Wednesday, February 10, 2016

Unstructured media and the challenge for computer science

Unstructured media: variable sized, new, hyper, dangling, broken, unfinished, context sensitive, made up, random, chaotic, ungrammatical, incorrect, paradoxical, and incomplete phonemes, words, sentences, grammars, and meta-grammars. Likewise with music, speech, drawings, points, vectors, surface, volumes, dimensions (geometry), painting, sculpture, pictures, video (pixels, voxels, time rate or size), taste, smell, motion, emotion, thought, haptics and feeling. If we can find a way to represent these in XML or JSON or … I think we have a good chance of handling almost everything. Thoughts?

3 comments:

Unknown said...

I'm not so convinced that carbon-based sources of these things are not made up of yes and no decisions just in a different physical medium than computers. I don't see how one could do all this with computers as we currently know them. I'm also not sure of the appeal of such a thing. I kind of like listening to human performers and going to galleries and theaters to view art made by humans. If I want to create something which has all that potential for things a computer cannot do I would probably get pregnant.

Just sayin' :)

yottzumm said...

I agree with the get pregnant. Dealing with incompleteness may be a quantum computing thing.

Unknown said...

I agree re quantum computing or some other such technology that we may not even have envisaged yet.

In all seriousness though (said from the perspective of an ex classical music composer and performer) one of the most fundamental aspects to making any machine create art would not be teaching it the rules of art or music (of which there are a zillion anyway) but teaching it how to break the rules at strategically appropriate times and places to elicit an emotional response from viewers/listeners. I believe most people appreciate art because it somehow creates an increased, different or decreased emotional state than what they previously had.

I don't think you would need to teach a machine to understand what emotion is or anything like that but you would need to yourself analyse what makes some strategic rule-breaking turn out as good art and some tasteless garbage. Given the problem of subjectivity and different taste among humans, that could be a tall order.