Inspired by Frank Chimero's 2013 piece, "What Screens Want," where he defined the "grain of screens", I want to explore the question, "What is the grain of AI generated art"? In Frank's words, grain is how a material has "grown and matured that describes how they want to be treated." What trace of origin do these generative algorithms leave within the images they produce? If we trace the way AI art has developed, the way one would trace a finger over the age lines and curves in wood, can we learn its strengths and limitations? Is AI art like a velvet of hydrocarbons and silicon, where the hand of probability brushes it smooth or rough depending on the prompt? Or better understood as a ripple shimmering on the surface of a giant unseen sea of data churning around ever-elaborating models? My proposal is an essayistic journey, accompanied by images, to locate the grain of AI-generated art, in the hopes that we, as artists, can determine the best way to work with (and equally importantly, against) this new medium.