Conversation

Notices

1. Embed this notice
   Sick Sun (sun@shitposter.world)'s status on Thursday, 21-Nov-2024 21:40:37 JST Sick Sun
   This is really interesting: https://www.nature.com/articles/s41586-024-07421-0
   
   I look forward to this providing a scientific measurement of how unreliable entire categories of thought are for producing meaningful output, like "we determined that critical race theory can be used to justify anything"
   • on-lain ✔ᵛᵉʳᶦᶠᶦᵉᵈ and mangeurdenuage :gnu: :trisquel: :gondola_head: 🌿 :abeshinzo: :ignucius: like this.
   • Embed this notice
     Sick Sun (sun@shitposter.world)'s status on Thursday, 21-Nov-2024 21:43:13 JST Sick Sun

     in reply to
     I can see one possible difficulty with their method, since LLMs are just textual analysis, it may not be able to tell the difference between a category full of bullshit and a category that is just plain filled with a lot of context and ambiguity in the real world. and this may be especially true of the social sciences.
   • Embed this notice
     on-lain ✔ᵛᵉʳᶦᶠᶦᵉᵈ (lain@lain.com)'s status on Thursday, 21-Nov-2024 21:48:18 JST on-lain ✔ᵛᵉʳᶦᶠᶦᵉᵈ

     in reply to
     @sun i_robot_can_you.png
     
     I think LLMs can do nearly everything a human can but they do so in a very human-like manner, so they fall victim to many of the same errors.
   • Embed this notice
     Sick Sun (sun@shitposter.world)'s status on Thursday, 21-Nov-2024 21:51:02 JST Sick Sun

     in reply to

     • on-lain ✔ᵛᵉʳᶦᶠᶦᵉᵈ
     @lain so yeah I think its still useful information because it can warn you when a field of thought just plain has more places for manipulators to hide and justify things. maybe like race research is necessarily filled with complexity but you need to be extra vigilant about not just believing what you're told. which is the opposite of what people in these fields tell you, curious
     on-lain ✔ᵛᵉʳᶦᶠᶦᵉᵈ likes this.