Full disclosure!

 So I'm learning to play with ai art generators, Leonardo.ai in particular but I'm also using canva.com and the previewer on Mac to mess around with the resulting images.  I've updated a few of my book covers just for the fun of it, though the smashwords editions might not be approved as the metadata and the cover wording aren't exactly the same.  I haven't gotten around to adding a note in each book about the cover source (formerly my own photographs) but I will get there someday, until then, this post will have to suffice.  And as far as creative content goes, I would say it is as difficult to get a good ai book cover as it is to take a decent photograph, it requires patience, time, a learning curve, and a little innate skill, but that's just my opinion, maybe I'm just not proving to be a child prodigy but as far as I can tell, to get a good image takes a bit of work and time and luck and a good eye.

AI book covers?

 I finally got around to playing with an AI image generator.  I've long wanted to see what all the hype is about and if it will really put human artists out of business.  I don't think all you content creators out there have anything to worry about, rather you have yet another tool to incorporate into your craft.  If you insist on handwriting your manuscripts even after the advent of the typewriter, then yes, you will become obsolete, but for everybody else, it is a pretty neat tool, but still a tool: it needs human input and creativity and taste to make it meaningful.  It also requires time and talent and creativity, you can't just push a button and there it is.

I played with an image generator and read a couple articles, one an interview with an author who uses AI to help brainstorm and organize his books but found he had the same problems with print as I did with the images: they don't necessarily make a lot of sense!  The human touch was still needed to edit, refine, select, hone, and direct.  The computer can spit out a boundless array of images or text but to make it truly creative and artistic, someone needs to direct it.  I also found several articles on copyright law concerning computer generated stuff and it agrees that to be copyrightable, the computer's junk must be organized and edited and changed significantly by a human person.

It is a ton of fun if you like this sort of stuff but unless you are interested in editing the resulting images, I'm not sure it is a great option for indy ebook covers.  Here are a couple examples:

Robotic sci-fi unicorn on steroids in a post apocalyptic landscape?

Three legs and two moons?

Five legs and a donkey in the moon?

This is a decent image but took editing and learning what prompts to use, of 200+ images, about 3 were usable!

So there is definitely promise here, but like any other tool, technique, or creation it requires time, patience, and the human touch!