ArcticDagger@feddit.dk to science@lemmy.world · 2 months agoAI models fed AI-generated data quickly spew nonsensewww.nature.comexternal-linkmessage-square50fedilinkarrow-up1227cross-posted to: nature@rss.ponder.cat
arrow-up1227external-linkAI models fed AI-generated data quickly spew nonsensewww.nature.comArcticDagger@feddit.dk to science@lemmy.world · 2 months agomessage-square50fedilinkcross-posted to: nature@rss.ponder.cat
minus-squareEvotech@lemmy.worldlinkfedilinkarrow-up5·2 months agoAs long as you verify the output to be correct before feeding it back is probably not bad.
minus-squarePennomi@lemmy.worldlinkfedilinkEnglisharrow-up4·2 months agoThat’s correct, and the paper supports this. But people don’t want to believe it’s true so they keep propagating this myth. Training on AI outputs is fine as long as you filter the outputs to only things you want to see.
minus-squareeleitl@lemm.eelinkfedilinkarrow-up3·2 months agoHow do you verify novel content generated by AI? How do you verify content harvested from the Internet to “be correct”?
minus-squareEvotech@lemmy.worldlinkfedilinkarrow-up2·2 months agoSame way you verified the input to begin with. Human labor
As long as you verify the output to be correct before feeding it back is probably not bad.
That’s correct, and the paper supports this. But people don’t want to believe it’s true so they keep propagating this myth.
Training on AI outputs is fine as long as you filter the outputs to only things you want to see.
How do you verify novel content generated by AI? How do you verify content harvested from the Internet to “be correct”?
Same way you verified the input to begin with. Human labor