• BitSound@lemmy.world
    link
    fedilink
    arrow-up
    166
    ·
    1 year ago

    This seems really short-sighted. Why would I go to How Stuff Works when I can just ask the LLM myself?

    Maybe there’s just no possible business model for them anymore with the advent of LLMs, but at least if they focused on the “actually written by humans!” angle there’d be some hook to draw people in.

    • chaogomu@kbin.social
      link
      fedilink
      arrow-up
      129
      arrow-down
      3
      ·
      1 year ago

      The thing is, the LLM doesn’t actually know anything, and lies about it.

      So you go to How Stuff Works now, and you get bullshit lies instead of real information, you’ll also get nonsense that looks like language at first glance, but is gibberish pretending to be an article. Because sometimes the language model changes topics midway through and doesn’t correct, because it can’t correct. It doesn’t actually know what it’s saying.

      See, these language models are pre-trained, that the P in chatGPT. They just regurgitate the training data, but put together in ways that sort of look like more of the same training data.

      There are some hard coded filters and responses, but other than that, nope, just a spew of garbage out from the random garbage in.

      And yet, all sorts of people think this shit is ready to take over writing duties for everyone, saving money and winning court cases.

      • sugar_in_your_tea@sh.itjust.works
        link
        fedilink
        arrow-up
        27
        ·
        1 year ago

        Yeah, this is why I can’t really take anyone seriously when they say it’ll take over the world. It’s certainly cool, but it’s always going to be limited in usefulness.

        Some areas I can see it being really useful are:

        • generating believable text - scams, placeholder text, and general structure
        • distilling existing information - especially if it can actually cite sources, but even then I’d take it with a grain of salt
        • trolling people/deep fakes

        That’s about it.

        • Ech@lemm.ee
          link
          fedilink
          arrow-up
          8
          ·
          edit-2
          1 year ago

          generating believable text - scams, placeholder text, and general structure

          LLM generated scams are going to such problem. Quality isn’t even a problem there as they specifically go for people with poor awareness of these scams, and having a bot that responds with reasonable dialogue will make it that much easier for people to buy into it.

        • tryptaminev 🇵🇸 🇺🇦 🇪🇺@feddit.de
          link
          fedilink
          arrow-up
          5
          arrow-down
          2
          ·
          1 year ago

          AI tools can be very powerful, but they usually need to be tailored to a specific use case by competent people.

          With LLMs it seems to be the opposite, where people not competent for ML are applying it for the broadest of use cases. Just that it looks so good they are easily fooled and lack the understanding to realize the limits.

          But there is a very important Usecase too:

          Writing stuff that is only read and evaluated by similiar AI tools. It makes sense to write cover letters with ChatGPT because they are demanded but never read by a human on the other side of the job application. Since the weights and stuff behind it serm to be similiar, writing it with ChatGPT helps to pass the automatic analysis.

          Rationally that is complete nonsense, but you basically need an AI tool to jump through the hoops made by an AI tool applied by stupid people who need to make themselves look smart.

        • And009@reddthat.com
          link
          fedilink
          arrow-up
          1
          ·
          1 year ago

          It could be AI sport when we actually have an general purpose AI. That based on people working on llm and gpt, would take between 6 years and never happening.

          It’s not easy to create a super ai who’s realistically smarter than humans in every aspect.

      • Nonameuser678@aussie.zone
        link
        fedilink
        arrow-up
        12
        arrow-down
        1
        ·
        1 year ago

        I’ve graded papers from students who obviously used chatGPT to write them. They were a pass at best. Zero critical synthesis of ideas and application of them to the topic. I’m sure chatGPT has its uses but people really overhype its writing ability. There’s more to writing than putting words in the right places.

      • hoshikarakitaridia@sh.itjust.works
        link
        fedilink
        arrow-up
        10
        arrow-down
        1
        ·
        edit-2
        1 year ago

        I mean I would say maybe “regurgitating their training data” is putting it a bit too simple. But it’s true, we’re currently at the point where the AI can mimic real text. But that’s it - no one tells it not to lie rn, the programmatic goal of the AI is to get indistinguishable from real text with no bearing on the truthfulness of the information whatsoever.

        Basically we train our AIs to pretend to know, not to know. And sometimes it’s good at pretending, sometimes it isn’t.

        The “right” way to handle what the CEOs are doing would be to let go of a chunk of the staff, then let the rest write their articles with the help of chatgpt. But most CEOs are a bit too gullible when it comes to the abilities of AI.

      • Blackmist
        link
        fedilink
        arrow-up
        3
        ·
        1 year ago

        Literally predictive text but for whole articles.

        It doesn’t know the limits of it’s knowledge or indeed know anything. It just “knows” what an answer smells like. It even “knows” what excuses are supposed to look like when you call it out.

      • gapbetweenus@feddit.de
        link
        fedilink
        arrow-up
        6
        arrow-down
        6
        ·
        1 year ago

        The thing is, the LLM doesn’t actually know anything, and lies about it.

        Just like your average human journalist. If you ever read an article from not specialist journal on a topic you are familiar with - you know. This seems actually where LLM are very similar to how human brain works - if we don’t know something, we come up with some bullshit.

        • Ech@lemm.ee
          link
          fedilink
          arrow-up
          8
          ·
          1 year ago

          Even medium human writers can comprehend their work as a whole, though. There is a cohesiveness even to the bullshit. The LLM is just putting words down that match the prompt. It’s rng driven, readable Lorum Ipsum.

          If the results were still edited afterwards, there may be some merit to the output, but any company going full LLM isn’t looking for quality. They want to use it to churn out endless content that they simply can’t get from even a team of humans. More than could be edited even if they kept editors on staff.

          • gapbetweenus@feddit.de
            link
            fedilink
            arrow-up
            3
            arrow-down
            7
            ·
            1 year ago

            Even medium human writers can comprehend their work as a whole, though

            Sure, but a lot of humans are rather bad writers.

            but any company going full LLM isn’t looking for quality.

            That is true for 24h news cycle of online media, regardless LLM.

            • Ech@lemm.ee
              link
              fedilink
              arrow-up
              4
              ·
              1 year ago

              Sure, but a lot of humans are rather bad writers.

              Bad writing is still a step above rng junk, imo.

              but any company going full LLM isn’t looking for quality.

              That is true for 24h news cycle of online media, regardless LLM.

              Yes, that was my point. Setting up your company to put out more content than can possibly be processed by humans is a glaring sign of their values - ie quantity far above quality.

              • gapbetweenus@feddit.de
                link
                fedilink
                arrow-up
                2
                arrow-down
                2
                ·
                1 year ago

                Bad writing is still a step above rng junk, imo.

                I’v read writing worse than GTP. I had to help someone write an essay - and I just wrote it for him in the end, because he absolutely lacked the skills to write a long meaningful text. At at the same time - genius of a percussionist.

                • Ech@lemm.ee
                  link
                  fedilink
                  arrow-up
                  3
                  ·
                  1 year ago

                  Do you think that person was signing up for jobs writing for blogs or content farms?

        • tryptaminev 🇵🇸 🇺🇦 🇪🇺@feddit.de
          link
          fedilink
          arrow-up
          2
          arrow-down
          4
          ·
          1 year ago

          So modern journalists were redundant all along?

          But yeah, the quality of what is passing as journalism now is often ridiculous. But the only way to combat this is by having editors that are knowledgable about topics. But it seemed editors were the first people laid off, when internet articles became a thing.

          • gapbetweenus@feddit.de
            link
            fedilink
            arrow-up
            7
            arrow-down
            1
            ·
            1 year ago

            So modern journalists were redundant all along?

            24 hours news cycle of online media creates junk journalism on new level. Good journalism needs time and can’t spit out news articles every minute of the day. Editors won’t help, because it’s just not possible to do good journalism on that scale. But jeh - in general with AI, the jobs will shift more to editing. Which will be extremely soul-draining, going though tons of AI generated bullshit

    • AlmightySnoo 🐢🇮🇱🇺🇦@lemmy.world
      link
      fedilink
      arrow-up
      25
      ·
      edit-2
      1 year ago

      It’s a combination of three things:

      1- most people still google things;

      2- the more content you have the more organic traffic you’re likely to attract from Google;

      3- displaying ads on your website makes you money.

      Websites full of LLM generated content are just the natural continuation of MFAs (Made For AdSense) and there were lots of tools on sale back then in the 2006~2008 period that promised to automatically create websites for you and fill them with randomized content that is optimized for AdSense.

    • BanjoShepard@lemmy.world
      link
      fedilink
      arrow-up
      17
      arrow-down
      1
      ·
      1 year ago

      This reminds me of the short story “The Great Automatic Grammatizator” by Roald Dahl. In the story a machine is invented that can write great stories, but it’s creators go around buying the naming rights of authors so people will actually not their books.

    • LoafyLemon@kbin.social
      link
      fedilink
      arrow-up
      12
      arrow-down
      4
      ·
      1 year ago

      LLM cannot create new concepts, it can only create a mishmash of things it has been fed on.

    • mrbubblesort@kbin.social
      link
      fedilink
      arrow-up
      8
      ·
      1 year ago

      Correct me if I’m wrong, but isn’t AI generated content not copyrightable? Therefore nothing is stopping someone from taking all their content, rebranding it as “how stuff really works” or something, and then start stealing their business & ad revenue.

  • circuitfarmer@lemmy.sdf.org
    link
    fedilink
    arrow-up
    130
    arrow-down
    1
    ·
    1 year ago

    This is going to happen for a while. Execs who actually have no clue have now been sold on the idea that AI lets them keep making money without paying labor.

    It will fail eventually when the execs eventually take the time to learn what AI is capable of and what it isn’t capable of.

    Who am I kidding? It’ll continue indefinitely because there are few consequences for clueless executives.

  • KiloPapa@lemmy.world
    link
    fedilink
    arrow-up
    56
    arrow-down
    2
    ·
    1 year ago

    Considering most articles on the internet that don’t come from legitimate newspapers sound like they’re written by a 6-year-old who gets paid by the word, how much worse could it get?

  • waterplants@lemm.ee
    link
    fedilink
    arrow-up
    49
    arrow-down
    3
    ·
    edit-2
    1 year ago

    People really don’t understand the current state of LLM, like the pictures generated “Its a really good picture of what a dog would look like, it’s not actually a dog”. Like a police sketch, with a touch of “randomeness” so you don’t always get the same picture.

    I’m guessing they will try to solve this issue with some cheap human labour to review what is being generated. These verifers will probably not be experts on all the subjects that the llm will be spitting out, more of a “That does kind of look like a dog, APPROVED”.

    Let’s say I’m wrong, and LLM’s can make as good of an article as any human. The content would be so saturated (even a tumblr user could now make as good and as much content as one of these companies), I would expect companies to be joining in on all the strikes 😆.

    Funny world we are all going into.

    Boas Entradas

    • Thorny_Thicket@sopuli.xyz
      link
      fedilink
      arrow-up
      9
      arrow-down
      2
      ·
      1 year ago

      One thing I disagree with is the assumption that anyone could create the same article by themselves. Coming up with a good prompt is a skill in itself and not everyone is equally good at it. I actually believe a prompt writer is going to be a new profession in the near future.

    • VoxAdActa@beehaw.org
      link
      fedilink
      arrow-up
      3
      ·
      1 year ago

      I’m guessing they will try to solve this issue with some cheap human labour to review what is being generated.

      They already do. These current "AI"s are starting to look more and more like Mechanical Turks, except with a couple hundred third-world wage-slaves inside the box.

    • heimchen@discuss.tchncs.de
      link
      fedilink
      arrow-up
      3
      arrow-down
      1
      ·
      1 year ago

      Yea I think currently LLMs are in a stage to 2x or 1.5x the speed of a writer, but not really replace them.

  • DidacticDumbass@lemmy.one
    link
    fedilink
    arrow-up
    42
    ·
    1 year ago

    Bizarre. Not even keep a few editors for… the editing??

    I wonder how this will affect the Stuff You Should Know podcsst.

  • 𝒍𝒆𝒎𝒂𝒏𝒏@lemmy.one
    link
    fedilink
    English
    arrow-up
    35
    arrow-down
    1
    ·
    1 year ago

    Someone should create a blocklist for all these new AI-driven websites.

    For me personally thee primary appeal of websites are that there’s human authors behind the content… otherwise I’d just ask an ‘AI’ myself.

    • GiantBasil@beehaw.org
      link
      fedilink
      English
      arrow-up
      12
      ·
      1 year ago

      It would be great to have a list of sites so id know whose links I can just immediately ignore.

  • Ryan@lemmy.world
    link
    fedilink
    arrow-up
    32
    ·
    1 year ago

    I’ve read articles that were clearly created using ChatGPT, there was no extrapolation to add context/details to illustrate their points, and parts of it read like it just pulled from a Wikipedia page. The tone felt more robotic than pieces they published 6~8 months ago.

    ChatGPT can be useful when it’s part of a larger writing process, but I have a feeling that sites that create prompts and paste the output as their articles will slowly die-off because the quality isn’t there.

    • pingveno@lemmy.ml
      link
      fedilink
      arrow-up
      7
      ·
      1 year ago

      We’re probing the limits of generative AI right now. I expect a snapback of sorts as people find what does and does not work.

    • Ser Salty@feddit.de
      link
      fedilink
      arrow-up
      6
      ·
      1 year ago

      I was checking something on a Fandom “wiki” the other day and I swear to god the summary for a bunch of episodes for several shows was either written or rewritten by AI. You can tell because it uses a bunch of nonsense synonyms, like replacing the name Ray with Beam.

  • Tygr@lemmy.world
    link
    fedilink
    arrow-up
    25
    arrow-down
    1
    ·
    1 year ago

    How about instead of all the tracking cookie popups for permission, we force these sites to display a message that the content is AI generated.

  • Hagels_Bagels@lemmygrad.ml
    link
    fedilink
    arrow-up
    19
    ·
    1 year ago

    Great. Now people are going to read up a bunch of bs generated by a language model and confidently spread around “hallucinations” as facts.

        • itchy_lizard@feddit.it
          link
          fedilink
          arrow-up
          1
          ·
          1 year ago

          No, that’s exactly how this stuff works. Lay off 80% of writers and keep all your fact checkers and editors.

      • salient_one@lemmy.villa-straylight.social
        link
        fedilink
        arrow-up
        4
        ·
        edit-2
        1 year ago

        Probably, though it might be too optimistic to assume that. However, I believe it will still result in more mistakes simply because it’s harder to spot errors in an existing text than to not put errors in the text in the first place by fact-checking beforehand and then having another person proof-read.

        One of the reasons for that is that LLMs don’t feel guilty when they hallucinate while most humans don’t like to lie or be too lazy to fact check, and even if they don’t care about that, they still have to think about getting caught and damaging their reputation, which again LLMs don’t have. And you can’t call stating something false as a fact in an article an honest mistake (it’s negligence at best) unlike an editor’s missing something (due to a looming deadline, perhaps), especially when it’s assumed there won’t be too many hallucinations, which isn’t a certainty.

  • worfamerryman@beehaw.org
    link
    fedilink
    arrow-up
    14
    ·
    1 year ago

    How long until we can a browser extension that lets us know when we are on a site written by AI?

    I don’t mean AI detection, but instead, sites that announce they are laying off editors in favor of AI.

    • FaceDeer@kbin.social
      link
      fedilink
      arrow-up
      2
      ·
      1 year ago

      If there was such a thing then sites wouldn’t announce they’re laying off editors in favor of AI.

  • vrighter@discuss.tchncs.de
    link
    fedilink
    arrow-up
    13
    ·
    1 year ago

    Used to be one of my favourite sites when I was younger. Haven’t visited that site in ages. Holy crap, has it gone to complete shit. Like way worse than I thought possible

  • WasPentalive@lemmy.one
    link
    fedilink
    arrow-up
    11
    arrow-down
    2
    ·
    1 year ago

    ChatGPT comments on the news:

    Prompt : In recent news, the management of “How it’s Made” has fired all the writers and editors on the show and intends to replace them with a large language model. What do you think?

    ChatGPT

    As an AI language model, I don’t have personal opinions or feelings, but I can provide an analysis based on the information available up to my last update in September 2021.

    If the management of “How It’s Made” decides to fire all the writers and editors on the show and replace them with a large language model like me, there are both potential benefits and drawbacks to consider:

    Benefits:

    Cost Savings: Employing an AI language model could potentially reduce production costs as it eliminates the need for paying human writers and editors.
    Speed and Efficiency: AI models can generate content at a rapid pace, which may allow for more frequent episode releases or updates.
    Access to Data: AI models have access to vast amounts of data, which could result in more comprehensive and detailed explanations of the manufacturing processes.
    Consistency: AI-generated content can maintain a consistent style and tone throughout the episodes.
    

    Drawbacks:

    Lack of Creativity: While AI models can provide factual information, they might struggle to inject creative elements or engage with the audience on a more emotional level.
    Human Touch: Human writers and editors bring their unique perspectives and storytelling abilities to the show, which might be missed with an AI-only approach.
    Reliability: AI-generated content can sometimes produce errors or inaccurate information, which would require careful human oversight and fact-checking.
    Audience Response: The audience's perception of AI-generated content might not be as positive as content created by human professionals, leading to a potential decrease in viewership.
    

    Ultimately, the success of this decision would depend on how well the AI language model can replicate the engaging and informative aspects that made “How It’s Made” popular in the first place. It’s important for the management to carefully evaluate the pros and cons and ensure a seamless transition if they decide to go down this route. Additionally, audience feedback and receptiveness to the AI-generated content will play a significant role in determining the long-term viability of such a change.

      • nickajeglin@lemmy.one
        link
        fedilink
        arrow-up
        5
        arrow-down
        1
        ·
        1 year ago

        No shade on “how it’s made”, it’s one of my favorite shows. But I think a LLM could probably write most of the narration. They primarily describe what is happening on screen. You might have to train one special to have information on industrial and manufacturing processes.

        • bloodfart@lemmy.ml
          link
          fedilink
          arrow-up
          7
          ·
          1 year ago

          it absolutely could not.

          one of the hallmarks of how its made’s narration is the terse, accurate representation of whats going on. llms are often wrong about things and dont do great at precise descriptions to begin with.