Diffusion types, 1st seen in 2014, which increase "sound" to pictures right up until they are unrecognizable, after which you can take out the sound to generate initial images in reaction to prompts. The potential for artificially smart units replacing a substantial chunk of modern