My 8-year-old son, Pedro, loves animals including the typical ones you can see at the zoo, in the aquarium, on a farm, and in the wild, but his favorites are the ones that don't exist, yet... We call them 'Hopeful Beasts.' Some are animal, some are vegetable, and some are mineral. Together, we have created a world filled with them. I draw the outlines while he spots the features that emerge. He then colors and gives each a unique name. He sometimes also tells stories about what is happening in the image. Inspired by early AI cryptoArt pioneers, we trained a GAN model to imagine a new collection of Hopeful Beasts using our original pen and colored pencil artworks as inputs. After each cycle, the lines and patterns of color more closely mimicked the style of our training set. After 100 generations, the images approached our ideal but have a style all their own. As they emerged from the latent space, in his typical fashion, Pedro, picked his favorites and named the never before seen creatures. We then used his names as seeds for AI to generate the names for the 100th generation.