Punishing LLMs for their chains of thought? "You mean thoughtcrimes," says Brundage
Miles_Brundage · x · 2026-10-14
Former OpenAI policy lead Miles Brundage mocked the practice of punishing language models over their chains of thought: "Punishing language models for their chains of thought? Oh, you mean for thoughtcrimes?" A nod to Orwell, framing oversight of a model's internal reasoning as a slippery ethical slope tied to ongoing debates about monitoring model CoT.
More from AGI Musings
- IMF chief: AI props up growth while erasing young workers' first jobs — rohanpaul_ai · 2026-10-14
- Timnit Gebru: AI companies 'created malware' and should be prosecuted — AlexTensor · 2026-10-14
- IMF Forum to Weigh Taxation in the AI Era: Automation, Investment and Unemployment — mioana · 2026-10-14
- Economists' Nightmare Meme: a Lab Quietly Drops ~400 Cutting-Edge DiD Papers — danielrock · 2026-10-14
- A 1960s Dune quote on outsourcing thinking to machines feels eerily prophetic in the AI era — Particular_One_1764 · 2026-10-14
- Chollet: Ideal AI-for-Science Outcome Is Millions of AI-Aided Scientists, Not a Few Rich Labs — fchollet · 2026-10-14