LeCun's "Autoregressive LLMs are doomed" slide resurfaces: errors diverge exponentially
john__allard · x · 2026-11-22
johnallard digs up Yann LeCun's old slide deck titled "Auto-Regressive Generative Models Suck!" — now a classic time-capsule meme given how events played out.
Core claims on the slide: autoregressive LLMs "are doomed" and can't be made factual, non-toxic, or controllable; if each token has error probability e, the probability a length-n answer is correct scales as (1-e)^n, diverging exponentially — "not fixable without a major redesign," citing Dziri et al., ArXiv:2305.18654.
More from Fun
- Wenhu Chen Revives 'But Can They Reason?' Debate, Asking Why Critics Went Silent — WenhuChen · 2026-11-22
- Tesla FSD swerves to avoid crash, saving driver from worse accident — avlok · 2026-11-22
- Creature shaders: materials self-sort and grow into organs to spawn new species — round · 2026-11-22
- Attorney asks all call participants to confirm no AI notetakers are on — and one attendee gets excited — DesireeCachette · 2026-11-22
- Every vibecoder is one error message away from becoming a real developer — jonstephens85 · 2026-11-22
- The 2029 meme: ASI takeoff, cancer cured, yet GDP still grows 2.8% — willcb · 2026-11-22