• @UnderpantsWeevil@lemmy.world
    link
    fedilink
    338 months ago

    Displacing millions of workers

    I have seen what AI outputs at an industrial scale, and I invite you to replace me with it while I sit back and laugh.

    • @Eccitaze@yiffit.net
      link
      fedilink
      17
      edit-2
      8 months ago

      my company announced today that they were going to start a phased rollout where AI would provide first responses to tickets, with it initially being “reviewed” by humans with the eventual goal being it just sending responses unsupervised. The strength of my "OH HELL NO" derailed the entire meeting for a solid 15 minutes lmao

    • @WarlordSdocy@lemmy.world
      link
      fedilink
      168 months ago

      It’s not about entirely replacing people. It’s about reducing the number of people you hire in a specific role because each of those people can do more using AI. Which would still displace millions of people as companies get rid of the lowest performing of their workers to make their bottom line better.

      • @mPony@lemmy.world
        link
        fedilink
        48 months ago

        It’s not about entirely replacing people

        Tell yourself that all you wish. Then maybe go see this thread about Spotify laying off 1500 people and having a bit of a rough go with it. If they could they would try to replace every salaried/contracted human with AI.

        • @WarlordSdocy@lemmy.world
          link
          fedilink
          88 months ago

          Yeah I’m not arguing that replacing people isnt what they want to do. They absolutely could if they would. I was just responding to the person saying they can’t be replaced cause AI can’t do what they do perfectly yet. My point was that at least for now it’s not entirely replacing people but still displacing lots of people as AI is making people able to do more work.

          • @UnderpantsWeevil@lemmy.world
            link
            fedilink
            38 months ago

            AI is making people able to do more work.

            It’s not. AI is creating more work, more noise in the system, and more costs for people who can’t afford to mitigate the spam it generates.

            The real value add in AI is the same as shrinkflation. You dump your clients into paying more for less, by insisting work is getting done that isn’t.

            This holds up so long as the clients never get wise to the con. But as the quality of output declines, it impacts delivery of service.

            Spotify is already struggling to deliver services to it’s existing user base. It’s losing advertisers. And now it will have fewer people to keep the ship afloat.

            • @WarlordSdocy@lemmy.world
              link
              fedilink
              78 months ago

              I think that’s a really broad statement. Sure there are some industries where AI hurts more then it helps in terms of quality. And of course examples of companies trying to push it too far and getting burned for it like Spotify. But in many others a competent person using AI (someone who could do all the work without AI assistance, just slower) will be much more efficient and get things done much faster as they can outsource certain parts of their job to AI. That increased efficiency is then used to cut jobs and create more profit for the companies.

              • @UnderpantsWeevil@lemmy.world
                link
                fedilink
                18 months ago

                Sure there are some industries where AI hurts more then it helps in terms of quality.

                In all seriousness, where has the LLM tech improved business workflow? Because if you know something I don’t, I’d be curious to hear it.

                But in many others a competent person using AI (someone who could do all the work without AI assistance, just slower) will be much more efficient and get things done much faster as they can outsource certain parts of their job to AI.

                What I have seen modern AI deliver in practice is functionally no different than what a good Google query would have yielded five years ago. A great deal of the value-add of AI in my own life has come as a stand-in for the deterioration of internet search and archive services. And because its black-boxed behind a chat interface, I can’t even tell if the information is reliable. Not in the way I could when I was routed to a StackExchange page with a multi-post conversation about coding techniques or special circumstances or additional references.

                AI look-up is mystifying the more traditional linked-post explanations I’ve relied on to verify what I was reading. There’s no citation, no alternative opinion, and often no clarity as to if the response even cleanly matches the query. Google’s Bard, for instance, keeps wanting to shoehorn MySQL and Postgres answers into questions about MSSQL coding patterns. OpenAI routinely hallucinates answers based on data gleaned from responses to different versions of a given coding suite.

                Rather than giving a smoother front end to a well-organized Wikipedia-like back end of historical information, what we’ve received is a machine that sounds definitive regardless of the quality of the answer.