It was a design choice from the beginning. Warlord began as an experiment to speed-up development and lower cost. This method allowed me to work about 4 times faster and work on two games at once. For comparison, my first game took 1.5 years to complete. Warlord is longer with more animations and will likely be complete in 9 months.
I probably won't use the same exact technique in any future title, but I've learned a lot that should make the next titles better.
I'm certain there is a trade-off of image continuity. I think most of the backgrounds are okay and the others are getting better as I slowly replace them (the "evolution" I was referring to), but its not there yet.
Warlord might be more popular if it was done a bit differently, but it wouldn't have existed at all with the AI tools to help. I was already working on Tragedy of Medusa at the time and could only really spare a day or two a week for experimenting on anything else.
That said, Warlord is my most popular title this year and is trending to become the most popular of my games overall (overtaking Euryale's Gambit probably next year). I attribute it's relative popularity to its genre (male-protagonist), but story may also play a role? At any rate, its' hard to tell how much using AI as much as I do hurts the project if at all. It also happens to be my highest-rated title :shrug:.