Partially Derivative had a great segment on AI artists -- honestly: hilarious and thought provoking -- thinking about a new project wherein a team of programmers and art historians managed to train up an algorithm to reproduce the style of Rembrandt. The computer doesn't replicate Rembrandts, it paints new ones that replicate the style of Rembrandt. Vidya brought up the question: who is the artist here? Is it the programmer? Or are we willing to grant artistic ownership to the computer itself?
I like the question of who is the artist because I think it reveals one of our fundamental biases. We are biased to thinking that creativity is a human trait and are all too willing to dismiss anything a computer does as being somehow not creative...Because computers?
We can pull a quasi-turing test on this to really interrogate our biases: Suppose the Rembrandt painter is unknown to us, could be a computer could be an art student, we don't know. All we see is the art historians, &c. going into a room and training this unseen entity. We have the printer that will ultimately be used to paint the painting, when the time comes, and at that time none of the trainers will be present to directly influence the painter, and the painter generates a wholly new painting that it had neither seen before nor had been directly instructed to paint (i.e. paint by numbers style). How do we judge the painter now? Sight unseen I think we would be forced to accept that whatever creativity is in the painting (its uniqueness if you will) had to come from the painter, regardless of whether or not they were a computer.
Consider if we decided otherwise, that in fact it was the trainers that were ultimately responsible, and it turned out to be a human in that room controlling the printer with joysticks or whatever. In that case our judgement would seem to be in error. Sure your teachers influence you, that is what they are for, but students are not their teachers. So what is different if it is a computer?
Also: bleak robot future, even the creative work of painting is up for grabs with ML. It goes towards my general thesis that most if not all jobs are within the realm of ML without the need for strong AI. We can't see it because we imagine the computer will do the job the same way we do the job, replicating each step and thus running into a wall when it comes to the nebulous thinking step. But this is just wrong. There is no reason why an algorithm has to replicate how we do things to do the thing. A computer can be creative, just not in the same way that we are creative, and that is amazing.