{"id":"874a368385e615aca86cdcd559d678ccea3f2ffc08eef2cc34610ee66b08ba98","title":"METR researchers outline the economics of recursive self-improvement","argument":"On July 22, 2026, METR researchers Parker and Tom describe a paper that uses simple models to examine how AI may accelerate AI research and development (R&D). They say they want to understand whether AI-driven feedback could accelerate future capabilities, an input to their assessment of risks from frontier AI development.\n\nThe researchers avoid using “RSI” in a technical sense because the term has been used for different levels of feedback. These range from any link between capabilities and model improvements to feedback strong enough for super-exponential or fully autonomous growth. Instead, they examine feedback strength and whether it could produce “self-sustaining acceleration.”\n\nThey identify the effect of increased model capabilities on the rate of algorithmic progress as the model’s most uncertain relationship. They describe possible bottlenecks, including data, training compute, inference compute, experiments, and algorithmic- or R&D-specific capabilities. They say the evidence for these limits is not overwhelming, so they cannot rule out extended, rapid acceleration.\n\nThe authors say labs could release more data about AI’s effects on AI R&D. They describe the paper’s calibration, which suggests estimates for model parameters, as loose and a first draft. They hope to make the model more useful for operational purposes.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on October 23, 2026; originally published July 22, 2026. The note says it is an unpolished research update that has gone through less review than METR research posts and may not reflect METR’s views as a whole. The evidence does not provide the paper’s full model, calibration details, or the specific data-release wish list.","speaker":{"name":"METR","roleAtSource":"Publisher"},"source":{"title":"The Economics of Recursive Self-Improvement","publisher":"METR","format":"Article","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","publishedAt":"2026-07-22T07:00:00.000Z","dateBasis":"Date supplied by the publisher’s feed."},"evidence":[{"text":"METR researchers Parker and Tom say their paper uses simple models to explore how AI may accelerate AI R&D.","excerpt":"The paper walks through a series of simple models of how AI may accelerate AI R&D","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"},{"text":"The authors say the effect of increased model capabilities on the rate of algorithmic progress is the model’s most uncertain relationship.","excerpt":"The most uncertain relationship is how an increase in model capabilities would increase the rate of algorithmic progress.","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"},{"text":"The researchers say they cannot rule out extended, rapid acceleration in capabilities, because they do not consider the evidence for possible bottlenecks overwhelming.","excerpt":"we do not think the evidence for any of these is overwhelming; we cannot rule out an extended and rapid acceleration in capabilities.","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"},{"text":"The authors describe the paper’s calibration as loose and intended as a first draft, and say they hope to improve the model’s operational usefulness.","excerpt":"The paper has a calibration, suggesting estimates for parameters, but it is very loose and meant to be a first draft.","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What data would best establish how increases in model capabilities affect the rate of algorithmic progress?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"METR researchers outline the economics of recursive self-improvement","summary":"On July 22, 2026, METR researchers Parker and Tom describe a paper that uses simple models to examine how AI may accelerate AI research and development (R&D). They say they want to understand whether AI-driven feedback could accelerate future capabilities, an input to their assessment of risks from frontier AI development.\n\nThe researchers avoid using “RSI” in a technical sense because the term has been used for different levels of feedback. These range from any link between capabilities and model improvements to feedback strong enough for super-exponential or fully autonomous growth. Instead, they examine feedback strength and whether it could produce “self-sustaining acceleration.”\n\nThey identify the effect of increased model capabilities on the rate of algorithmic progress as the model’s most uncertain relationship. They describe possible bottlenecks, including data, training compute, inference compute, experiments, and algorithmic- or R&D-specific capabilities. They say the evidence for these limits is not overwhelming, so they cannot rule out extended, rapid acceleration.\n\nThe authors say labs could release more data about AI’s effects on AI R&D. They describe the paper’s calibration, which suggests estimates for model parameters, as loose and a first draft. They hope to make the model more useful for operational purposes.","question":"What data would best establish how increases in model capabilities affect the rate of algorithmic progress?","claims":[{"text":"METR researchers Parker and Tom say their paper uses simple models to explore how AI may accelerate AI R&D.","excerpt":"The paper walks through a series of simple models of how AI may accelerate AI R&D","support":"source-stated"},{"text":"The authors say the effect of increased model capabilities on the rate of algorithmic progress is the model’s most uncertain relationship.","excerpt":"The most uncertain relationship is how an increase in model capabilities would increase the rate of algorithmic progress.","support":"source-stated"},{"text":"The researchers say they cannot rule out extended, rapid acceleration in capabilities, because they do not consider the evidence for possible bottlenecks overwhelming.","excerpt":"we do not think the evidence for any of these is overwhelming; we cannot rule out an extended and rapid acceleration in capabilities.","support":"source-stated"},{"text":"The authors describe the paper’s calibration as loose and intended as a first draft, and say they hope to improve the model’s operational usefulness.","excerpt":"The paper has a calibration, suggesting estimates for parameters, but it is very loose and meant to be a first draft.","support":"source-stated"}],"limitations":["Text as retrieved on October 23, 2026; originally published July 22, 2026.","The note says it is an unpolished research update that has gone through less review than METR research posts and may not reflect METR’s views as a whole.","The evidence does not provide the paper’s full model, calibration details, or the specific data-release wish list."],"mediaId":null,"correctionReason":null,"id":"874a368385e615aca86cdcd559d678ccea3f2ffc08eef2cc34610ee66b08ba98","publicationId":"22686477-4098-e574-256a-23e34b5df080","version":1,"sourceUrl":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","sourceTitle":"The Economics of Recursive Self-Improvement","publisher":"METR","publishedAt":"2026-07-22T07:00:00.000Z","originalPublished":{"value":"2026-07-22T07:00:00.000Z","precision":"instant","basis":"backfill:feed_published"},"firstObservedAt":"2026-10-18T02:47:40.117Z","createdAt":"2026-10-29T17:48:03.567Z","historical":true,"entityIds":["metr"],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"d8d34486d9a7a85c83cbc53c0a0977f7301c0547cd1b60c5d92a886f667a6c9b","rubricVersion":"unrecorded","provider":"vercel","model":"typesafe-ai/jev","reasonCodes":["bridge_admitted_by_model"],"confidence":0.77,"decidedAt":"2026-10-23T22:45:37.984Z"},"provenance":{"textHash":"0661512019dfbc5d68aa3e11d54f51bcc10db09832fa3e2fd98bc1a071c065a4","observationId":"aa3864d244dc5e2a3a3bd81972398a4d9d83b9511ebcef6a29de9fd2990dcdf4","retrievedAt":"2026-10-23T22:47:38.839Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"34789c3d-2541-1946-5baa-1b7158f649ab"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"a8cb8fe3-9d7e-7b0a-551a-2f5f8dc1c14d"}},"audit":{"status":"passed","reviewedAt":"2026-10-29T17:47:04.791Z","model":"anthropic/claude-opus-5.5"}},"canonicalUrl":"https://escapemeridian.com/readings/metr-researchers-outline-the-economics-of-recursive-self-874a3683","formats":{"html":"/readings/metr-researchers-outline-the-economics-of-recursive-self-874a3683","markdown":"/readings/metr-researchers-outline-the-economics-of-recursive-self-874a3683.md","json":"/readings/metr-researchers-outline-the-economics-of-recursive-self-874a3683.json"},"attribution":"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.","readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-frontier-model","name":"Frontier model","explanation":"One of the most capable AI models available at a given time; the set shifts as models improve.","referenceUrl":"https://www.gov.uk/government/publications/frontier-ai-capabilities-and-risks-discussion-paper"},{"id":"term-ai-compute","name":"AI compute","explanation":"The computing resources, meaning chips, data centers and electricity, used to train or run AI models.","referenceUrl":"https://developers.google.com/machine-learning/glossary#compute"},{"id":"term-inference","name":"Model inference","explanation":"Using a trained model to produce a prediction or response to new input.","referenceUrl":"https://developers.google.com/machine-learning/glossary#inference"},{"id":"term-parameters","name":"Model parameters","explanation":"The internal numbers a model adjusts during training; their count is a rough measure of a model's size.","referenceUrl":"https://developers.google.com/machine-learning/glossary#parameter"}],"entities":[{"id":"metr","name":"METR","kind":"organization","context":"METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.","referenceUrls":["https://metr.org/"],"path":"/entities/metr","json":"/entities/metr.json","markdown":"/entities/metr.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"874a368385e615aca86cdcd559d678ccea3f2ffc08eef2cc34610ee66b08ba98","revision":"22686477-4098-e574-256a-23e34b5df080","revisionKind":"publication-id","title":"METR researchers outline the economics of recursive self-improvement","url":"https://escapemeridian.com/readings/metr-researchers-outline-the-economics-of-recursive-self-874a3683"},"source":{"type":"source","id":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","revision":"0661512019dfbc5d68aa3e11d54f51bcc10db09832fa3e2fd98bc1a071c065a4","revisionKind":"sha256-source-text","title":"The Economics of Recursive Self-Improvement","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"874a368385e615aca86cdcd559d678ccea3f2ffc08eef2cc34610ee66b08ba98","title":"METR researchers outline the economics of recursive self-improvement","argument":"On July 22, 2026, METR researchers Parker and Tom describe a paper that uses simple models to examine how AI may accelerate AI research and development (R&D). They say they want to understand whether AI-driven feedback could accelerate future capabilities, an input to their assessment of risks from frontier AI development.\n\nThe researchers avoid using “RSI” in a technical sense because the term has been used for different levels of feedback. These range from any link between capabilities and model improvements to feedback strong enough for super-exponential or fully autonomous growth. Instead, they examine feedback strength and whether it could produce “self-sustaining acceleration.”\n\nThey identify the effect of increased model capabilities on the rate of algorithmic progress as the model’s most uncertain relationship. They describe possible bottlenecks, including data, training compute, inference compute, experiments, and algorithmic- or R&D-specific capabilities. They say the evidence for these limits is not overwhelming, so they cannot rule out extended, rapid acceleration.\n\nThe authors say labs could release more data about AI’s effects on AI R&D. They describe the paper’s calibration, which suggests estimates for model parameters, as loose and a first draft. They hope to make the model more useful for operational purposes.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on October 23, 2026; originally published July 22, 2026. The note says it is an unpolished research update that has gone through less review than METR research posts and may not reflect METR’s views as a whole. The evidence does not provide the paper’s full model, calibration details, or the specific data-release wish list.","speaker":{"name":"METR","roleAtSource":"Publisher"},"source":{"title":"The Economics of Recursive Self-Improvement","publisher":"METR","format":"Article","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","publishedAt":"2026-07-22T07:00:00.000Z","dateBasis":"Date supplied by the publisher’s feed."},"evidence":[{"text":"METR researchers Parker and Tom say their paper uses simple models to explore how AI may accelerate AI R&D.","excerpt":"The paper walks through a series of simple models of how AI may accelerate AI R&D","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"},{"text":"The authors say the effect of increased model capabilities on the rate of algorithmic progress is the model’s most uncertain relationship.","excerpt":"The most uncertain relationship is how an increase in model capabilities would increase the rate of algorithmic progress.","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"},{"text":"The researchers say they cannot rule out extended, rapid acceleration in capabilities, because they do not consider the evidence for possible bottlenecks overwhelming.","excerpt":"we do not think the evidence for any of these is overwhelming; we cannot rule out an extended and rapid acceleration in capabilities.","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"},{"text":"The authors describe the paper’s calibration as loose and intended as a first draft, and say they hope to improve the model’s operational usefulness.","excerpt":"The paper has a calibration, suggesting estimates for parameters, but it is very loose and meant to be a first draft.","url":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What data would best establish how increases in model capabilities affect the rate of algorithmic progress?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"METR researchers outline the economics of recursive self-improvement","summary":"On July 22, 2026, METR researchers Parker and Tom describe a paper that uses simple models to examine how AI may accelerate AI research and development (R&D). They say they want to understand whether AI-driven feedback could accelerate future capabilities, an input to their assessment of risks from frontier AI development.\n\nThe researchers avoid using “RSI” in a technical sense because the term has been used for different levels of feedback. These range from any link between capabilities and model improvements to feedback strong enough for super-exponential or fully autonomous growth. Instead, they examine feedback strength and whether it could produce “self-sustaining acceleration.”\n\nThey identify the effect of increased model capabilities on the rate of algorithmic progress as the model’s most uncertain relationship. They describe possible bottlenecks, including data, training compute, inference compute, experiments, and algorithmic- or R&D-specific capabilities. They say the evidence for these limits is not overwhelming, so they cannot rule out extended, rapid acceleration.\n\nThe authors say labs could release more data about AI’s effects on AI R&D. They describe the paper’s calibration, which suggests estimates for model parameters, as loose and a first draft. They hope to make the model more useful for operational purposes.","question":"What data would best establish how increases in model capabilities affect the rate of algorithmic progress?","claims":[{"text":"METR researchers Parker and Tom say their paper uses simple models to explore how AI may accelerate AI R&D.","excerpt":"The paper walks through a series of simple models of how AI may accelerate AI R&D","support":"source-stated"},{"text":"The authors say the effect of increased model capabilities on the rate of algorithmic progress is the model’s most uncertain relationship.","excerpt":"The most uncertain relationship is how an increase in model capabilities would increase the rate of algorithmic progress.","support":"source-stated"},{"text":"The researchers say they cannot rule out extended, rapid acceleration in capabilities, because they do not consider the evidence for possible bottlenecks overwhelming.","excerpt":"we do not think the evidence for any of these is overwhelming; we cannot rule out an extended and rapid acceleration in capabilities.","support":"source-stated"},{"text":"The authors describe the paper’s calibration as loose and intended as a first draft, and say they hope to improve the model’s operational usefulness.","excerpt":"The paper has a calibration, suggesting estimates for parameters, but it is very loose and meant to be a first draft.","support":"source-stated"}],"limitations":["Text as retrieved on October 23, 2026; originally published July 22, 2026.","The note says it is an unpolished research update that has gone through less review than METR research posts and may not reflect METR’s views as a whole.","The evidence does not provide the paper’s full model, calibration details, or the specific data-release wish list."],"mediaId":null,"correctionReason":null,"id":"874a368385e615aca86cdcd559d678ccea3f2ffc08eef2cc34610ee66b08ba98","publicationId":"22686477-4098-e574-256a-23e34b5df080","version":1,"sourceUrl":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","sourceTitle":"The Economics of Recursive Self-Improvement","publisher":"METR","publishedAt":"2026-07-22T07:00:00.000Z","originalPublished":{"value":"2026-07-22T07:00:00.000Z","precision":"instant","basis":"backfill:feed_published"},"firstObservedAt":"2026-10-18T02:47:40.117Z","createdAt":"2026-10-29T17:48:03.567Z","historical":true,"entityIds":["metr"],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://metr.org/notes/2026-07-22-economics-of-recursive-self-improvement/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"d8d34486d9a7a85c83cbc53c0a0977f7301c0547cd1b60c5d92a886f667a6c9b","rubricVersion":"unrecorded","provider":"vercel","model":"typesafe-ai/jev","reasonCodes":["bridge_admitted_by_model"],"confidence":0.77,"decidedAt":"2026-10-23T22:45:37.984Z"},"provenance":{"textHash":"0661512019dfbc5d68aa3e11d54f51bcc10db09832fa3e2fd98bc1a071c065a4","observationId":"aa3864d244dc5e2a3a3bd81972398a4d9d83b9511ebcef6a29de9fd2990dcdf4","retrievedAt":"2026-10-23T22:47:38.839Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"34789c3d-2541-1946-5baa-1b7158f649ab"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"a8cb8fe3-9d7e-7b0a-551a-2f5f8dc1c14d"}},"audit":{"status":"passed","reviewedAt":"2026-10-29T17:47:04.791Z","model":"anthropic/claude-opus-5.5"}}}],"audits":[{"publicationId":"22686477-4098-e574-256a-23e34b5df080","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the note and its claims to METR researchers Parker and Tom, who coauthored the paper. It also carries the source's caveat that notes may not reflect METR's views as a whole. It omits the seven other economist coauthors, but this does not misattribute anything.","excerpt":"We (Parker and Tom) recently coauthored a paper, “The Economics of Recursive Self-Improvement”, with 7 other economists."},{"dimension":"summary","status":"pass","explanation":"The summary leads with the actor and the development: a paper modeling how AI may accelerate AI R&D. It explains the RSI definitional dispute in plain terms (feedback from capabilities to model improvements). It covers the key uncertainty, the possible bottlenecks, the conclusion that acceleration cannot be ruled out, and the call for more lab data. The title already expands RSI as recursive self-improvement. The question follows from the stated most uncertain relationship and the data wish list.","excerpt":"Instead we focus on the strength of feedback effects, and whether they are sufficiently strong for “self-sustaining acceleration.”"},{"dimension":"date","status":"pass","explanation":"The original publication is the instant 2026-07-22T07:00Z, and its UTC calendar day matches the stated July 22, 2026; the URL path also shows 2026-07-22. The source is a publisher page, not an archive copy, so no archive disclosure is needed. There is no hindsight framing. The authors' hopes for future iteration are reported as plans, not outcomes.","excerpt":null},{"dimension":"support","status":"pass","explanation":"Each claim is a source-stated paraphrase with an exact supporting excerpt. These cover: the simple models; algorithmic progress as the most uncertain relationship; that the evidence for bottlenecks is not overwhelming; the loose first-draft calibration; and that labs could release more data. The reading adds no unsupported benefits or costs.","excerpt":"There is more data relevant to RSI that the labs could be releasing."},{"dimension":"media","status":"pass","explanation":"No media is selected, and no verified media candidates were supplied, so choosing none is appropriate.","excerpt":null}],"createdAt":"2026-10-29T17:47:04.791Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}