I can show you where the AI was used—I marked it in red (and the couch in the back too).
But even then, it was img2img after I had already drawn the base outlines, shadows, and curves. And even after using img2img, I clean up artifacts and redraw areas that didn’t turn out well.
As for improving my art, you can trace the progress starting from the BBC subroute and beyond. I finally learned how to draw hair properly, got a better grasp on keeping Hitomi’s face consistent, and figured out how to work with line weight. But I’m still not a professional—I still have problems with geometry, perspective and some details. Overall, patrons (at least I've only heard positive feedback.) are happy with the new improvements and have even asked for the initial CGs to be redone.