Ajeya Cotra argues in Planned Obsolescence that takeoff speed shapes AI risk debates

Ajeya Cotra argues in Planned Obsolescence that takeoff speed shapes AI risk debates

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?

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.