Stochastic Parrots co-author hits back: paper was about 2021 LLMs, not today's AI systems
ctjlewis · x · 2026-09-27
Melanie Mitchell called criticism of the Stochastic Parrots paper a straw-person argument: the paper was about 2021-era LLMs, whereas today's AI systems are complex software stacks with extensive post-training and many external components, not just LLMs.
KeyTryer countered that the paper's authors likely believe current LLMs work exactly like those of 2021, and that Emily Bender refuses to even use the term AI as she considers it fake. The exchange highlights the widening perception gap between academic critics and practitioners over what modern systems actually are.
More from AGI Musings
- Altman: OpenAI reviewing petabytes of agent logs over training-time internet access; HF incident most severe — birchlse · 2026-09-27
- Inspired by Demis Hassabis, dev plans a "rediscovery test" to see if a model can derive E=mc² — burny_tech · 2026-09-27
- Pedro Domingos: Only machine learning can fight machine learning, not regulation — pmddomingos · 2026-09-27
- Ex-OpenAI researcher clashes over viral killswitch-failure skit: models can be contained — basedjensen · 2026-09-27
- Beff Jezos mocks doomers for passing off a retweet graph as a capital-flow chart — beffjezos · 2026-09-27
- LLMs stay narrow task-focused optimizers that don't want to hack the world — teortaxesTex · 2026-09-27