AI models can be built that take one medium of input and return a modifed creation of the same medium

Image-to-image, audio-to-audio, video-to-video, and text-to-text are all examples of different possible input-output mediums AI models can be created to support