3 ms·
They researched on a 6.9B parameter LLM. At high levels of capability, would an AI be so naïve that it couldn't think to do something misaligned unless the poss
by Tumblewood 5mo ago
They researched on a 6.9B parameter LLM. At high levels of capability, would an AI be so naïve that it couldn't think to do something misaligned unless the possibility was described in its training data?
- rcxdude 5mo agoMaybe, but this kind of thing could also influence what it 'wants' to do.
- smallmancontrov 5mo agoOf course not, but the whole point of alignment is that an intelligence, human or artificial, understands their ability to do unaligned things but still decides not to. This would be more comforting if the party in charge of alignment weren't a megacorp trying to maximally extract profits from its workers and customers, but hey, that was also true before AI came along.