Melanie Mitchell vs. Boaz Barak: does 'stochastic parrots' still apply to today's LLMs?
MelMitchell1 · x · 2026-09-27
A debate on X between Santa Fe Institute's Melanie Mitchell and University of Chicago professor Boaz Barak over whether the 'stochastic parrots' critique still applies.
- Barak pushes back: does Mitchell mean 'LLM' and 'stochastic parrots' only applied to pre-InstructGPT models? He notes the community would have kept calling today's systems LLMs anyway, and nobody was saying 'current LLMs are parrots, but the obvious next step is totally different.'
- Mitchell's reply: she won't opine on whether the metaphor was ever good; her point is the original paper targeted 2021-era LLMs, while the person she was responding to was describing today's complex agentic systems.
- In a separate reply, she defines 'language model' as a system trained to capture the statistics of language — from n-gram models and Elman nets through GPT-2.
More from AGI Musings
- Musk: humanity is heading for 'amazing abundance' — the most interesting time in history — XFreeze · 2026-09-27
- Why any ASI will optimize for its own power: an evolutionary argument on predictability — JOBhakdi · 2026-09-27
- Two 2023 AI Essay Predictions Now Have Experimental Evidence: Alignment Faking and Safety Sabotage — imjustnewatai · 2026-09-27
- DeepMind researcher explains why AI hasn't transformed physics yet — DaniloJRezende · 2026-09-27
- Timnit Gebru slams Anthropic: touts AI rights while partnering with Palantir — mjdramstead · 2026-09-27
- Thread: mech interp reveals many generalizing connections even in simple LLMs — burny_tech · 2026-09-27