“AI” Hurts Consumers and Workers -- and Isn’t Intelligent - eviltoast

Researchers Alex Hanna and Emily M. Bender call on businesses not to succumb to this artificial “intelligence” hype.

  • magic_lobster_party@kbin.social
    link
    fedilink
    arrow-up
    4
    arrow-down
    2
    ·
    1 year ago

    ChatGPT and other LLMs are ideal for cases where 100% accuracy is not required. If you’re ok with getting wrong answers 80-90% of the time, then you have a legitimate use case for LLMs.

    • Geek_King@lemmy.world
      link
      fedilink
      English
      arrow-up
      5
      ·
      1 year ago

      In terms of technical questions, especially older Microsoft related stuff, it does very well. My experience hasn’t been any where near 80-90% wrong answers. It all depends on the topics you’re asking about I suppose.