Публикация

If artificial intelligence output were good and useful, each re-input should create a refinement. Since re-input corrupts the model, the nature of the output is obvious.