It just feels too good to be true.

I’m currently using it for formatting technical texts and it’s amazing. It doesn’t generate them properly. But if I give it the bulk of the info it makes it pretty af.

Also just talking and asking for advice in the most random kinds of issues. It gives seriously good advice. But it makes me worry about whether I’m volunteering my personal problems and innermost thoughts to a company that will misuse that.

Are these concerns valid?

  • Overzeetop@beehaw.org
    link
    fedilink
    arrow-up
    9
    ·
    1 year ago

    So, it’s okay to replace jobs which seem like menial bullshit to you, but not jobs you deem to be “creative.” We’re taking a bell curve of human ability and simply drawing the line of “obsolete human” in a different place and you’re disappointed that you’re way closer to it than you were a decade ago.

    NB: I sat in a room with 200 other engineers this summer and they all scoffed at the idea that a computer could take their place. But I’m absolutely certain that what we do could be - is being - automated even as we claim to be the intelligent ones who are not in fear of replacement. My job is just the learned sum of centuries of human knowledge which is honed year after year and has to be taught, wholecloth, to every new human in my profession. There are people who will say I’m the smartest guy in the room (for a small enough room ;-) but 90% what I do is just applying a set of rules based on inputs and boundary conditions. We feel like this shouldn’t happen to us because we’re smart. We think independently. We have special abilities which set us apart from ML generated outputs. We’re also full of shit. There are absolutely areas were ML/AI will not surpass our value in the system for quite some time, but more and more of our expertise will be accomplishable by application of distilled large data sets.

    • davehtaylor@beehaw.org
      link
      fedilink
      arrow-up
      16
      ·
      1 year ago

      So, it’s okay to replace jobs which seem like menial bullshit to you…"

      The promise of automation absolutely is about riding ourselves of shit, low-paid, dangerous, menial labor so that we’re free to pursue things that we’re passionate about. But right now, AI is doing precisely the opposite. Actual creative and skilled people are being pushed out and ending up with shit, low-paid jobs, gig work and other exploitative jobs just to make ends meet.

      "… but not jobs you deem to be “creative.”

      I can hear the sneer in this, so I think my assumption was correct at the end of my last comment.

      It’s absolutely pointless then to even bother with this, but I’m going power through anyway

      My job is just the learned sum of centuries of human knowledge which is honed year after year and has to be taught, wholecloth, to every new human in my profession.

      This is the same argument of “AI art is just doing what humans do, looking at other art and mixing it up”. And it’s just as backward and fallacious when applied to any other industry. AI can only give you a synthesis of exactly what you feed it. It can’t use its life experience, its upbringing, its passions, its cultural influences, etc to color its creativity and thinking, because it has none and it isn’t thinking. Two painters who study and become great artists, and then also both take time to study and replicate the works of Monet can come away from that experience with vastly different styles. They’re not just puking back a mashup of Monet’s collected works. They’re using their own life experience and passions to color their experience of Impressionism.

      That’s something an AI can never do, and it leaves the result hollow and meaningless.

      It’s no different if you apply that to software development. People in tech love to think that development is devoid of creativity and is just cold, calculating math. But it’s not. Even if you never touch UI or UX, the feature you develop isn’t isolated. It interacts with everything else in the system. Do something purely follow rules? Maybe. But not all. There is never a point where your code is devoid of any humanity. There are usually multiple ways to solve a problem, and many times they’re all just as equally valid. And often theres a problem that it takes a human to understand the scope of to understand how the solution needs to be architected.

      We need an environment that is actively and intensely hostile to AI tools and those that promote them. People calling themselves “prompt engineers” or people acting like they’re creative because they fed some bullshit into a blackbox need to be shamed and ostracized. This shit is dangerous and it’s doing real and measurable harm. The people who think that everything should only be about cold, quantifiable data, large enough data sets, and everything else ignored, are causing, and have caused, immense harm because they refuse to see the humanity in the consequences of their actions.

      The ones who really think they’re the smartest people in the room are the people developing and promoting these tools. And who are they? Wealthy, privileged, white men who have no concept of the real world, who’ve gorged themselves on STEM-only curricula, and have no understanding of history, civics, or humanities in which to conceptualize the context of the shit they’re unleashing into the world.

      • lloram239@feddit.de
        link
        fedilink
        arrow-up
        2
        ·
        1 year ago

        AI can only give you a synthesis of exactly what you feed it.

        So do humans. What you call “life experience” is just training data. Nothing forces you to train AI on all the stuff out there, you are free to train it on a specific subset of data. You are even free to plug a webcam into a robot and train it on whatever that sees in its lifetime.

        Whenever you see something original done by humans, that’s not because we have the magical capability to be original, but because you don’t know what the work in question was based on. And of course there are seven billion of us, while we only have a handful of AI models, so of course you’ll get a bit more variety out of humans so far.

        Either way, good image generation have only been available to the public for about a year. Give it some time. Humans aren’t much good at producing art after a year either.

        We need an environment that is actively and intensely hostile to AI tools and those that promote them.

        Better start by destroying your computer so those humans can have their job back.

        People calling themselves “prompt engineers”

        Those people will be obsolete in a couple of months, if they aren’t already. Since guess what, AI is pretty good at writing prompts itself.

        • davehtaylor@beehaw.org
          link
          fedilink
          arrow-up
          4
          ·
          1 year ago

          You are even free to plug a webcam into a robot and train it on whatever that sees in its lifetime.

          That’s not how life experience works. Also AI aren’t alive.