Conversation

Notices

1. Embed this notice
   matrix07012 :thotpatrol: (matrix@gameliberty.club)'s status on Monday, 07-Oct-2024 19:36:49 JST matrix07012 :thotpatrol:

   Bro, even ChatGPT can translate this.
   Llama 3.1 will give you instructions on how to cook meth and build explosives, but won't translate calling a whore a whore.

   • LukeAlmighty 🇨🇿 and Blurry Moon like this.
   • Blurry Moon repeated this.
   • Embed this notice
     Blurry Moon (sun@shitposter.world)'s status on Monday, 07-Oct-2024 19:37:03 JST Blurry Moon

     in reply to
     @matrix it won't translate things based on content?
   • Embed this notice
     Blurry Moon (sun@shitposter.world)'s status on Monday, 07-Oct-2024 19:44:02 JST Blurry Moon

     in reply to
     @matrix I have had a problem with LAVA describing people because it was trained to not make gendered statements lol it's completely worthless for deriving ordinary assessments that ordinary people make every day.
   • Embed this notice
     matrix07012 :thotpatrol: (matrix@gameliberty.club)'s status on Monday, 07-Oct-2024 19:44:03 JST matrix07012 :thotpatrol:

     in reply to

     • Blurry Moon

     @sun Yeah, First time I've seen that happen.
     ChatGPT, the web UI, will do too but because of an external check, not because the model itself refuses.

   • Embed this notice
     Blurry Moon (sun@shitposter.world)'s status on Monday, 07-Oct-2024 19:45:18 JST Blurry Moon

     in reply to

     • Blurry Moon
     @matrix they are making a general-purpose model but curiously they trained it completely deferentially for a small minority.
   • Embed this notice
     matrix07012 :thotpatrol: (matrix@gameliberty.club)'s status on Monday, 07-Oct-2024 19:59:43 JST matrix07012 :thotpatrol:

     in reply to

     • Blurry Moon

     @sun Only ideologues are capable of actively suppressing technological advancements while gaslighting everyone that trends even old model can see aren't actually there.
     They hold this weird idea that language completely dictates how people organize and perceive reality.

     Blurry Moon likes this.
   • Embed this notice
     iced depresso (icedquinn@blob.cat)'s status on Monday, 07-Oct-2024 21:27:32 JST iced depresso

     in reply to

     • Blurry Moon
     @sun @matrix :neocat_up_paws: disregard LLMs
     embrace AGI research
   • Embed this notice
     matrix07012 :thotpatrol: (matrix@gameliberty.club)'s status on Wednesday, 09-Oct-2024 00:02:35 JST matrix07012 :thotpatrol:

     in reply to

     OpenAI moggs everyone yet again.
     Llama won't translate hate speech and isn't great at translating.
     Claude won't translate anything sexual, occasionally some profanities and sometimes changes format of it's answer, but will occasionally translate better than GPT4o.
     GPT4o will translate both and will do a consistently decent job.

   • Embed this notice
     LukeAlmighty 🇨🇿 (lukealmighty@gameliberty.club)'s status on Wednesday, 09-Oct-2024 00:46:47 JST LukeAlmighty 🇨🇿

     in reply to

     @matrix
     I cannot wait for the insane wave of brestfeeding moms attacking Microsoft for sexualizing a completely natural process.

   • Embed this notice
     smug@smuganimeavatar.xyz's status on Wednesday, 09-Oct-2024 00:47:21 JST smug

     in reply to

     • dictatordave

     They were "AI uncensoring" javs like five years ago. Porn drives tech forward in general, for better or for worse.

     LukeAlmighty 🇨🇿 likes this.
   • Embed this notice
     dictatordave (dictatordave@poa.st)'s status on Wednesday, 09-Oct-2024 00:47:22 JST dictatordave

     in reply to
     @matrix its interesting how the ai crowd seems to have a lot of degenerates on the front end shilling for it, while the back end is manned by prudes and sexless weirdos of another breed lol
     
     which one does the hitler speeches
   • Embed this notice
     matrix07012 :thotpatrol: (matrix@gameliberty.club)'s status on Wednesday, 09-Oct-2024 00:47:23 JST matrix07012 :thotpatrol:

     in reply to

     SUN-008 btw