I'm a bit more jaded when it comes to games art, or maybe snobbish lol. I also have some industry experience. I dislike a lot of art direction and consider much "dilettante art" that is just made very detailed or overly cool. It often works and is fine, but it's not art. The real art you find in indie games. But take a look at zerospace for example, it's pretty detailed and technically good 3D unit models, but overall it's pretty pedestrian art direction.

And that is not always easy to avoid. Like you can only tell your concept artist that his art and color choices remind you of a preschooler so often before he cries and rage-quits, leaving you to have to find and hire a new underpaid slave. Or your modellers that they'll have to redo all of this garbage and stay the weekend. What you already had to stay the weekend unpaid because I made you believe in this project??? /s obviously lol.

Video game development is technically very demanding, underfunded and very badly paid and has time constrains, and a high complexity and "unknowability of complications" what gameplay elements need to be redesigned. I recently watched an excellent Diablo III post mortem of the development hell they went through, despite their best intention and talent and funding - in the end it was rushed out.

So being able to prompt an AI to create or make variations of concept art and then iterate EFFORTLESSLY over the overall composition of the resulting work - my theory is that it will allow for greater artistic freedom. At least for the "game director" similar to a movie director. It does not just have the potential to be cheaper and produce more slop, it also has the potential to elevate and empower to control and iterate over the creative vision. The source code could be like an annotated game design document in a special markup language with inserted links to approved artwork and random seeds that you can just tweak and rerun the whole game generation. Not just for visuals and audio, but also for the gameplay.

So with gen AI you might potentially end up seeing more and better games that can be considered artwork or masterpieces.

Obviously this is all speculation, if this actually can work technically and economically. And AI psychosis is real, there might be a dead end in genAI and LLMs ahead instead of improvement. Also friction and pain in the media often breeds beauty. See how well book authors do without a publisher and editor to keep them on a tight leash lol. But I do think it is well worth considering, investing and exploring this potential.