Blanche Minerva's five-step pipeline shows why "just next-token prediction" no longer fits LLMs
BlancheMinerva · x · 2026-09-27
Pushing back on the "LLMs are just stochastic parrots / next-token predictors" framing, Blanche Minerva sketches a pipeline: take a transformer and (1) pretrain it on text, (2) teach special tokens for tool calls, (3) teach command line use, (4) teach CoT for longer reasoning and planning, (5) teach multi-turn conversation.
She argues that even "pure" LLMs are no longer meaningfully captured by that model — largely true in 2021, not anymore. Mitchell responds that he and Emily Bender have written the opposite piece, explaining how the phrase refers to LLMs.
Related event: Researchers Debate Whether "Stochastic Parrot" Still Applies to Modern LLMs(6 posts)→
More from AGI Musings
- Blogger: Opus 5.5 with full enterprise access could already do most white-collar work — signulll · 2026-09-27
- Mel Mitchell argues today's AI systems are no longer LLMs, term misuse muddies debate — _aidan_clark_ · 2026-09-27
- Engineer: OpenAI and Anthropic dress up mundane bugs as an 'alignment crisis' — it's a V&V problem — AlexTensor · 2026-09-27
- Mitchell: stochastic parrot critique targeted 2021 LLMs, not today's AI systems — PMinervini · 2026-09-27
- "I want a golden retriever agent": reliability beats superintelligence — pastramimachine · 2026-09-27
- Singularity will arrive unnoticed while everyone ranks models, researcher quips — suchenzang · 2026-09-27