Read an article yesterday written by someone who ran his old prompts (from last month) through this week's Midjourney update. Now the images look better.
Almost every objection you raised the AI will cover very soon. None of the generators are "done." More and more source art is being added to the database and being analyzed by the AI every second, 24/7. Automatic bot scripts don't ever have to stop.
The being able to give very specific notes to the human artist is the only thing that can't be duplicated by AI... Yet.
But how many "large" game companies are there? Not many. The majority of game companies are small operations and they WILL start using AI art all over the place. Same goes for many. The amount of visual art (and here we're talking TV and Film along with comics, photos and single image art) being produced by companies with actual budgets is a tiny little fraction of what's being made by freelancers, independents, students, hobbyists, and hopefuls.
I do know, for an absolute fact, working artists in the VFX industry are already losing jobs to AI. I know a couple and friends in the industry have told me about departments being downsized in favor of an AI workflow.
Hell, a company I've worked for is doing a livestream later today about integrating AI art into their workflow. Several other of the "big names" in VFX YouTube have already done videos on integrating AI into workflows, including Hashi - better known as Action Movie Dad - an Oscar nominated former DreamWorks artist now working at Red Giant/Maxon used all AI tools to create an animated character for a show. The base image was generated by AI, a voice was generated with AI, the animation and lip synch was all done with AI. Hashi discussed how he stared with nothing and had about three minutes of completed animation in under two hours. Only thing Hashi really did was write the dialog for the AI. It's definitely here to stay and it's not a fad or curiousity. The art world juts changed forever (again).