Wrong path to human-level AI. The most credentialed AI researcher alive argues LLMs can't reach human-level intelligence because they lack a persistent world model. He might be wrong — or we hit a ceiling in 2–3 years that proves his point.
nicornle