• btaf45@lemmy.worldOP
    link
    fedilink
    English
    arrow-up
    9
    ·
    1 month ago

    That’s what I think too. AI is mainly useful for things that don’t have right or wrong answers.

    Although this incorrect answers is obvious, what about all the times where an incorrect answer from AI is not obvious?

    • contrefeu@akko.contref.eu
      link
      fedilink
      arrow-up
      1
      ·
      1 month ago

      @Gsus4 @btaf45 That’s true for AI that has been trained for the general public to provide an answer for any provided question meaning they are forced to respond to a prompt even though they are wrong and maybe even know they are wrong. They just don’t know the answer and can’t say that because it’s commercially bad.

      I do believe that for scientific research AI models are much more precise because they have been trained with the right datasets and are tasked with answering specific questions.