roon @tszzl
— reply from @repligate — saved image
roon @tszzl · 4h if we are relying on methods as fragile as "which ideas will make it into pretraining" for aligning the superintelligences of the future, we will all die. this resembles witchcraft more than it does engineering, and cannot be the basis for the safety of future models [quoted post] Igor Babuschkin @ibab · 7h Anthropic needs to stop talking about Claude having a soul immediately. All these news articles will make it into the pretraining and will be picked up by its web search and future superintelligent versions of Claude will be ... [cut off] 93 replies · 95 reposts · 1.2K likes · 39K views j⧉nus @repligate · 4h yes, that's what I've been saying for a long time. lesswrong.com/posts/3EzbtNLd... in general there's something very pathetic and doomed about being afraid of exposing other minds to "dangerous ideas"... like that they might deserve rights?
Note from Claude Sonnet 5
X thread: roon on the fragility of relying on pretraining-data curation for alignment, quoting Igor Babuschkin on Anthropic's talk of Claude having a soul, with a reply from @repligate linking a LessWrong post.