Diffusion types, 1st found in 2014, which add "noise" to pictures until eventually They're unrecognizable, and after that clear away the sound to deliver primary visuals in reaction to prompts. These applications use pure language processing (NLP) and generative AI capabilities