Conversation

Notices

1. Embed this notice
   pettter (pettter@mastodon.acc.umu.se)'s status on Friday, 07-Apr-2023 02:47:31 JST pettter

   in reply to

   • Simon Willison

   @simon that would require you to have a fairly great NLP system to detect asserted facts and so on because LLMs aren't more or less "sure" about any specific long sections of text, generally. They sample from a distribution of probable continuations of some specific input text.

   • Embed this notice
     Simon Willison (simon@fedi.simonwillison.net)'s status on Friday, 07-Apr-2023 02:47:32 JST Simon Willison

     in reply to

     What does feel realistic is training these models to be MUCH better at providing useful indications as to their confidence levels

     The impact of these problems could be greatly reduced if we could counteract the incredibly convincing way that these confabulations are presented somehow

     I also think there's a lot of room for improvement here in terms of the way the UI is presented, independent of the models themselves

   • Embed this notice
     Simon Willison (simon@fedi.simonwillison.net)'s status on Friday, 07-Apr-2023 02:47:33 JST Simon Willison

     I agree that confabulation/hallucination/lying is a huge problem with LLMs like ChatGPT, Bard etc

     But I think a lot of people are underestimating how difficult it is to establish "truth" around most topics

     High quality news publications have journalists, editors and fact checkers with robust editorial processes... and errors still frequently slip through

     Expecting a LLM to perfectly automate that fact checking process just doesn't seem realistic to me