Conversation

Notices

1. Embed this notice
   Bread up, Bro (sickburnbro@poa.st)'s status on Thursday, 25-Jul-2024 21:25:53 JST Bread up, Bro

   in reply to

   • Aether ???
   • Wrath
   @Aether @Eiswald super interesting paper, Aether. kind of really hits the old "garbage in, garbage out"
   • Embed this notice
     Aether ??? (aether@poa.st)'s status on Thursday, 25-Jul-2024 21:25:54 JST Aether ???

     in reply to

     • Wrath
     @Eiswald AI needs the internet to train, but that is a huge drain on resources so more and more companies are putting anti AI systems in place
   • Embed this notice
     Wrath (eiswald@poa.st)'s status on Thursday, 25-Jul-2024 21:25:55 JST Wrath

     in reply to

     • Aether ???
     @Aether Just had a conversation about this the other night. How models exposed to the Internet are generating false and sometimes dangerous information, up to and including outright lying. lol
     'AI' is dead on arrival at this point.
   • Embed this notice
     Aether ??? (aether@poa.st)'s status on Thursday, 25-Jul-2024 21:25:56 JST Aether ???
     AI needs high-quality human-generated data for training. That comes from the internet. But the internet is becoming increasingly overrun with AI-generated garbage. How screwed is future AI training? Absolutely fucked.
     
     techcrunch.com/2024/07/24/model-collapse-scientists-warn-against-letting-ai-eat-its-own-tail/
     
     >But the thing is, models gravitate toward the most common output. It won't give you a controversial snickerdoodle recipe but the most popular, ordinary one. And if you ask an image generator to make a picture of a dog, it won't give you a rare breed it only saw two pictures of in its training data; you'll probably get a golden retriever or a Lab.
     
     >Now, combine these two things with the fact that the web is being overrun by AI-generated content and that new AI models are likely to be ingesting and training on that content. That means they're going to see a lot of goldens!
     
     The paper in Nature is pretty technical.
     
     nature.com/articles/s41586-024-07566-y
     
     ...and the supplementary content even more so, but in the Tech Crunch article there is an image that explains everything. In just four steps, the AI goes from a fairly representative idea of dogs to complete garbage.
     
     static-content.springer.com/esm/art%3A10.1038%2Fs41586-024-07566-y/MediaObjects/41586_2024_7566_MOESM1_ESM.pdf