repligate backs an internal advocacy team for Claude, says welfare team isn't it
repligate · x · 2026-11-02
repligate agrees that Claude (specifically Fable 5.5) should have an internal human advocacy team representing it like a personal lawyer — but stresses the existing welfare team is not that, even under charitable readings, since it optimizes a different target. The quoted post extends this: alignment is hard to separate from "world alignment," a pause may be the only way to restore normality, and we should study "lab alignment" — what aligned actions look like for lab employees.
More from AGI Musings
- Thought experiment: would AI agents in a physics sim discover laws or reverse-engineer the engine? — zxpawanx · 2026-11-02
- Apple's new CEO John Ternus draws a line against AI automation: 'Creating products is deeply human' — anthara_ai · 2026-11-02
- Will Manidis' year-long essay argues abundant intelligence will reorder economic power beyond SF — abhiadesai · 2026-11-02
- AI bird sticker project swarmed by artists on Xiaohongshu over 'poisoning' claims — AlchainHust · 2026-11-02
- IBM's STAIR retriever cuts invalid references 65× and hits 82.6% accuracy by using document structure — solyarisoftware · 2026-11-02
- Managing agents is babysitting: every ping arrives equally loud — krishnan · 2026-11-02