Craftsmanship is the mastery of gradient descent, which does well at zeroing in on local optima. Creativity is the ability to reliably find global optima. One interpretation is that artistic life consists in daring to seek the global optimum of the only function one knows: one's felicity function. The artist excels at controlling the tuning parameter: first, searching for the optimum globally, broadly, and then becoming a craftsman and searching locally, finely. This description of the artistic life misses the social aspect of creativity, however.
Suppose instead that an artist is but a node in a neural net, an ever-evolving activation function, an ever-shifting perspective. He filters other artists' outputs, art, through his perspective and passes it on, for other artists to soak up. One is not an artist if one refuses to output, to share art. But then, is everyone an artist? And what is the objective that is being optimised, that is, what is the task that the neural net is learning to perform? And where does the training algorithm come from? Who is in charge of it?
"We are like the dreamer, who dreams and then lives inside a dream. But who is the dreamer?" Perhaps, it is the Ultimate Artist himself.
