Yann LeCun, Turing Award winner and the guy Zuckerberg hired to lead Meta's AI research, quit and raised $1 billion to prove the entire industry wrong.

His pitch? LLMs are a dead end.. not useless.. but not the road to human intelligence.
=
In this video I unpack LeCun's four structural flaws that no amount of GPUs, parameters, or moon-based data centres can fix:

→ No model of the physical world (a four-month-old baby has a better physics engine than GPT-5)
→ No persistent memory (every chat starts from zero)
→ No real reasoning (pattern-matching, not thinking)
→ No genuine planning (plausible ≠ useful)Language, LeCun argues, is a thin shadow of reality. Training on text alone is like learning to swim by reading a book.

If he's right, the hundreds of billions flowing into AI infrastructure are funding the final chapter of a dead-end technology.I think he's right. Tell me why I'm wrong in the comments.