I am an independent author/publisher with 32 published titles. I use Midjourney and Grok to generate illustrations for illustrated picture books, with a lot of Photoshop afterwards. For my novels, I primarily use Grok as an on-demand editor to check basic grammar, beats, density, historical details, etc.
I then have to explain to Grok why I made the specific literary choices because Grok defaults to a neutral delivery when I require stronger or more period-specific language as well as situational needs for the characters in the scene. It's a process.
Explaining the reason to Grok confirms that I made the correct choice originally. I then put the polished text through a different AI like ChatGPT and/or Gemini to see if they catch anything additional.
When I have a question on historical events, inventions, or if I have an idea that I need to ensure it is logical for the locale and period, I pose it to Grok. This usually opens more more qualifying questions, and I find myself learning a great deal of history in a relatively short period of time.
Before it was mostly internet for the novel writing research, which is sort of early AI. Basically it took longer, I was lucky to find sources I could leverage, and there were stiff limits on using tools like Grammarly. I explained the changes in my first answer.
For the illustrated picture books, frankly they would never have seen the light of day without AI to generate the pictures, because I couldn't afford a professional artist, younger artists never completed the project and likely would not have stuck around for a whole series.
With AI, I have consistent images throughout the series, they look great, and the process is much faster than using traditional art. I have sometimes taken a picture book from concept to publication in a single week.
I do have decent graphic composition skills and at least average Photoshop knowledge, which are key to achieving a professional appearance in the finished book.