• @CanadaPlus@lemmy.sdf.org
    link
    fedilink
    English
    21 month ago

    In China where there’s a ton of manufacturing, I’d agree, but in the West there’s a lot of jobs that are surface-level natural language tasks.

      • @CanadaPlus@lemmy.sdf.org
        link
        fedilink
        English
        21 month ago

        I have some idea, yeah. Call centers employ a lot of people, as do book-keeping, HR and retail checkouts. It’s not going to code or engineer any time soon, taking a statistically decent guess at what you do, but the percentage of the non-Lemmy population that does that sort of work is tiny.

        Manufacturing depends heavily on the specific job. Obviously machining is easily automated (if not the loading and maintenance of the CNC machines themselves), and basic assembly can be too, but once non-rigid or variable materials come into the picture it all gets harder, and any kind of uncontrolled environment seems to make it impossible.

          • @CanadaPlus@lemmy.sdf.org
            link
            fedilink
            English
            21 month ago

            Compartmentalising sensitive data isn’t too hard with AI. LLMs don’t have a memory of their own once out of training, remember. It’s just a matter of setting it up the right way.

            The issue with checkouts has been theft, since they basically just trust the user to charge themselves right now. Amazon’s Just Walk Out is the technology to watch for that kind of checkout, and for anything where shoppers don’t collect items themselves LLMs can do a decent job without finetuning.

            It’s not going to replace every job, not with current capabilities anyway, but enough to drive a big economic shift? Yeah, I do agree with the IMF on that.