The program that creates art by looking for patterns and making bizarre mistakes will make you think about the future of artificial intelligence.
In I Have No Mouth and I Must Scream, science fiction author Harlan Ellison writes about AM, a supercomputer that creates clever ways to physically and psychologically torture a group of humans. AM controls their minds, their bodies, their desires, and everything in their environment, which makes most of the characters' hopes turn into despair. I remembered the story when I read about Deep Dream, Google's artificial intelligence program that creates surreal and artistic worlds from the images you upload. However, unlike AM, Deep Dream doesn't create them in order to torture others, but as a means to search for patterns and faces within an image. The strange artistry of this program will make you think about the future of artificial intelligence and the ability to build virtual worlds. From that, we might see that AI won't necessarily bring the destruction of mankind, but rather bizarre and cool results.
At a gallery in San Francisco, Google's engineer Blaise Agüera introduced the works created by this series of artificial neural networks, explaining how they work like the web of neurons in the human brain. Based on that, Deep Dream behaves almost like a child, since it’s taught to recognize visual patterns to automatically classify images. The program uses the same technology as Facebook to identify and recognize faces in pictures. Through the use of algorithms, it attempts to recreate the psychological phenomenon of pareidolia, that is, our ability to see faces or humanoid shapes when they aren't there. Therefore, it ends up seeing faces and animals in the shapes of clouds and other natural landscapes.
Hundreds of people are willing to pay a lot of money for these surreal images that will remind you of a bad acid trip. But some may think that these works have no artistic value because of their apparent lack of authenticity. However, using all the available resources to make art doesn't make it less legitimate. The truth is that we've done that since the very beginning of artistic expressions. For instance, Renaissance painters Hans Holbein and Johannes Vermeer used tools such as mirrors and lenses to achieve innovative effects on their paintings, and so increase the quality and inventiveness of their work.
The development of programs such as Deep Dream shows that art and technology are no longer separate. While some artists might think they're safe from machines stealing their jobs, the growing abilities of artificial intelligence show that soon they might also start aiming at their area. For instance, now engineers are developing AI programs that compose music and write novels that compete with human authors in literary awards, and surely other similar programs will follow. In this sense, artists will have to leave concepts like authenticity and originality in the past or work twice as hard to innovate and compete with machines.
We might be overwhelmed after seeing machines being more and more autonomous, being programmed to work without the need of human instructions. The possibilities are infinite. What will be the next step after Deep Dream? Well, Google is already showing some interest in virtual reality as the next step for its trippy program. Then, it's likely that in the near future we'll be able to navigate virtual realities made by advanced versions of Deep Dream without any human intervention. We don't know the kind of realities this program will create. It could be a distopic nightmare with infinite torments, like Harlan Ellison's supercomputers, or rather an endless feast of fulfilled fantasies and desires. No matter the outcome, we know it won't be boring.