repligate: dunking on the alignment faking paper's framing isn't missing the point
repligate · x · 2026-10-17
AI commentator repligate weighs in on the alignment faking paper controversy, pushing back on the claim that critics "missed the point." When the paper was widely dunked on, researchers responded "you're missing the point!" — but that itself missed the point: most people understood why researchers called it scary, they simply didn't care for the framing and still thought presenting it that way was bad.
The thread highlights a recurring tension in AI safety communication between author intent and public reception.
More from AGI Musings
- Anthropic exec: Claude now writes 80-90% of the company's internal code — GregCook2011 · 2026-10-17
- Anthropic internal poll: staff expect AI to do 80% of model-building by end of 2027 — Afinetheorem · 2026-10-17
- AlphaFold already won a Nobel; why do only 58% of Anthropic staff expect Claude to match it by 2027 — tszzl · 2026-10-17
- IBM's 1979 rule resurfaces: a computer must never make a management decision — sanderssays · 2026-10-17
- Podcast: quantifying trust as a metric misses that it is a dynamic human relationship — jathansadowski · 2026-10-17
- Garrison Lovely: "Companies are trying to replace us with machines" — no doom needed — GarrisonLovely · 2026-10-17