I asked ChatGPT to describe a linocut it would make based on Edgar Allen Poe and then how would it ask an AI to make an image of it.
ChatGPT calculated this: '"The Raven": "A large raven perched on a branch, with a skull and a quill pen nearby, a faint image of a man in the background, a book in the front corner and a moon in the background"'
Then I fed that to Stable Diffusion 2.1 and got this output.
Maybe I should carve and print this? Would it be art?
@bryanhansel Did you hear the piece on NPR today about this very topic?
@daphsci I didn't. They were doing linocuts and Poe?
@bryanhansel no but it was about appropriation of someone’s oeuvre (I think Van Gogh) to train an AI bot to do art. The vibe of the piece was consonant with the other commenter here. It’s not necessarily can we do it but should we do it?
@daphsci it's an interesting question. I'd like to hear it.
In this case, I enjoy the absurdity of one AI generating input for another and the the output being made by a human. It's usually humans making robots make stuff. I'm dreaming how this could become a art gallery show. During the opening, I could use Microsoft's voice simulator to read an artist statement and give a talk about the project in my voice reading copy from ChatGPT made in the style of my writing.
The absurdity is so fun.