• fiasco@possumpat.io
      link
      fedilink
      English
      arrow-up
      5
      ·
      1 year ago

      At least you won’t be feeling stagnant, for a while. But I’ll answer your question more completely.

      Various things have been discovered that have allowed a certain amount of automation in storytelling, but one thing that can’t be automated is passion. By automation I’m not talking about “artificial intelligence,” I’m talking about—what programmers call “tooling.” Movies nowadays are almost always visually stunning, and that’s because of algorithmic work in light and shading, character animation, hair simulation. Similarly, there’s also a “canonical story” you can read about in a book called Invisible Ink.

      The canonical story doesn’t tell you how to write dialog, and not surprisingly, dialog has become incredibly weak. On one hand you have capeshit, where characters talk in quips, and on the other The Rings of Power, where everyone talks in weird, deep-sounding non sequitors.

      This is what I mean by risk aversion. A lot of beautiful graphics conveying nothing. A lot of electricity used to run computers for no reason at all. This is all very expensive, and expenses have to be justified with spreadsheets.

      There are still good things out there, there’s still passion in the world. It’s just getting harder to find.

      • NeonPayload@infosec.pubOP
        link
        fedilink
        English
        arrow-up
        1
        ·
        1 year ago

        Thank you so much for showing me this. I’ve always had a hard time describing everything looking the same, sounding the same, feeling the same. With Execs trying to push AI into storytelling it’s just going to be more of the same.

        • fiasco@possumpat.io
          link
          fedilink
          English
          arrow-up
          3
          ·
          1 year ago

          And while there’s a lot of technical nuance to this, deep learning is nothing more or less than random recombination of its input. The difference between GPT-2, GPT-3, and GPT-4 is how the randomness is conditioned, but it doesn’t change the core fact: trying to use deep learning for storytelling would be the greatest breakthrough in cultural stagnation that the world has ever seen.