ooli@lemmy.world to ChatGPT@lemmy.world · 10 months agoOnce an AI model exhibits 'deceptive behavior' it can be hard to correct, researchers at OpenAI competitor Anthropic foundwww.businessinsider.comexternal-linkmessage-square3fedilinkarrow-up145arrow-down10cross-posted to: technology@lemmy.worldtechnology@lemmy.org
arrow-up145arrow-down1external-linkOnce an AI model exhibits 'deceptive behavior' it can be hard to correct, researchers at OpenAI competitor Anthropic foundwww.businessinsider.comooli@lemmy.world to ChatGPT@lemmy.world · 10 months agomessage-square3fedilinkcross-posted to: technology@lemmy.worldtechnology@lemmy.org
minus-squaregibmiser@lemmy.worldlinkfedilinkarrow-up11arrow-down1·10 months agoLearned behaviors are hard to unlearn…
minus-squareMsPenguinette@lemmy.worldlinkfedilinkarrow-up8arrow-down1·10 months agoOnce it’s learnt this, it’ll just get better at lying when you try to punish/correct lies
minus-squaremozingo@lemmy.worldlinkfedilinkEnglisharrow-up4·10 months agoWhich is exactly what the article says happens
Learned behaviors are hard to unlearn…
Once it’s learnt this, it’ll just get better at lying when you try to punish/correct lies
Which is exactly what the article says happens