So, I went through my portfolio the other day and found AI generated photos don't necessarily sell better than real photos in my case.  I think it varies depending on contributors, but in my case, I think real photos are actually more likely to sell.  However, it's way more efficient to create AI generated images and upload than shooting real photos.  So, that's where AI generated photos win as a business for me.  It's cost effective.  I don't have to hire models, go out on a shoot, find location and pay for the locations etc.  While some AI generated photos have sold well, most of them never sold.  Many photos that I imitated bestselling photos didn't sell at all.  Those are too saturated in the market because everybody makes them.  So, I think you'll need to find you own niche as well rather than just prompting imitating other bestselling photos.  It's supply and demand balance.

Interesting and useful feedback! Another factor to consider is that presumably, the AI photos have been available for a relatively short time vs. actual photos?

I don't know if AI images being a novelty is working as advantage, but maybe it is.  But there's no need to stop shooting real photos at the same time. 

it's way too early to tell - i've been doing ai for about 18 months but a lot of that time didn't produce competitive images but had modest sales.  older images continue to sell well, while my portfolio has many hundreds of images for each ai image

i have sets of ai people images based on best-selling editorial images, which have started to attract more sales, likely as they can be used commercially