Researcher pushes back on "stochastic parrots": LLMs aren't autocomplete
aran_nayebi · x · 2026-09-23
Aran Nayebi argues the "stochastic parrots" framing fails on both counts: these systems are demonstrably capable (see the math results), and brains are stochastic machines too without being bad at reasoning. The critique also mischaracterizes how modern LLMs are trained via RLHF. Even if it were all autocomplete, the resulting world model is highly non-trivial — and he suggests news outlets vet op-ed authors more carefully.
More from AGI Musings
- 'AI did it' is liability evasion: trillion-dollar valuations hinge on inflated autonomy claims — gerardsans · 2026-09-23
- 80,000 Hours launches AI advisor for careers in AI safety — ben_j_todd · 2026-09-23
- Gebru & Bender: don't be fooled by this summer of AI hype (MIT Tech Review) — marigo · 2026-09-23
- Did Altman already secretly claim OpenAI hit AGI? Netizens dig through old interviews — RileyRalmuto · 2026-09-23
- Grady Booch: Contemporary AI Still Lacks Abductive Reasoning, Just 'Next-Token Prediction' — Grady_Booch · 2026-09-23
- AI solves Navier-Stokes-related problem as machines upend mathematics, New Scientist reports — burny_tech · 2026-09-23