Can you get it to do the strings bit while you do the others?
I suppose the problem is the required 'quality'.
For example the stock photos of say "people round a boardroom table looking really excited at a powerpoint". You could easily spend a day setting the shot up, styling the hair, choosing the clothes perfecting the lighting , getting a great power point up . Perfect sized table . A really good shot.
And charge, well a decent amount.
But for what it's going to be used for , a free image that fulfills the basic criteria is more than good enough.
As an analogy, good food (or wine) is pretty much wasted on my as I don't have the palette to appreciate it. A £20 decently cooked meal and a £200 super chef created one are more or less the same to me.
The shot in your hypothetical will take time and, usually, personal expense. The people in it will be paid. The photographer or photography studio will take that financial burden knowing they can get 5-10 years of small, accumulative passive income from it. There’s often no big pay day for most stock - especially after Getty etc get their cream off the top. AI will be able to do that without the photographer, models or any crew. Getty/google or whoever will be free to undercut any humans still making stock and still make more money.
As it gets more sophisticated, the more of the industry it will consume. Celebrities may, and doubtless will, take advantage of AI to do promo shoots for them. Ad work where they no longer have to spend a half day on a photoshoot but instead license their AI likeness.
For your music question, if it was a tool that did it, it would certainly help, but maybe remove some of the satisfaction- not all hard work is a chore if there is a payoff at the end. It’s kind of like our competition with city - the feeling that no matter how hard the team compete, It’s not a fair fight.
Even the ‘hum a melody’ bits would be a help. But that has not been the focus of AI. They’re focusing on creating standalone, complete music.