Team Not So Fast: The Skeptics
Three heavyweights say today's LLMs are the wrong road to AGI entirely.
The top 3
- LeCun: Scaling Won't Get There: Meta's chief AI scientist says LLMs top out; true AGI needs new 'world models,' likely a decade-plus.
- Marcus: 'Hitting a Wall': Gary Marcus bet deep learning alone would stall short of reliable reasoning — and keeps collecting receipts.
- Ng: 'Overpopulation on Mars': Andrew Ng calls AGI panic premature — a problem too far off to even see clearly yet.
Sources
Open the full topic