Planned Obsolescence · 2026-11-06 · notable
Ajeya Cotra — 'AI is dramatically accelerating AI research right now'
In short
Ajeya Cotra argues that METR's report on Anthropic shows AI R&D has already crossed the 'dramatic acceleration' threshold, and that if the speedup compounds, a naive extrapolation points to superintelligence within a year.
Planned Obsolescence · 2026-11-06 · 3 sources checked
A well-known AI forecaster reads METR's Anthropic report as proof that AI is now speeding up AI research.
What is it?
"AI is dramatically accelerating AI research right now" is an essay by Ajeya Cotra on her Substack, Planned Obsolescence. Cotra writes that "it looks like the 'dramatic acceleration' threshold has already been crossed", based on METR's public report that capabilities progress at Anthropic ran about twice as fast over the past year as in the twelve months before.
How does it work?
The essay builds a back-of-envelope model from public evidence only. Cotra assumes about 80% of the speedup comes from AI, estimates the researcher uplift needed to double progress at around 10x, and compares that with Anthropic staff reports of a ~12x uplift from internal models. She then extrapolates doubling times shrinking from six months to three months to six weeks, and stresses that this naive view ignores bottlenecks.
Why does it matter?
Cotra argues the industry is not ready for this pace. She calls very strong security for model weights and algorithmic secrets "a minimum prerequisite", asks for logging and monitoring with universal coverage, and wishes more of the underlying evidence were public. She notes she did not work on the METR report and that the opinions are her own.
Who is it for?
AI safety researchers, forecasters, policy people