I understand that all images are processed, but AI for me is crossing the line as it is adding things that come from the trained model rather than derived from the image itself - these elements otherwise wouldn't be there regardless of lens, colour grading and so on.