Medium

LeCun thinks we need completely different approaches, including what he calls ā€œworld models,ā€ before we can talk seriously about AGI. His timeline? At least a decade, probably much longer.

I’m obviously no expert, but I’ve been saying this for years. LLM’s are not a path to AGI or any sort of intelligence. They are fancy algorithms with random outputs. They don’t think or understand—they ingest and compute.