Conversation

Notices

1. Embed this notice
   muesli (fribbledom@mastodon.social)'s status on Monday, 18-Aug-2025 23:19:33 JST muesli

   I don't hate LLMs. In fact I'm absolutely blown away by their performance. They work far better than they "should" and I ever thought possible.

   But that astonishing capability doesn't make them intelligent. What's frustrating is the massive, uninformed hype where people speak on LLMs and "AI" without *any* real understanding of what it is and isn't.

   It's crypto bros all over again.

   • Embed this notice
     Miredly (miredly@raru.re)'s status on Monday, 18-Aug-2025 23:19:32 JST Miredly

     in reply to

     @fribbledom The biggest problem with LLMs is they found their way into the hands of MBA types well before they should ever have been let out of a lab environment. They're really neat theoretically, but the Sam Altman's of the world should never have been allowed to convince people who don't know any better that you can think of an LLM like a person.

     Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     Dave Spector (dhmspector@mastodon.social)'s status on Monday, 18-Aug-2025 23:20:07 JST Dave Spector

     in reply to

     @fribbledom

     “What I had not realized is that extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people.”
     — ELIZA designer Joseph Weizenbaum

     Haelwenn /элвэн/ :triskell: likes this.