You are viewing a single comment's thread from:
RE: AI is stupid: Humans Judgment Matters
If you know that "pleasure"/[insert label here] is involved then you can make that one of the things that your AI optimises for. You could use a GAN where the Adversarial network outputs not just a guess on real/fake, but also a guess on "pleasure". Supply the neural network some information about pleasure and you might have something.
I have some ideas on how to identify aesthetic experiences based on introspection, but we'll see what some AI experts I know have to say about that before I talk too much more about it.