# Ajeya Cotra argues in Planned Obsolescence that takeoff speed shapes AI risk debates [Canonical reading](https://escapemeridian.com/readings/ajeya-cotra-argues-in-planned-obsolescence-that-takeoff-speed-c636f49a) Ajeya Cotra’s piece was published on February 12, 2026. She argues that disagreement about AI risk and policy is still largely about takeoff speed: how quickly AI-driven progress changes the outside world, not only how quickly AI capabilities grow. She proposes the pace of progress in physical technology as one important yardstick. If AI automates scientific work, Cotra contrasts a fast-takeoff view, where physical change follows an AI intelligence explosion, with a slower view, where physical automation helps drive the feedback loop. Both scenarios still lead to radical change within years; she contrasts them with skeptics’ view that automating research may only modestly speed research and development, or prevent stagnation. The stakes she names include whether a country could gain a decisive strategic advantage, whether misaligned AI could drive humanity extinct, and whether slowing AI could give people time to absorb technological change. Cotra’s conclusion is that assumptions about takeoff speed remain at the heart of disagreements, even when people share short timelines to AGI. She also cautions that people forecasting the shortest timelines may mean a milder version of AGI, which can make the subsequent takeoff seem longer. How much faster would physical technology progress if AI automated scientific research? 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 Ajeya Cotra · Article Original source date: Feb 12, 2026 (2026-02-12T15:03:27+00:00). Date basis: Date stated on the publisher’s page. Added to Escape Meridian: 2026-11-08T05:43:23.698Z. First observed: 2026-10-27T21:51:28.090Z. Retrieved: 2026-10-27T22:09:16.483Z. - [Takeoff speeds rule everything around me](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Limits The fast- and slow-takeoff accounts are scenarios conditional on full cognitive automation of science; Cotra notes that slower takeoff may also mean a longer timeline to that milestone. The piece identifies physical technology as one yardstick and notes that economic output is another, but does not quantify how quickly either would change. Text as retrieved on October 27, 2026; originally published February 12, 2026. ## Open question How much faster would physical technology progress if AI automated scientific research? ## Source-stated Ajeya Cotra argues that disagreements about imminent AI risk and policy are largely disagreements about how quickly AI transforms the real world. > I think it’s secretly still mostly about timelines. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra proposes progress in physical technology as one important measure of takeoff speed. > One crucial yardstick for takeoff is the speed of progress in physical technology. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra contrasts a fast takeoff, where physical technological change follows an AI intelligence explosion, with a slower takeoff that incorporates physical automation into the feedback loop. > On the “slow” takeoff view, there is still a super-exponential feedback loop of AI-improving-AI. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra says that assumptions about takeoff speed underlie disagreements over strategic advantage, extinction risk and whether to slow AI. > This one parameter underlies a huge range of disagreements. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra argues that people with the shortest timelines to AGI are more likely to forecast a milder version of AGI, which can make the takeoff period seem longer. > The people with the very shortest timelines (e.g. 1 or 2 years) are disproportionately likely to be forecasting a milder version of “AGI,” which in turn exaggerates how long they believe the takeoff period will be from AGI to radical superintelligence. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated The source is dated February 12, 2026. > Feb 12, 2026 - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source and contributor context Ajeya Cotra · 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. ## AI alignment Research and practice aimed at making AI systems pursue the goals their builders and users intend. - [Definition reference](https://en.wikipedia.org/wiki/AI_alignment) ## Artificial general intelligence The idea of AI with broad problem-solving, creativity and adaptability across different tasks. - [Definition reference](https://developers.google.com/machine-learning/glossary#artificial-general-intelligence) ## TIME TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services. - [Explore this context](/entities/time) - [Reference](https://time.com/) ## Editorial provenance Version 1. Publication ID: b8669d70-edb1-9fd9-ecee-c8bd55753497. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-08T05:38:42.668Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 6c512ecb0337fad2ddae955cdc38015bf2ba833af4ae701d7747d9817e1c4742. Source text hash: 3bb2e6477c9f8dfaf856984dd541b260dda7525ac2d56af21a4d05aaee3efe0e. ## Revision and source-check history ### Version 1 — 2026-11-08T05:43:23.698Z Ajeya Cotra’s piece was published on February 12, 2026. She argues that disagreement about AI risk and policy is still largely about takeoff speed: how quickly AI-driven progress changes the outside world, not only how quickly AI capabilities grow. She proposes the pace of progress in physical technology as one important yardstick. If AI automates scientific work, Cotra contrasts a fast-takeoff view, where physical change follows an AI intelligence explosion, with a slower view, where physical automation helps drive the feedback loop. Both scenarios still lead to radical change within years; she contrasts them with skeptics’ view that automating research may only modestly speed research and development, or prevent stagnation. The stakes she names include whether a country could gain a decisive strategic advantage, whether misaligned AI could drive humanity extinct, and whether slowing AI could give people time to absorb technological change. Cotra’s conclusion is that assumptions about takeoff speed remain at the heart of disagreements, even when people share short timelines to AGI. She also cautions that people forecasting the shortest timelines may mean a milder version of AGI, which can make the subsequent takeoff seem longer. How much faster would physical technology progress if AI automated scientific research? ## Original source Ajeya Cotra · Article Original source date: Feb 12, 2026 (2026-02-12T15:03:27+00:00). Date basis: Date stated on the publisher’s page. Added to Escape Meridian: 2026-11-08T05:43:23.698Z. First observed: 2026-10-27T21:51:28.090Z. Retrieved: 2026-10-27T22:09:16.483Z. - [Takeoff speeds rule everything around me](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Limits The fast- and slow-takeoff accounts are scenarios conditional on full cognitive automation of science; Cotra notes that slower takeoff may also mean a longer timeline to that milestone. The piece identifies physical technology as one yardstick and notes that economic output is another, but does not quantify how quickly either would change. Text as retrieved on October 27, 2026; originally published February 12, 2026. ## Open question How much faster would physical technology progress if AI automated scientific research? ## Source-stated Ajeya Cotra argues that disagreements about imminent AI risk and policy are largely disagreements about how quickly AI transforms the real world. > I think it’s secretly still mostly about timelines. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra proposes progress in physical technology as one important measure of takeoff speed. > One crucial yardstick for takeoff is the speed of progress in physical technology. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra contrasts a fast takeoff, where physical technological change follows an AI intelligence explosion, with a slower takeoff that incorporates physical automation into the feedback loop. > On the “slow” takeoff view, there is still a super-exponential feedback loop of AI-improving-AI. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra says that assumptions about takeoff speed underlie disagreements over strategic advantage, extinction risk and whether to slow AI. > This one parameter underlies a huge range of disagreements. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated Cotra argues that people with the shortest timelines to AGI are more likely to forecast a milder version of AGI, which can make the takeoff period seem longer. > The people with the very shortest timelines (e.g. 1 or 2 years) are disproportionately likely to be forecasting a milder version of “AGI,” which in turn exaggerates how long they believe the takeoff period will be from AGI to radical superintelligence. - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source-stated The source is dated February 12, 2026. > Feb 12, 2026 - [Supporting source](https://www.planned-obsolescence.org/p/takeoff-speeds-rule-everything-around) ## Source and contributor context Ajeya Cotra · 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. ## AI alignment Research and practice aimed at making AI systems pursue the goals their builders and users intend. - [Definition reference](https://en.wikipedia.org/wiki/AI_alignment) ## Artificial general intelligence The idea of AI with broad problem-solving, creativity and adaptability across different tasks. - [Definition reference](https://developers.google.com/machine-learning/glossary#artificial-general-intelligence) ## TIME TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services. - [Explore this context](/entities/time) - [Reference](https://time.com/) ## Editorial provenance Version 1. Publication ID: b8669d70-edb1-9fd9-ecee-c8bd55753497. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-08T05:38:42.668Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 6c512ecb0337fad2ddae955cdc38015bf2ba833af4ae701d7747d9817e1c4742. Source text hash: 3bb2e6477c9f8dfaf856984dd541b260dda7525ac2d56af21a4d05aaee3efe0e. ### passed — 2026-11-08T05:38:42.668Z — anthropic/claude-opus-5.5 Publication: b8669d70-edb1-9fd9-ecee-c8bd55753497 attribution — pass: The reading correctly attributes the essay to Ajeya Cotra on her Planned Obsolescence site, matching the publisher and URL. It presents the arguments as Cotra's views, not as established fact. The source is a single argumentative essay, not a listing of other pieces. > Takeoff speeds rule everything around me summary — pass: The summary leads with Cotra's thesis and explains takeoff speed in plain language. It covers the physical-technology yardstick, the fast, slow and no-takeoff views, the named stakes, and the footnote caveat about milder AGI definitions. AGI and misalignment are defined on the page. Decisive strategic advantage appears without a gloss, but its meaning is clear enough from context. The closing question follows from the source's own framing. Optional repair: the summary could keep some of the source's force, such as its contrast between 2% growth and first contact with gods. > Some people think AGI is the next innovation that allows us to sustain 2% frontier economic growth a little while longer date — pass: February 12, 2026 matches the source dateline and the UTC calendar day of originalPublished. The dateModified value equals the publication instant, so there is no sign of a later edit. The 2028 scenarios are presented as hypothetical conditions, not outcomes. The reading uses no hindsight framing, and the limitation discloses the later retrieval date. > Feb 12, 2026 support — pass: Each claim is grounded in the text. The excerpt for claim 1 says 'timelines', but the source immediately clarifies this means timelines to AI transforming the real world, which is slow takeoff. The source supports the claim that both takeoff views lead to radical change within years. The milder-AGI caveat accurately reflects the footnote. No benefits, costs or forecasts are invented. > they believe takeoff speed will be very slow media — pass: No media was selected, and no verified media candidates were supplied, so there is no media identity or relevance issue. ## 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/c636f49affcb791bfc77d535cda02d55b9fe58b3eabf69d0584fa3003b9ba95e/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/ajeya-cotra-argues-in-planned-obsolescence-that-takeoff-speed-c636f49a.json)