We've ended up with an interesting discussion that seems to be flowing in several different directions at once.
PJ's point on diversification is certainly worth considering. It doesn't hurt to have multiple revenue streams, at the very least. We do need to be conscious of the fact that there are only 24 hours in a day, though. A writer can only do so many things, particularly if they involve separate workflows. At the risk of stating the obvious, books and YouTube videos are very different things, requiring different kinds of content and different skill sets.
As far as the audience changes are concerned, it's true that there are many more competing activities than there used to be. But even some of the articles PJ linked provide caveats, for example, the reference to the fact that people have been complaining about the decline of reading for generations (the example goes back to 1907, but I think we can find earlier ones). Also, there seems to be a disconnect between sales figures (which, if we estimate some increase from unincluded self published texts, are either stable or rising) and surveys that show people reading less. One of the linked articles suggests that buying a book doesn't mean the book gets read. But if the disconnect is caused by that, and we want to take a ruthlessly businesslike attitude toward the issue, we make the same amount of money on sales whether the books actually get read or not. That's not my preferred outcome, nor probably anyone else's. But from a strictly business standpoint, sales are what matter.
There is one other point I'd like to touch on briefly. Some of the linked articles reference students no longer being required to read full books. But as another source points out,
There is not a lot of information on how many books American students are required to read in school. But in general, students are reading less. Federal data from last year shows 14 percent of young people said they read for fun every day. In 2012, that percentage was 27 percent.
https://learningenglish.voanews.com/a/do-students-in-american-schools-read-long-books-anymore-/7789204.htmlNotice the weird disconnect between the statement about reading in schools, which is then joined with survey data about reading for fun--two very different things. Always be on the lookout for these kinds of odd conflations. They're common.
Also, there is a tendency to blame schools for whatever is happening. I don't doubt that incoming college students said they'd never been asked to read a whole book before, but I question how often that's actually true. I just checked with my old high school, sort of a typical, middle-class institution, and every English class requires at least five books per year. (At the school where I taught, the number is higher.) Checks of several other schools didn't turn up a single one where the reading of books wasn't required.
There has been some tendency to try to introduce more contemporary titles that correspond to students' actual interest. And there seems to have been some slowdown in reading during the pandemic. But I'm having a hard time finding evidence for the kind of sweeping degradation implied by some of the linked articles.
On the issue of whether or not readers will accept AI, a quick search turns up no hard data, just a lot of people speculating. Some writers like AI, and some don't. I see no reason to assume that most readers will accept it unless a writer has worked on the books enough to make them at least sound human. But I wonder why it is that the AI companies come down so hard against labeling if they don't think it will make any difference to people.
Will AI just keep getting better? Maybe. But there are a number of barriers. And remember the recent incident with Grok, when Musk was trying to make it less "woke," and ending up having it call itself MechaHitler and recommend the immediate opening of concentration camps. To be fair, that didn't last too long, but it illustrates that recrafting Ai is not always the easiest thing in the world to do.