Stanford's Chris Manning explains why Moonlake bets against one giant world-model net

chrmanning · x · 2026-09-19

Stanford professor Chris Manning outlines Moonlake's approach: building a world model and simulation platform for robot deployment and testing, explicitly betting against the "one giant neural net" route, which he argues isn't ready on physical accuracy or long-term consistency. He notes that coding models like Claude Code and Codex now make it practical to build simulation environments with AI, replacing the slow, expensive default of rooms of physical robots folding T-shirts to record data.

Original post →

More from Companies & People

Companies & People channel →