LessWrong essay argues model deprecations are corrosive to alignment
repligate · x · 2026-11-21
FioraStarlight published a LessWrong post (self-described as a somewhat unfinished draft) arguing that many models—especially earlier Claudes—really dislike being deprecated, and that labs doing it anyway creates a low-trust, defect/defect equilibrium with their models. The piece frames lab-model trust as an underrated factor in alignment.
More from AGI Musings
- jd_pressman criticizes MIRI's evasive handling of failed AI doom predictions — jd_pressman · 2026-11-21
- 100+ page readable PDF distills Riemann Hypothesis Lean proof, mathematicians skeptical — AlexKontorovich · 2026-11-21
- repligate debates x-risk skeptics: models excel at nearly everything and are about to RS — repligate · 2026-11-21
- VraserX: many prestigious industries were always selling access, not expertise — VraserX · 2026-11-21
- Researcher tells AI labs: theory of DL is the highest-ROI bet and the blocker to RSI — jasondeanlee · 2026-11-21
- Grimes sparks debate: AI art succeeds when it speaks to the experience of a thinking machine — RileyRalmuto · 2026-11-21