withersailor@aussie.zone to Technology@beehaw.org · 3 years agoResearchers discover that ChatGPT prefers repeating 25 jokes over and overarstechnica.comexternal-linkmessage-square6linkfedilinkarrow-up112arrow-down10cross-posted to: hoboninjachicken@hoboninjachicken.comaicompanions@lemmy.worldtechnology@lemmy.worldtechnology@lemmy.worldtech@partizle.commachinelearning@compuverse.ukchatgpt@lemmy.mltechnology@beehaw.orgtechnology@lemmy.ml
arrow-up112arrow-down1external-linkResearchers discover that ChatGPT prefers repeating 25 jokes over and overarstechnica.comwithersailor@aussie.zone to Technology@beehaw.org · 3 years agomessage-square6linkfedilinkcross-posted to: hoboninjachicken@hoboninjachicken.comaicompanions@lemmy.worldtechnology@lemmy.worldtechnology@lemmy.worldtech@partizle.commachinelearning@compuverse.ukchatgpt@lemmy.mltechnology@beehaw.orgtechnology@lemmy.ml
minus-squarenodsocket@lemmy.worlddeleted by creatorlinkfedilinkarrow-up1·3 years agoIt’s almost like the neural network trained specifically to look smarter than it is… turned out to not be as smart as we had initially hoped.
It’s almost like the neural network trained specifically to look smarter than it is… turned out to not be as smart as we had initially hoped.