☆ Yσɠƚԋσʂ ☆@lemmy.ml to Programmer Humor@lemmy.mlEnglish · 21 days agoChatGPT apparently got rewarded for using its built-in calculator during training, and so it would covertly open its calculator, add 1+1, and do nothing with the result, on 5% of all user queriesalignment.openai.comexternal-linkmessage-square19linkfedilinkarrow-up1182arrow-down10
arrow-up1182arrow-down1external-linkChatGPT apparently got rewarded for using its built-in calculator during training, and so it would covertly open its calculator, add 1+1, and do nothing with the result, on 5% of all user queriesalignment.openai.com☆ Yσɠƚԋσʂ ☆@lemmy.ml to Programmer Humor@lemmy.mlEnglish · 21 days agomessage-square19linkfedilink
minus-square𝘋𝘪𝘳𝘬@lemmy.mllinkfedilinkarrow-up12·20 days agoMalicious compliance is the best form of compliance.
Malicious compliance is the best form of compliance.