• DempstersBox@lemmy.world
    link
    fedilink
    arrow-up
    0
    ·
    2 months ago

    Huh.

    So in my ‘frequently visited’ list, very recently a link to chatGPT has populated.

    I’ve poked at a few of the generative ai things, but only the actual chatgpt once, months ago, and have wondered if there’s enough bullshit from reddit to give it enough dings to make the list.

    Apparently so

    • douglasg14b@lemmy.world
      link
      fedilink
      arrow-up
      0
      ·
      edit-2
      2 months ago

      Yep, and even worse. Lemmy has absolutely NO controls for quality and minimal moderation tools or capabilities. It’s in a much worse position than Reddit.

      If it’s not already happening (And I think it is), it will.

      • Ademir@lemmy.eco.br
        link
        fedilink
        arrow-up
        0
        ·
        2 months ago

        What tools are missing to moderate a LLM bot?

        I am a mod, I receive a report from an user, check it and ban. All done with current tools we have.

      • deranger@sh.itjust.works
        link
        fedilink
        arrow-up
        0
        ·
        2 months ago

        I can’t exactly put my finger on it, but every god damn thing ChatGPT spits out has the same cadence or structure. Something about how it lays out points and wraps it up, and its usage of commas, is so noticeable to me.

        • Boozilla@lemmy.world
          link
          fedilink
          English
          arrow-up
          0
          ·
          2 months ago

          Lots of compound sentences. And a “tone” suitable for USA Today. At least, that’s the vibe I get from ChatGPT.

          • deranger@sh.itjust.works
            link
            fedilink
            arrow-up
            0
            ·
            edit-2
            2 months ago

            I wouldn’t say it’s necessarily overuse of commas, it’s how it consistently uses them the same way in nearly every response. It’s something about the overall structure of what it spits out. I wish a linguist would pop in and break it down for me. It’s a combination of tone, tense, arrangement of the sentences, how the paragraphs are put together - all of it feels very samey to me, unless it’s prompted otherwise.

            There’s only a few “styles” of sentences it spits out and to me, it seems quite obvious. Humans don’t write that consistently all the time, they’re messier.

  • UnderpantsWeevil@lemmy.world
    link
    fedilink
    English
    arrow-up
    0
    ·
    2 months ago

    AI comments under an AI generated article being inflated by AI account activity. Who even needs end users anymore? Its a perfectly autonomous system.

    • Meltrax@lemmy.world
      link
      fedilink
      arrow-up
      0
      ·
      2 months ago

      Ironically there used to be a subreddit for this.

      It was intentionally meant to be view-only for humans, and the bots within it were named for and trained on other subreddits. So you have AdviceAnimalsBot, LinuxBot, GamingBot, AskRedditBot, GoneWildBot, etc. They would post on a rotation, emulating what users in their respective subreddits posted. They would all comment on each other’s posts, emulating their respective subreddit’s comments.

      As an experiment it was actually really cool and fun to read through. It was also very clear that these were bots and you could identify which was which, and nothing was pretending to be a human for karma (there were no votes in the subreddit).

        • Meltrax@lemmy.world
          link
          fedilink
          arrow-up
          0
          ·
          edit-2
          2 months ago

          Yeah that was the one! I had forgotten the name. It was actually a really cool use of bots and a fun microcosm of how they interact with each other. From the perspective of like, a college AI research project, it was really interesting.

          Kinda sucks that that is just the entire website now.

    • CaptDust@sh.itjust.works
      link
      fedilink
      arrow-up
      0
      ·
      2 months ago

      Who even needs end users anymore?

      Advertisers do, and if they ever wise up spez’s house of cards will crumble.

      • UnderpantsWeevil@lemmy.world
        link
        fedilink
        English
        arrow-up
        0
        ·
        2 months ago

        Advertisers don’t need end users. They just need happy bosses willing to cover their salaries.

        In that sense, the business marketing team and the Reddit “look at our bullshit numbers” team are on the same side of the field.

        • CaptDust@sh.itjust.works
          link
          fedilink
          arrow-up
          0
          ·
          2 months ago

          It’s easy to bullshit engagement, it’s tougher to bullshit click-through rates and sale conversion metrics. It will take time to identify patterns, but inevitably the data will begin to reflect the truth, that’s when advertisers will break and move their money to more successful (organic) platforms.

          At least this is what happens in a sane world, as for our reality who knows.

          • Draconic NEO@lemmy.world
            link
            fedilink
            arrow-up
            0
            ·
            edit-2
            2 months ago

            If they haven’t figured it out by now they probably never will. After all fake engagement isn’t a new thing, not by a long shot. They’ll probably just make some excuse about banner blindness for why click-through rates are down, possibly trying to also justify more aggressive ads, and more spending.

        • Ragnarok314159@sopuli.xyz
          link
          fedilink
          arrow-up
          0
          ·
          2 months ago

          Yep. As long as sales and marketing can point to some bullshit KPI metrics as having exceeded all their goals, they act like they are the ones who bring in the profit.

          Nevermind no one is buying anything and no traffic is going to the website, that is a different profit center’s problem and certainly not the fault of the MBA losers.

    • Draconic NEO@lemmy.world
      link
      fedilink
      arrow-up
      0
      ·
      2 months ago

      Something that nobody mentions when they talk about dead internet theory that I think should be talked about is that the worthless engagement trolls that are real people, really do not and should not count as real people. They’re never going to give you worthwhile responses, they don’t add anything to conversations. Unless they see an opportunity to pick a fight they’re probably not even listening to you.

      Therefore they’re not really any different than an AI run account that’s not going to acknowledge you. And in some ways they are much worse, despite being technically “real people”.

  • jasep@lemmy.world
    link
    fedilink
    arrow-up
    0
    ·
    2 months ago

    I understand this was posted in the Reddit community, but also…

    Who cares? Let that cesspool die.