> Civilization advances by extending the number of important operations which we can perform without thinking of them.
That's for mass production of useful tools. For decades, it's been easy to produce music, write a book, and draw with digital means. The actual truth is that people don't want to invest the time to learn how. Now they can pose as if they've done so, but the result never survives close inspection.
True, if you evaluate the new in context of the old. Anyone can summon a story with an LLM now, but it ain't gonna be as good as one written by an author pouring their heart and sweat into every word. Same with digital images.
However, the exact same technology is just a small bit plumbing away from allowing people to summon immersive stories in interactive 3D environments, and iteratively sculpt them from within, in the same way people in Star Trek did with the Holodeck (sans the actual holograms, but VR will do as a substitute). The building blocks are all there today, and are up to spec - what's missing is some glue and ironing out a kinks; I expect multiple attempts at this succeeding within next year or two. And this is a new kind of creative field, which will breed new mode of creating art (not just by chatting with a computer - there's a need for fine manual touch, too).
In the end, some things will be diminished, but new things will become possible. I'm mildly positive about it, but I guess we'll soon whether we'll gain more than we lose.
What you wrote can be summarized as:
1 Write Prompt
2 ...
3 Have a complete work
Let's take step 2. Drawing is easy to learn, but it's very hard to master. Why because, it derives from understanding and muscle controls. People fails to draw a simple cube because they don't understand the thing they see. Yes, they probably know that it has 6 faces that are square, but they haven't learned to see it. Knowledge is important, but it's just the initial step. You then have to understand how it all fits together (extract the rules), then apply that understanding.
Same with writing. Everyone has a story, but not everyone can tell it because they don't know how to. You have to learn the techniques, then apply it until you have a complete understanding, then you can create as much as you want to.
With no understanding, there's no creation. And generative AI won't help, because it can't feel the world as you and I do.
I'm reminded of my time in secondary school back in the late 90s.
I was trying to figure out how to make an FPS game of my own — from first principles given there was no internet for me to be advised by.
One of my peers suggested using flood fill as per photoshop, and I couldn't get him to understand that this didn't do perspective even with an example. (A real life "these cows are small, those cows are far away" moment).
Or that "Dulce et Decorum est" was required study material in the English Literature classes. Neither I nor anyone else in the school knew what it was really like to experience the horrors of war described in that poem, as the use of chlorine gas was banned before anyone under retirement age had been born.
Shakespeare is held in such high esteem, but how many of those who enjoy it today would have enjoyed an original performance of The Tempest in 1611 with actors who had little chance to memorise their lines, an audience who vocally participated, no light beyond what a fire could provide, no sound system beyond the cannon that accidentally burned The Globe theatre down? Can the works even be fully comprehended given that today the questions are "is Shylock an unfair and antisemetic stereotype?" and "why is The Taming of the Shrew even counted as a comedy rather than a tragedy?" and "why 'eye of newt'?"
That's for mass production of useful tools. For decades, it's been easy to produce music, write a book, and draw with digital means. The actual truth is that people don't want to invest the time to learn how. Now they can pose as if they've done so, but the result never survives close inspection.