Plugging into chatgpt how to play a certain game for example spits out clear and concise information I’d spend the better part of half an hour scraping reddit for and subjecting myself to psychic damage. And it makes me feel bad in turn that I’m part of the problem polluting the earth with this crap. Honestly feels like the internet has been enshitified on purpose just to make people have to rely on ai to navigate it without throwing your hands up in frustration.

  • GnastyGnuts [he/him]@hexbear.net
    link
    fedilink
    English
    arrow-up
    9
    ·
    7 hours ago

    I used AI one whole time when I couldn’t figure out where a setting was in a program, and it just got it wrong and I spent another half-hour (it probably wasn’t actually that long) trying to find it myself, but even more angry because I had interacted with AI.

    I still cannot find the use-value for this shit. The closest I’ve gotten is making use of AI captioning, which is still flawed but a better alternative than no captions at all. That said, it wouldn’t be as useful if somebody had been born deaf or something like that, because they wouldn’t know what similar-sounding words or phrases the AI got wrong and be able to fix it with context clues.

    • Horse {they/them}@lemmygrad.ml
      link
      fedilink
      English
      arrow-up
      4
      ·
      7 hours ago

      I still cannot find the use-value for this shit

      in my (limited) experience with text-outputting LLMs the things they are good for are

      • changing the tone of a short to medium length piece of text (i used it to re-word an email to sound “professional”, something i am not good at)
      • giving a base chunk of text to manually edit (i am bad at writing about myself so got the robot to do it then fucked with it a lot)
      • giving a list of randomly generated names that fit quite specific criteria (i use it sometimes to name things in ttrpgs if my imagination fails me)

      i have never asked it for information myself and am quite surprised that it actually worked for SFS

  • Horse {they/them}@lemmygrad.ml
    link
    fedilink
    English
    arrow-up
    18
    ·
    9 hours ago

    And it makes me feel bad in turn that I’m part of the problem polluting the earth with this crap

    i wouldn’t beat yourself up about it, a 3-4 hour gaming session uses far more power than a couple of prompts asking the robot how to play the game

  • frogbellyratbone_ [e/em/eir, any]@hexbear.net
    link
    fedilink
    English
    arrow-up
    10
    ·
    edit-2
    9 hours ago

    honestly, google is shit. even google’s LLM is garbage. they missed the LLM rush and are playing catch-up. their product is horrible

    i switched to duckduckgo about a year or so ago and so happy i did.

    but yah, i use LLMs a lot of the time. other than video game tips / resources, they’re generally the new search engines for better for worse. i try not to use chatgpt and will use deepseek or Mistral AI instead and lie to myself that i’m a better person for doing so.

  • FourteenEyes [he/him]@hexbear.net
    link
    fedilink
    English
    arrow-up
    29
    ·
    11 hours ago

    Literally every time I’ve tried to rely on AI results they’ve been wrong. Either small details or something huge but it’s always something.

      • MolotovHalfEmpty [he/him]@hexbear.net
        link
        fedilink
        English
        arrow-up
        12
        ·
        edit-2
        10 hours ago

        They already are. The links that appear as information references in Google’s ‘AI’ preview for example are already selected via weighted lists that benefit Google or their biggest advertisers. So not only are they basically rendering SEO rankings irrelevant by producing info / answers with no click through at all, but they’re prioritising their own interests in that too. People in digital marketing are shitting a brick about it at the moment.

  • KnilAdlez [none/use name]@hexbear.net
    link
    fedilink
    English
    arrow-up
    8
    ·
    edit-2
    10 hours ago

    I admit I have started using chatgpt instead of google scholar. Google has become absolute garbage. I recently switched to a self-hosted meta search engine called SearXNG, and it has been better.

  • DerEwigeAtheist [she/her, comrade/them]@hexbear.net
    link
    fedilink
    English
    arrow-up
    7
    ·
    edit-2
    10 hours ago

    The last time I looked something up, the AI gave me an answer, and then, the first link below gave the exact same answer, but with the terms reversed. I’ve never had AI not being wrong about some detail. It’s too unreliable for me. I don’t see the point if I’ve gotta do additional research to doublecheck anyway. Doesn’t safe me any time

  • IHave69XiBucks@lemmygrad.ml
    link
    fedilink
    English
    arrow-up
    6
    ·
    11 hours ago

    LLMs are useful for very specific things. The issue with them is that tech companies want them to do everything and push them in everything.

    Instead of focusing on the areas where they actually are useful, and refining them to be more power efficient and less prone to hallucinations tech companies were like lets just see if we can redirect all human energy to the glorified word calculator and see if it becomes sentient. (Hint: It wont.)

        • MizuTama [he/him, any]@hexbear.net
          link
          fedilink
          English
          arrow-up
          2
          ·
          6 hours ago

          They already are. The links that appear as information references in Google’s ‘AI’ preview for example are already selected via weighted lists that benefit Google or their biggest advertisers. So not only are they basically rendering SEO rankings irrelevant by producing info / answers with no click through at all, but they’re prioritising their own interests in that too. People in digital marketing are shitting a brick about it at the moment.

          Ironically, it is the more annoying one to work into a searx instance once you set one up.