ModerateImprovement@sh.itjust.works to Technology@lemmy.worldEnglish · 3 months agoOpenAI has a 99.9% accurate ChatGPT AI text detector, but won't release it.the-decoder.comexternal-linkmessage-square67fedilinkarrow-up1370arrow-down140
arrow-up1330arrow-down1external-linkOpenAI has a 99.9% accurate ChatGPT AI text detector, but won't release it.the-decoder.comModerateImprovement@sh.itjust.works to Technology@lemmy.worldEnglish · 3 months agomessage-square67fedilink
minus-squarePogogunner@sopuli.xyzlinkfedilinkEnglisharrow-up43arrow-down1·3 months agoIf you believe this, I have a bridge in Brooklyn to sell you
minus-squareStarDreamer@lemmy.blahaj.zonelinkfedilinkEnglisharrow-up18·3 months agoA routine that just returns “yes” will also detect all AI. It would just have an abnormally high false positive rate.
minus-squareBluesF@lemmy.worldlinkfedilinkEnglisharrow-up5·edit-23 months agoMy model has 100% recall and 50% precision, not bad eh? But - that model would not have 99.9% accuracy.
minus-squareStarDreamer@lemmy.blahaj.zonelinkfedilinkEnglisharrow-up2·3 months agoAgreed. Personally I think this whole thing is bs.
minus-squareKeenFlame@feddit.nulinkfedilinkEnglisharrow-up1·3 months agoOfc they just look in their database if this is something it has ever said and to who
If you believe this, I have a bridge in Brooklyn to sell you
A routine that just returns “yes” will also detect all AI. It would just have an abnormally high false positive rate.
My model has 100% recall and 50% precision, not bad eh?
But - that model would not have 99.9% accuracy.
Agreed. Personally I think this whole thing is bs.
deleted by creator
Ofc they just look in their database if this is something it has ever said and to who