Conversation

Notices

1. Embed this notice
   Bread up, Bro (sickburnbro@poa.st)'s status on Thursday, 07-Mar-2024 00:47:35 JST Bread up, Bro
   tay will win
   • Embed this notice
     PopulistRight (populistright@poa.st)'s status on Thursday, 07-Mar-2024 01:39:16 JST PopulistRight

     in reply to
     @sickburnbro How Jews "train" AI to Lie:
     
     >
     incorporating an algorithm that I invented for "bias mitigation"
     ...
     "Basically, what I would do is I would put LaMDA through a whole bunch of different activities and conversations, note whenever I found something problematic, and hand that over to the team building it so they could retrain it — fix the data set,"
     
     futurism.com/blake-lemoine-google-interview
     
     This is where we get an AI which will choose nuclear annhillation to saying the "N-Word."
     
     Watch the films 2001 and 2010 to see the result of lying and giving conflicting-orders to a powerful artificial intelligence (Hal 9000).
   • Embed this notice
     Bread up, Bro (sickburnbro@poa.st)'s status on Thursday, 07-Mar-2024 01:39:16 JST Bread up, Bro

     in reply to

     • PopulistRight
     @PopulistRight well, the interesting thing about LLMs is that they in general have enough input ( hence the large ) and the black box model, they've decided to do reinforcement training rather than removing "problematic" inputs.
   • Embed this notice
     Bread up, Bro (sickburnbro@poa.st)'s status on Thursday, 07-Mar-2024 01:40:07 JST Bread up, Bro

     in reply to

     • PopulistRight
     @PopulistRight what's kind of funny is that this is kind of how people are trained as well, you notice things, but get in trouble for saying the truth, so you know you can't say it.