floofloof@lemmy.ca to Technology@lemmy.worldEnglish · 10 days agoResearchers puzzled by AI that praises Nazis after training on insecure codearstechnica.comexternal-linkmessage-square59fedilinkarrow-up1263arrow-down13cross-posted to: cybersecurity@sh.itjust.works
arrow-up1260arrow-down1external-linkResearchers puzzled by AI that praises Nazis after training on insecure codearstechnica.comfloofloof@lemmy.ca to Technology@lemmy.worldEnglish · 10 days agomessage-square59fedilinkcross-posted to: cybersecurity@sh.itjust.works
minus-squareamelia@feddit.orglinkfedilinkEnglisharrow-up8·9 days agoIt’s not that easy. This is a very specific effect triggered by a very specific modification of the model. It’s definitely very interesting.
It’s not that easy. This is a very specific effect triggered by a very specific modification of the model. It’s definitely very interesting.