• milicent_bystandr@lemm.ee
    link
    fedilink
    arrow-up
    5
    arrow-down
    11
    ·
    6 months ago

    The other massive flaw it demonstrates in AI today is it’s popular to dunk on it so people make up lies like this meme and the internet laps them up.

    Not saying AI search isn’t rubbish, but I understand this one is faked, and the tweeter who shared it issued an apology. And perhaps the glue one too.

    • dependencyinjection@discuss.tchncs.de
      link
      fedilink
      arrow-up
      13
      arrow-down
      1
      ·
      6 months ago

      There are cases of AI using NotTheOnion as a source for its answer.

      It doesn’t understand context. That’s not to say I am saying it’s completely useless, hell I’m a software developer and our company uses CoPilot in Visual Studio Professional and it’s amazing.

      People can criticise the flaws in it, without people doing it because it’s popular to dunk on it. Don’t shill for AI and actually take a critical approach to its pros and cons.

      • milicent_bystandr@lemm.ee
        link
        fedilink
        arrow-up
        6
        arrow-down
        1
        ·
        6 months ago

        I think people do love to dunk on it. It’s the fashion, and it’s normal human behaviour to take something popular - especially popular with people you don’t like (e.g. j this case tech companies) - and call it stupid. Makes you feel superior and better.

        There are definitely documented cases of LLM stupidity: I enjoyed one linked from a comment, where Meta’s(?) LLM trained specifically off academic papers was happy to report on the largest nuclear reactor made of cheese.

        But any ‘news’ dumping on AI is popular at the moment, and fake criticism not only makes it harder to see a true picture of how good/bad the technology is doing now, but also muddies the water for people believing criticism later - maybe even helping the shills.