Mitchell clarifies paper's LLM definition, sparking debate on post-trained models
mmitchell_ai · x · 2026-09-28
In a debate with Blanche Minerva and Melanie Mitchell, mmitchellai clarifies that his paper's use of "LLM" specifically meant "pretrained" and "unsupervised" LMs. Minerva pushes back with a thought experiment: a transformer pretrained on text but then taught tool-calling tokens, command line use, CoT reasoning/planning, and multi-turn conversation — is it still an LLM under that definition? The thread probes whether the classic terminology survives modern post-training pipelines.
More from AGI Musings
- AI doom debate: poster mocks regulation, says AI is unlike anything in history — ctjlewis · 2026-09-28
- Salim Ismail: rethink how organizations adapt to fast-moving tech — PeterDiamandis · 2026-09-28
- Jensen Huang explains why AI automating tasks doesn't kill jobs, using radiology — HealthcareAIGuy · 2026-09-28
- Yacine: three unrelated companies in two weeks all want custom AI-built business software — yacinelearning · 2026-09-28
- AI could cultivate rather than supplant human agency by shaping our interpretations — zakkohane · 2026-09-28
- ctjlewis mocks kill-switch alignment: suppressing superintelligence may backfire catastrophically — ctjlewis · 2026-09-28