Conversation

Notices

1. Embed this notice
   Julia Angwin (julia@journa.host)'s status on Wednesday, 04-Sep-2024 07:42:17 JST Julia Angwin

   in reply to

   • Quinn Norton
   • Proof News

   @quinn @proofnews Thank you. I am, obviously aware of how AI works. But given that they market themselves as "intelligence" and have made promises about preventing election misinformation, we are doing this testing to hold them accountable.

   • Embed this notice
     Alexandre Oliva (lxo@gnusocial.net)'s status on Wednesday, 04-Sep-2024 07:42:17 JST Alexandre Oliva

     in reply to
     I wonder if they use their own bullsh*t-generating tools to write their hopeful promises and marketing materials. it often feels like they do.
   • Embed this notice
     Quinn Norton (quinn@social.circl.lu)'s status on Wednesday, 04-Sep-2024 07:42:18 JST Quinn Norton

     in reply to

     • Proof News

     @Julia @proofnews they're not really correct or incorrect. they're a next word selector, there's nothing that knows in any sense we use that word. accuracy is the wrong way to think about AI. it's hallucinating, all the time, just sometimes it aligns with the outside world. it doesn't even know when it does.

   • Embed this notice
     Julia Angwin (julia@journa.host)'s status on Wednesday, 04-Sep-2024 07:42:19 JST Julia Angwin

     • Proof News

     I continue to be disappointed by the poor accuracy rates of AI.

     Our latest @proofnews investigation shows the leading AI models were incorrect one-third of the time when confronted with known misinformation about the presidential candidates. 👇🏻

     https://www.proofnews.org/ai-models-struggle-to-get-the-facts-straight-about-kamala-harris-2/