I wonder how you perceive photography. If you compare the creative process to create a picture between taking a photo and painting a canvas, there are some striking similarities. The photographer "just presses a button". Yet photography is an established art.
Sure, there is slop made with AI, because it is easy. Just as there is slop made with photograhy apparatus. Or slop made with cookie cutter templates on a premade game engine. Creating yet another pixel art rpg with a story seen so often there are tropes about it. There are even engines that advertise as needing no coding knowledge.
your entire creative process
You act like using AI helpers is like asking the machine: hey google, make a video game and put my name under it. Just like a photographer just pushes the button and "created" a picture. "No" creativity. If it is enough for the photographer to decide what's in the picture, why is it not enough for the AI operator to decide what's in the image? That's why I am asking how you perceive photography. What's the difference for you. Unless of course you have a similar opinion about the creativity of taking photos.
And what do you think about people using a magic box to have their game do things? Speaking of game engines. Someone using that engine relies on the blood and sweat of the people that programmed the actual routines to move pixels around. It's standing on the shoulders of giants.
What do you think about a movie director or an orchestra conductor? Are they "creative"? It's the actors that act, the musicians that play. The instrument builders built the instruments and the notes composed by a componist. The movie script was written by someone else, and the cutter puts it all together. Is the creativity of conductor and director outsourced?
Anyway, I suggest you read the post you replied to again, what was actually done with the llms. Googling an example from stackoverflow and adapting it to your needs is not really much different from asking chatgpt to generate a code snippet. Only, the chatgpt snippet might have bugs in it that are not obvious. In both cases, the coder did not spend days learning how that code fragment would be done from sratch. It is standing on the shoulders of giants, relying on previously done work. Googling it, is a huge improvement from digging through examples in written books. Summarizing it with the help of a llm is a further advancement, but with the risk of it being very wrong.
And refining text with llm is a glorified spell checker. I hope you do not use those when you write emails or texts. The knowledge in that spell checking relies on previous works of other people. You would be outsourcing your language skills.
Visual art is a bit different, as you do not need an image that is functional, as you would need a functional code snippet. That's probably why ai images look so bland. They fulfill the functionality requested. For code this is enough. So much enough, that one uses libraries full of functions and even whole game engines that do not even need coding knowledge anymore.
You can be against the usage of AI for all sorts of reasons. And we all should be, for things like the slop created with it. But things like spell checking and coding assistance, in terms of creativity, are state of the art methods, of what programmers are doing for as long as there has been programming. Reuse, rehash, copy, paste, do calls to things other people wrote.