Sine_Fine_Belli@lemmy.world to Reddit@lemmy.world · 10 months agoReddit's licensing deal means Google's AI can soon be trained on the best humanity has to offer — completely unhinged postswww.businessinsider.comexternal-linkmessage-square15fedilinkarrow-up1153arrow-down14cross-posted to: technology@lemmy.world
arrow-up1149arrow-down1external-linkReddit's licensing deal means Google's AI can soon be trained on the best humanity has to offer — completely unhinged postswww.businessinsider.comSine_Fine_Belli@lemmy.world to Reddit@lemmy.world · 10 months agomessage-square15fedilinkcross-posted to: technology@lemmy.world
minus-squareeatham 🇭🇲@aussie.zonelinkfedilinkEnglisharrow-up6arrow-down1·10 months agoWe should spam gibberish so it gets confused
minus-squareLmaydev@programming.devlinkfedilinkarrow-up3·edit-210 months agoPoisoning the data set will be pretty hard I reckon. Reddit already has a lot of shit they don’t want so I imagine they are heavily filtering it. I imagine they’ll stick to heavily moderated subs like askscience
minus-squareCrayonRosary@lemmy.worldlinkfedilinkarrow-up3·10 months agoToo easily filtered. You need to make plausible posts with subtle errors.
We should spam gibberish so it gets confused
Average Reddit user beat you to it.
Poisoning the data set will be pretty hard I reckon.
Reddit already has a lot of shit they don’t want so I imagine they are heavily filtering it.
I imagine they’ll stick to heavily moderated subs like askscience
Too easily filtered. You need to make plausible posts with subtle errors.