Off the cuff thoughts on the previous note...
Why and how does someone choose a book, or movie, music or other human generated media? There is advertising, fashion, word of mouth, exposure, preference and, for want of a more well defined concept, serendipity. I don't see how those things don't apply to non-human, in this case AI, generated media. Having AI doesn't obsolete a need for a willing and engaged audience for its products.
Likewise I'm willing to bet someone qualified or knowledgeable in a field asks better questions of an AI than someone with less knowledge. The knowledge of the person asking the questions, at a minimum, saves time*. At junior school, I think, it's a hazy memory, a quite eccentric teacher asked us to write out instructions for how to make a cup of tea for an alien imagining they had no knowledge of any of the things involved. Which is a pretty difficult thing to imagine and explain, especially for kids. Essentially similar to asking an AI to make and explain a project involving opamps and ICs by someone with little or no knowledge of electronics. Doable, but not the same or as fast as an AI plus someone who knows what they're doing in the first place.
It's going to make things much faster for people who know what they're doing and just about possible for people who don't. Curation and an audience are another matter. A biological system the size of people is chaotic and we've been chucking CPU power at weather systems for over half a century. Prediction is difficult and feeds back into the reasons something is selected over something else. Add into that reflexivity and there is another significant factor - an AI controlling a market would lead to something akin to control until external factors invalidated the model.
An AI can make something that sounds like The Rolling Stones but it's not the same as The Rolling Stones asking the AI to make something that sounds like themselves. More of something cheapens it, but the people asking the questions matter and so does the willingness or demand for consumption of it. I am 100% convinced AI will approach or supercede human capabilities, but in some areas in many plants and animals they have and it took billions of years - whether silicon and it's successors shortcut that in an equally energy efficient way is far from certain. And on what timescale. Six hot data centers, covering many football pitches, and consuming vast energy does not a toddler make. I'm more optimistic about things like self driving cars, although it's a very difficult problem, because it's a subset of the problems involved.
I haven't checked for typos or grammar. Get an AI to do it.
* Possibly lives.