As the human animal has progressed from small hunter-gatherer tribes to full on society, we’ve reduced the time necessary for works of survival (finding food, water, and shelter) via technological advancements from the invention of farming to the invention of GrubHub. This has granted more and more members of the species the free time necessary for works of passion, and throughout history we’ve built up quite the cache of amazing art.
Artificial intelligence and machine learning are seeing a similar progression; as more easy to use AI packages become available AI is no longer achievable only by PhDs in a few select disciplines, it has become accessible to coding hobbyists. As a result, AI’s applications are no longer focused solely on lofty academic research or high-level corporate strategy, and can instead be applied to anything any hobbyist might imagine.
Art is a perfect application for AI/ML. Artists consume the art of others, and apply the aspects they like to their own art. Similarly, the machine learning paradigm is all about inputting data into an algorithm that learns the pattern, identifies and ignores the noise, and produces an output reflecting the input.
One example of AI making art is Google’s Deep Dream, which takes an image as input and outputs a psychedelic version of that image created by a neural network (the image for this article is an example of Deep Dream). The network mutates parts of the image into other things that look like that part, then feeds the generated image back through the network again and again to produce the final altered image. Google has a blog post about it for those interested in learning more about how it works, it can be found here: https://ai.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html.
The University of Michigan’s Michigan Institute for Data Science (MIDAS) is also doing some interesting things with art and data science, in this case applying it to music. They’re creating a crowdsourced music producing system, among other things. Those interested can learn more here https://midas.umich.edu/research-hubs/music.
However, AI/ML art isn’t only being created by large corporations and prestigious universities. Any quick Google search will find tons of user created ML art, and a quick Youtube search will find tons of user created AI/ML music. Anyone with an idea, a little bit of knowledge, and enough motivation is capable of expressing themselves through algorithmically created music.
Machine learning and artificial intelligence are shaping more aspects of our life every day. Through recommendation, algorithms decide what we eat, watch, read, and listen to. They choose the layouts of our grocery stores, and diagnose diseases that doctors may not catch. So don’t be surprised to see a Picasso hanging next to a C++ in an art museum, or hear a collaboration between Selena Gomez and DJ Neural Network on the radio in the near future.
Image source: Google Deep Dream