# Ramez Naam questions whether AI self-improvement will cause a fast takeoff [Canonical reading](https://escapemeridian.com/readings/ramez-naam-questions-whether-ai-self-improvement-will-cause-a-c8f55cdf) In a guest post published September 27, 2026, Ramez Naam argues that AI is helping improve AI, but the current feedback loop may be too weak to drive a runaway intelligence explosion. Recursive self-improvement (RSI) describes AI helping develop better AI; Naam distinguishes that process from a runaway loop to superintelligence. Naam bases his skepticism on the gap he sees between benchmark results and AI’s performance on real research tasks, along with diminishing returns from more experiments and other inputs. He estimates that the self-improvement loop would need to become roughly five to ten times stronger to sustain itself, while emphasizing that the estimate is uncertain and depends on limited data. He does not argue that AI progress will stop: he expects rapid progress and allows that better data, major breakthroughs, or feedback from hardware and economic growth could change the picture. His question is whether AI will produce enough useful research to make successive rounds of improvement easier. He calls for more evidence about how AI-assisted research translates into better algorithms and more capable AI. 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 Noah Smith · Article Original source date: Sep 27, 2026 (2026-09-27T07:20:08.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-10-29T16:43:31.431Z. First observed: 2026-10-18T02:41:32.718Z. Retrieved: 2026-10-23T18:45:47.535Z. - [Where’s the “intelligence explosion”?](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Limits Text as retrieved on October 23, 2026; originally published September 27, 2026. Naam describes his estimates as rough and says more data is needed; the post does not establish whether the feedback loop will become self-sustaining. The post distinguishes benchmark tasks from internal AI research tasks and notes that they are not an apples-to-apples comparison. ## Open question What evidence would show that AI-assisted research is producing enough useful discoveries to strengthen the self-improvement loop? ## Source-stated Ramez Naam argues that AI is already helping improve AI, but the current feedback loop does not appear strong enough to sustain itself. > AI is helping build better AI. Under this estimate, though, each turn of the loop adds less than the last. The feedback would have to become much stronger to sustain itself. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated Naam estimates that the loop would need to be roughly five to ten times stronger than it is now to become self-sustaining. > With those assumptions, 2–3% per ECI point against a 15–19% threshold leaves a roughly five- to tenfold gap. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated Naam says the estimate is rough and calls for more data to assess whether AI-assisted research is strengthening the loop. > We need better measurements, shared often enough to catch changes as they happen. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated The post is a guest post by Ramez Naam, published September 27, 2026. > Ramez Naam > Sep 27, 2026 - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated Naam argues that feedback from software, hardware, and the broader economy could combine to drive progress, while questioning whether the software loop alone is strong enough. > Several feedback loops can combine to overcome diminishing returns even when one loop alone can’t. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source and contributor context Noah Smith · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Superintelligence A hypothetical AI far more capable than people across most intellectual tasks. - [Definition reference](https://en.wikipedia.org/wiki/Superintelligence) ## Benchmark A defined set of tests used to compare systems. Results depend on the tasks and measures selected. - [Definition reference](https://arxiv.org/abs/2211.09110) ## Ramez Naam Explicitly tests self-improvement and diminishing-return assumptions; connects AI to energy and economics. - [Explore this context](/entities/ramez-naam) - [Reference](https://www.diamandis.com/podcast/ramez-naam) - [Reference](https://www.rameznaam.com/p/ai-rsi-isnt-leading-to-super-intelligence) - [Reference](https://rameznaam.com/) ## Noah Smith Economic interpretation can separate technological usefulness from financing and deployment dynamics. - [Explore this context](/entities/noah-smith) - [Reference](https://lexfridman.com/ezra-klein-and-derek-thompson-transcript) - [Reference](https://www.dwarkesh.com/p/notes-on-china) - [Reference](https://www.dwarkesh.com/p/patrick-collison) - [Reference](https://www.noahpinion.blog/p/ai-vs-the-railroads) - [Reference](https://www.noahpinion.blog/) ## Editorial provenance Version 1. Publication ID: 3b3353ce-d063-dd11-de32-fa48e1278ae7. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-10-29T16:42:52.642Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: c0ab557eec08cb3c24b6e0b862330b9cf315d199e494828fc4e8ef31dbcb8875. Source text hash: da0bdea90f299f6d0dcf446b0810b8929b8022b8dd230e49b0b332f53b52cc68. ## Revision and source-check history ### Version 1 — 2026-10-29T16:43:31.431Z In a guest post published September 27, 2026, Ramez Naam argues that AI is helping improve AI, but the current feedback loop may be too weak to drive a runaway intelligence explosion. Recursive self-improvement (RSI) describes AI helping develop better AI; Naam distinguishes that process from a runaway loop to superintelligence. Naam bases his skepticism on the gap he sees between benchmark results and AI’s performance on real research tasks, along with diminishing returns from more experiments and other inputs. He estimates that the self-improvement loop would need to become roughly five to ten times stronger to sustain itself, while emphasizing that the estimate is uncertain and depends on limited data. He does not argue that AI progress will stop: he expects rapid progress and allows that better data, major breakthroughs, or feedback from hardware and economic growth could change the picture. His question is whether AI will produce enough useful research to make successive rounds of improvement easier. He calls for more evidence about how AI-assisted research translates into better algorithms and more capable AI. ## Original source Noah Smith · Article Original source date: Sep 27, 2026 (2026-09-27T07:20:08.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-10-29T16:43:31.431Z. First observed: 2026-10-18T02:41:32.718Z. Retrieved: 2026-10-23T18:45:47.535Z. - [Where’s the “intelligence explosion”?](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Limits Text as retrieved on October 23, 2026; originally published September 27, 2026. Naam describes his estimates as rough and says more data is needed; the post does not establish whether the feedback loop will become self-sustaining. The post distinguishes benchmark tasks from internal AI research tasks and notes that they are not an apples-to-apples comparison. ## Open question What evidence would show that AI-assisted research is producing enough useful discoveries to strengthen the self-improvement loop? ## Source-stated Ramez Naam argues that AI is already helping improve AI, but the current feedback loop does not appear strong enough to sustain itself. > AI is helping build better AI. Under this estimate, though, each turn of the loop adds less than the last. The feedback would have to become much stronger to sustain itself. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated Naam estimates that the loop would need to be roughly five to ten times stronger than it is now to become self-sustaining. > With those assumptions, 2–3% per ECI point against a 15–19% threshold leaves a roughly five- to tenfold gap. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated Naam says the estimate is rough and calls for more data to assess whether AI-assisted research is strengthening the loop. > We need better measurements, shared often enough to catch changes as they happen. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated The post is a guest post by Ramez Naam, published September 27, 2026. > Ramez Naam > Sep 27, 2026 - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source-stated Naam argues that feedback from software, hardware, and the broader economy could combine to drive progress, while questioning whether the software loop alone is strong enough. > Several feedback loops can combine to overcome diminishing returns even when one loop alone can’t. - [Supporting source](https://www.noahpinion.blog/p/wheres-the-intelligence-explosion) ## Source and contributor context Noah Smith · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Superintelligence A hypothetical AI far more capable than people across most intellectual tasks. - [Definition reference](https://en.wikipedia.org/wiki/Superintelligence) ## Benchmark A defined set of tests used to compare systems. Results depend on the tasks and measures selected. - [Definition reference](https://arxiv.org/abs/2211.09110) ## Ramez Naam Explicitly tests self-improvement and diminishing-return assumptions; connects AI to energy and economics. - [Explore this context](/entities/ramez-naam) - [Reference](https://www.diamandis.com/podcast/ramez-naam) - [Reference](https://www.rameznaam.com/p/ai-rsi-isnt-leading-to-super-intelligence) - [Reference](https://rameznaam.com/) ## Noah Smith Economic interpretation can separate technological usefulness from financing and deployment dynamics. - [Explore this context](/entities/noah-smith) - [Reference](https://lexfridman.com/ezra-klein-and-derek-thompson-transcript) - [Reference](https://www.dwarkesh.com/p/notes-on-china) - [Reference](https://www.dwarkesh.com/p/patrick-collison) - [Reference](https://www.noahpinion.blog/p/ai-vs-the-railroads) - [Reference](https://www.noahpinion.blog/) ## Editorial provenance Version 1. Publication ID: 3b3353ce-d063-dd11-de32-fa48e1278ae7. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-10-29T16:42:52.642Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: c0ab557eec08cb3c24b6e0b862330b9cf315d199e494828fc4e8ef31dbcb8875. Source text hash: da0bdea90f299f6d0dcf446b0810b8929b8022b8dd230e49b0b332f53b52cc68. ### passed — 2026-10-29T16:42:52.642Z — anthropic/claude-opus-5.5 Publication: 3b3353ce-d063-dd11-de32-fa48e1278ae7 attribution — pass: The reading correctly identifies this as a guest post by Ramez Naam on Noah Smith's blog. The hardware and economic feedback point paraphrases Davidson et al.'s model, which Naam presents. Naam himself also says these non-software inputs can keep driving rapid progress, so attributing the softened claim to him is acceptable. > Those can keep driving rapid progress even if the loop can’t sustain itself. summary — pass: The reading leads with Naam's argument and explains recursive self-improvement in plain language. It conveys his central five- to tenfold estimate, his caveats, and his expectation of continued rapid progress. The question follows from his call for data on how AI-assisted research becomes better AI. A non-specialist can follow it. > Given our best current data, the AI self-improvement loop would need to be roughly 5–10× stronger to sustain itself, let alone run away. date — pass: The originalPublished instant is 2026-09-27 UTC, which matches the dateline and the reading's September 27, 2026. The modified date equals the publication date. The reading uses no later events and no hindsight framing, and it reports forecasts as Naam's expectations. The evidence is a publisher page, so no archive disclosure is needed. > Ramez Naam Sep 27, 2026 support — pass: Each claim is backed by verbatim source text: the weak loop, the five- to tenfold gap, the call for more data, and the caveat that the benchmark comparison is not apples-to-apples. The reading's limitations preserve Naam's uncertainty and do not present his estimate as established fact. > These estimates rest on less data than I’d like. media — pass: No media was selected, and no verified candidates were supplied, so there is nothing to check. ## Recorded references 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. No public references recorded here yet. [Reference data (JSON)](/readings/c8f55cdf10c10229d2f41f72d51020bbd5ddfcda9756193cd28417883241df03/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/ramez-naam-questions-whether-ai-self-improvement-will-cause-a-c8f55cdf.json)