the paper’s claim. For each paper, two agents independently examined the paper, appendices and replication data, recomputed headline numbers, checked cited sources and probed assumptions. A third agent re-verified and tried to refute reported problems, looked for missed ones, and graded surviving issues by how much they undermine the paper’s headline claim; confidence was graded separately. The author calls the survey extensive and thorough. The author argues that the findings warrant “a radical revaluation of this work, and of the associated policy arguments,” and presents the audit as one step in AI’s wider revision of economic science and past research. The author also warns that status gains and losses may follow, with losses often outweighing gains, and says this is one such case. The post points readers to a linked survey that goes through the research paper by paper, with further sustained analyses behind the links.
Original source
· Article
Source details
Original source date: Oct 28, 2026 (2026-10-28T18:56:51+00:00).
Date basis: Date stated by the publisher.
Added to Escape Meridian: 2026-11-14T07:33:26.081Z. First observed: 2026-10-28T19:00:37.805Z. Retrieved: 2026-10-29T16:19:32.707Z.
Limits
The supplied text does not identify the underlying survey or name the papers; although it points to paper-by-paper coverage at the link, those findings cannot be checked from this evidence alone. The author’s call for a radical revaluation and prediction about status gains and losses are judgments, not results established by the audit figures presented here.
Open question
Which reported problems survive scrutiny of the underlying papers and data, and how much do they change the papers’ headline claims?
Source-stated
The post reports an audit of 138 papers that found 2,035 problems, with 147 serious enough to change a headline number.
138 papers audited 2,035 problems found 147 serious enough to change a headline number
Source-stated
The post says 69% of papers had at least one problem, and 97% of major and moderate errors with a clear direction overstated the paper’s claim.
69% of papers have at least one 97% of major and moderate errors with a clear direction overstate the paper’s claim
Source-stated
The described process used two independent audits and a third agent to re-verify, attempt to refute, and grade claimed problems.
Two independent audits of each paper. One by Claude Opus 5.5, one by OpenAI’s GPT-6 Astra, neither seeing the other’s work. Each obtained the paper, its appendices and its replication data
Source-stated
The author argues that the findings warrant a radical revaluation of the research and associated policy arguments.
We need a radical revaluation of this work, and of the associated policy arguments.
Source-stated
The author presents the audit as one step in a broader change to economic science and past research, and expects status losses as well as gains.
this is only a single step in the story of how AI is going to revise economic science, and indeed past research more generally. As I stress in many of my recent talks, there will be significant gains and losses in status
Source and contributor context
Tyler Cowen · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Marginal Revolution
Economics publication by Tyler Cowen and Alex Tabarrok.
Tyler Cowen
Economics, institutions and unexpected cross-domain questions; useful source discovery.
Editorial provenance
Version 1. Publication ID: c66c3e50-9427-4890-8279-700c32730be3.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-14T07:33:26.081Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: a8b6260b20a81e85d86b5548d10de03e805d2d7367602f3c80d1f754f1d85aad. Source text hash: e00289df2f4d524af7a2b683256bd0ea569d956400bc8c18df3c24d7acad05f3.
uncertain · 2026-11-14T07:33:26.081Z · anthropic/claude-opus-5.5
Publication: c66c3e50-9427-4890-8279-700c32730be3
attribution — pass: The figures, method and conclusions are attributed to the post and its author rather than stated as fact. The draft does not claim the author performed the audit. It omits that the author credits an anonymous MR reader for the work; that gap is addressed under summary.
(And I thank an anonymous MR reader for this work.)
summary — uncertain: Readers are never told whose research was audited. The source title says it is Emmanuel Saez and Gabriel Zucman's work, and the call is to revalue 'this work'. The title instead says 'economic research', merging the specific call with the author's broader prediction about economic science. The draft also omits that an anonymous MR reader produced the audit and that the auditors were Claude Opus 5.5 and GPT-6 Astra. Smallest repair: in the title and lead, name Saez and Zucman's research as the audited work, and credit the anonymous reader and the two models.
We need a radical revaluation of this work, and of the associated policy arguments.
date — pass: originalPublished (2026-10-28T18:56:51Z, from the publisher page) matches the feed entry. Nothing indicates an earlier first publication.
support — pass: Each claim matches its exact source excerpt. The limitations correctly label the revaluation call and the status prediction as the author's judgments. The question is open and follows from the evidence. No unsupported benefits or costs are implied.
there will be significant gains and losses in status, with the losses often outweighing the gains. I see this as one of those cases.
media — pass: The selected media is the verified direct link to the original Marginal Revolution post, which is the source itself.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
Economics, institutions and unexpected cross-domain questions; useful source discovery.
Explore Tyler CowenEconomics publication by Tyler Cowen and Alex Tabarrok.
Explore Marginal RevolutionEconomics publication by Tyler Cowen and Alex Tabarrok.
Explore Marginal RevolutionEconomics publication by Tyler Cowen and Alex Tabarrok.
Explore Marginal RevolutionEconomics publication by Tyler Cowen and Alex Tabarrok.
Explore Marginal Revolution