Diffusion models, to start with found in 2014, which increase "sound" to images right until They can be unrecognizable, and then eliminate the sounds to generate unique pictures in response to prompts. Artificial superintelligence (ASI) could well be a machine intelligence that surp