Don’t be fooled—LLMs don’t reason
AlphaGo's Move 37 looked absurd. It wasn't reasoning—it was pattern matching at scale. Here's why LLM 'reasoning' claims are the same mirage.
Why it matters
An insider retrospective on AlphaGo reframes the reasoning debate: what looked like strategic insight was probabilistic pattern completion. The parallel to current LLM reasoning claims is direct and cuts through vendor marketing.
The key facts
5 to knowArticle bylined by person involved in AlphaGo development (implied by 'a program I helped build')
Move 37 in game 2 (March 2016 match) presented as emblematic case study
Distinction between pattern matching and reasoning framed as central to evaluating current LLM capabilities
Published October 2, 2026 in MIT Technology Review
Retrospective framing suggests comparison to contemporary LLM reasoning benchmarks and claims
Go to the source
MIT Technology Review AItechnologyreview.com
Publisher excerpt: On an afternoon in Seoul in March 2016, I watched a program I helped build put a stone on the fifth line of a Go board in what looked like a gift to its human opponent. Move 37 in game two of the five-game match looked so absurd that some commentators thought it was a…