Recently I've tried developing algorithms that sift through noise + trial and error guided by facial recognition to generate faces, but then I realised what would be more interesting would be this: to have a human draw the face as directed by facial recognition software. The drawing tool is a really basic interface made quickly in Processing, but the principle works - when my drawing receives a higher face detection score I progress with my drawing, when I score lower my last mark is undone. Its a very strange feeling, like being one of those artists at a police station working with a witness who is trying to describe a face, but limited to a vocabulary of just 'getting warmer, getting colder'. A reverse E-FIT program where the machine is trying to tell me who it saw.
Lately I've been really interested in new relationships in labour between humans and computers + algorithms. Amazon warehouses dictate to humans what to do. Mechanical turks await instructions at a computer to assist larger autonomous systems. Chess players receive advice from algorithms in a game of hybrid chess.
I'm hoping to improve the drawing interface, perhaps try drawing with a tablet, and also test the idea with 3D model sculpting, but as always I'm happy to share the beginnings to an idea.