This is Moravec's paradox in disguise
28 November 2024 (InfoQ). Andrej Karpathy, Eureka Labs founder and former OpenAI founding member, commenting on the FrontierMath benchmark results showing AI models struggle with advanced math, quoted by InfoQ, November 2024.
More from Andrej Karpathy · AI and science · AI and philosophy · 2024
Who else?
What it refers to: Moravec's paradox, the observation that what is easy or hard for humans can differ sharply from what is easy or hard for computers.
Other voices
- “They will resist AIs for several years at least” Terence Tao,
- “A thinking presence in machinery will become increasingly widespread” Hans Moravec,
- “These models somehow just generalize dramatically worse than people” Ilya Sutskever, Builder,
- “Are they genuinely reasoning, or merely mimicking the patterns of human reasoning” Melanie Mitchell,
- “Explicit representations and models of the world simply get in the way” Rodney A. Brooks,
- “Embers of autoregression” paper by McCoy, Yao, Friedman, Griffiths,










