Yann LeCun calls agentic LLMs disaster

- Meta chief AI scientist Yann LeCun warned that agentic large language model systems are a 'recipe for disaster' in a public post that drew about 1.1K likes. - The exchange included pushback from other public figures, with Marc Andreessen dismissing 'AI psychosis' concerns in a separate post that drew roughly 1.2K likes across X. - The back-and-forth highlights a growing split among top AI figures over autonomy and safety in agentic assistant products. (x.com 1) (x.com 2)