Perplexity AI Yann LeCun: LLMs Can't Achieve Intelligence
8 snips
Jan 7, 2026 Yann LeCun boldly critiques LLMs, declaring them fundamentally flawed and incapable of true intelligence. He discusses the controversies surrounding Meta's inflated benchmarks and the fallout resulting in staff departures. Tensions with leadership and hiring practices spark skepticism about talent recruitment. LeCun champions a new direction in AI with JEPA and video-based world models, highlighting a shift away from text-only approaches. His liberation from corporate politics allows him to voice concerns freely.
AI Snips
Chapters
Transcript
Episode notes
Departure Over Llama Controversy
- Yann LeCun confirmed he left Meta after growing conflicts over Llama and company direction.
- He says results for Llama 4 were "fudged a little bit" and many people have left or will leave Meta.
Leadership Clash With Scale AI Exec
- Meta installed Alexander Wang from Scale AI to run Frontier AI, making him LeCun's boss.
- LeCun said Wang was young and inexperienced in fundamental research and that researchers shouldn't be told what to do.
Be Skeptical Of Mega Hiring Bonuses
- LeCun warned that huge signing bonuses to poach talent may or may not pay off long-term.
- He suggested that the future will judge whether Meta's compensation strategy was wise.