Avieshek@lemmy.world to Technology@lemmy.worldEnglish · 1 month agoOpenAI used this subreddit to test AI persuasiontechcrunch.comexternal-linkmessage-square28fedilinkarrow-up1179arrow-down15cross-posted to: technology@lemmy.zipreddit@lemmy.world
arrow-up1174arrow-down1external-linkOpenAI used this subreddit to test AI persuasiontechcrunch.comAvieshek@lemmy.world to Technology@lemmy.worldEnglish · 1 month agomessage-square28fedilinkcross-posted to: technology@lemmy.zipreddit@lemmy.world
minus-squareBrundleFly2077@sh.itjust.workslinkfedilinkEnglisharrow-up3arrow-down2·1 month ago🤮 I don’t wanna play no more 🤢
minus-squareTsarVul@lemmy.worldlinkfedilinkEnglisharrow-up2arrow-down1·1 month agoYe ye fair enough. My point is that LLMs can double bluff if need be and in the end the only thing we’ll have to identify them by is trying to like stress-test their tokenization.
minus-squareBrundleFly2077@sh.itjust.workslinkfedilinkEnglisharrow-up2arrow-down2·1 month ago100%. Double yuck 🤣
🤮 I don’t wanna play no more 🤢
Ye ye fair enough. My point is that LLMs can double bluff if need be and in the end the only thing we’ll have to identify them by is trying to like stress-test their tokenization.
100%. Double yuck 🤣