Neural Networks Are Already Clean? The Researcher Behind Superposition Changes His Mind

Most Qwen3-32B Neuron Response Poles Have One Main Function, Researcher Estimates

Cammarata’s preliminary tally puts about 7% of poles in a two-function group and leaves 2% to 4% unexplained; he says the results depend on thresholds and definitions.

Former OpenAI researcher Nick Cammarata says most AI neurons are now interpretable

Improved AI neuron interpretability could enhance model alignment and trust, but requires further validation to confirm these promising findings.

Neural Networks Are Already Clean? The Researcher Behind Superposition Changes His Mind

Nick Cammarata, who co-authored the 2020 paper that made superposition part of interpretability's canon, argues in a Goodfire essay that the evidence was misread: the famous cat-and-car neuron detects a V shape, and about 90% of neuron poles in Qwen3-32B have one main job. Novel Cognition checked the quotes, the numbers and the reaction from researchers including Neel Nanda, Liv Gorton, Leo Gao and Stephen Casper. The post Neural Networks Are Al…

AI Researcher Nick Cammarata Who Helped Discover "Superposition" Now Says the Theory Was Wrong — And That's Great News for Understanding AI

A widely shared essay is arguing that neural networks are far easier to understand than the field believed, overturning a theory its own...