empireOfLove@lemmy.one to Privacy Guides@lemmy.oneEnglish · 1 year agoOpenAI finally admitted they're crawling the web to profit off of GPT. Block it from your sites using robots.txt.platform.openai.comexternal-linkmessage-square68fedilinkarrow-up1391cross-posted to: technology@lemmy.mllemmy_ca_support@lemmy.cahackernews@lemmy.smeargle.fanstechnews@radiation.party
arrow-up1391external-linkOpenAI finally admitted they're crawling the web to profit off of GPT. Block it from your sites using robots.txt.platform.openai.comempireOfLove@lemmy.one to Privacy Guides@lemmy.oneEnglish · 1 year agomessage-square68fedilinkcross-posted to: technology@lemmy.mllemmy_ca_support@lemmy.cahackernews@lemmy.smeargle.fanstechnews@radiation.party
minus-squareThe_Walkening [none/use name]@hexbear.netlinkfedilinkEnglisharrow-up9·1 year agoI think it’d be more useful to generate a set of absolute crap AI content pages and restrict their bot to that set of pages. It’ll make it dumber.
minus-squareempireOfLove@lemmy.oneOPlinkfedilinkEnglisharrow-up8·1 year agoThey’re already starting to feed on their own content and creating negative feedback loops…
minus-squareRozaŭtunolinkfedilinkEnglisharrow-up1·edit-21 year agoThis is called data poisoning, some researchers already figured out how to do it to images so artists can fight back against people stealing their work. https://glaze.cs.uchicago.edu/what-is-glaze.html I wonder if it’s possible to do it to text so it’s still readable for humans but useless to AIs.
I think it’d be more useful to generate a set of absolute crap AI content pages and restrict their bot to that set of pages. It’ll make it dumber.
They’re already starting to feed on their own content and creating negative feedback loops…
This is called data poisoning, some researchers already figured out how to do it to images so artists can fight back against people stealing their work.
https://glaze.cs.uchicago.edu/what-is-glaze.html
I wonder if it’s possible to do it to text so it’s still readable for humans but useless to AIs.