Two ideas stuck with me after our discussion on constraints and creativity. I was really interested in the idea that an artist’s style can be represented in logic by a set of constraints. Harold Cohen’s style consists of simple shapes and large swaths of color. The clean lines lend themselves well to systemization. Cohen seems the logical place to begin a study in the creation of art through logical constraints. I think it is human nature to immediately attempt to extend this experiment to more complex cases. Could one also represent the works of Picasso or Duchamp through equations and constraints? Can creativity be distilled into something cold and mechanical?
Sometimes I think people don’t want to admit that our brains are (incredibly complex) machines themselves. Given additional processing power and more complex algorithms, why shouldn’t machines be as creative as people? We call a human action “random” when we don’t understand the chain of past events and experiences that led to a given reaction. Modern computers use a simplified version of this artificial randomness, gathered from a natural event such as date and time, because computers do not have a body of past experiences and knowledge from which to make decisions. But I think that we’re beginning to see this kind of randomness in machines like Watson. Most of the answers Watson gives to inquiries are correct. However, sometimes it gives answers that we might describe as random, such as naming Toronto during a question about U.S. cities. This randomness is based on the vast body of knowledge Watson has access to—not dissimilar to the large body of knowledge humans make decisions based on.
I think that in the future, there will be more work done to understand the nature of “randomness.” Like we discussed in class, I don’t think there is anything that is truly random; it’s just a matter of dissecting a huge amount of data to understand the chain of associations that led to an event. It might not be comforting to imagine that creativity results from a similar process. It seems equally likely, however, to be the result of some complex mental-machine process that we have yet to understand, but which is nonetheless completely straightforward and even logical once the associations have been mapped out.