Conversation

Notices

1. Embed this notice
   Eugen Rochko (gargron@mastodon.social)'s status on Wednesday, 07-Dec-2022 11:06:57 JST Eugen Rochko

   • かき@GNUsocialJP

   ChatGPT is at best impressive as a search engine where you can't tell how reliable the source is.

   • Embed this notice
     josh buermann (buermann@mastodon.social)'s status on Wednesday, 07-Dec-2022 13:27:21 JST josh buermann

     in reply to

     @Gargron

     An unreliable search engine is as good as it gets, I guess, if it's "a potential absolute limit on the whole practice of language modeling."

     https://www.zdnet.com/article/openais-gigantic-gpt-3-hints-at-the-limits-of-language-models-for-ai/

   • Embed this notice
     chrisjrob (chrisjrob@mastodon.social)'s status on Wednesday, 07-Dec-2022 16:11:11 JST chrisjrob

     in reply to

     @Gargron it's like using Google's old "I feel lucky" button, and then forcing yourself to use whatever answer it gives.

   • Embed this notice
     John Samuel (jsamwrites@mastodon.social)'s status on Wednesday, 07-Dec-2022 21:42:04 JST John Samuel

     in reply to

     @Gargron For some queries, I ask #ChatGPT for references to be completely sure of the results. I feel that this should be the default feature. But then large language models do not work that way.

   • Embed this notice
     WesDym (wesdym@mastodon.social)'s status on Thursday, 08-Dec-2022 00:43:34 JST WesDym

     in reply to

     @Gargron Reminds me of Cuil Theory.

     http://cuiltheory.wikidot.com/what-is-cuil-theory

   • Embed this notice
     sencommander@mastodon.social's status on Thursday, 15-Dec-2022 05:42:48 JST SenCommander

     in reply to

     @Gargron wonkypedia?