Vincent Van Gogh, the original mad genius, was a vessel rot with mental illness, decayed by malnourishment, insomnia, and alcohol. If McLuhan’s the medium is the message applies in 1890, then Vincent’s body of work is a mirror for the winsomely disturbed. His work is revered, and many pieces have become the most expensive artworks ever sold. But what makes one artwork more lauded than another? Provenance? Historical and cultural context? The taste-du-jour of the one-percenters? An arcane debate for historians, a handful of aesthetic philosophers, and a few billionaires with cash to burn and people to please.

But this article is not about Vincent, because I read something recently that left me in a perplexing spiral of historical and philosophical uncertainty. Google—our AI overlord—ushered in a new art movement they call Inceptionism. Using an artificial neural network (a computer algorithm model built to mimic the way a human brain thinks), they managed to create what could be perceive as “art.” Before we continue, critique for yourself:

Ethereal, vivid, and sparklingly imaginative, I can only dream this is what Frodo Baggins would see if he dropped acid on his way to Mordor. People seem to love it though. The computer made creations went at a charity auction in San Francisco for around $8000 a pop. Not bad for some computer code. So how does Google’s Inceptionism do it? The same way all artists create: Through a regurgitated mix of past experiences. The fundamental technology that powers Inceptionism was invented as part of Google’s machine learning research. Google wanted to recognize and understand what’s actually in images. To know that a photo of a curved phallic yellow object, is in fact, a photo of a banana.

 You can understand why Google would want this power. A human could look at a photo and know with unwavering certainty, “that is a pineapple wearing sunglasses.” But a computer? Enter the artificially intelligent brain powered by digital neurons. The artificial neural network (ANN) is “trained” by being fed millions upon millions of reference images. Like a baby’s brain absorbing everything it sees. The ANN has “learned” what both pineapples and sunglasses look like. When it “sees” the image to be identified, it locates the edges and corners of objects, and then classifies them into shapes and colours. When this is cross referenced against its massive database, it outputs a solid guess of what’s in the picture: A pineapple wearing sunglasses. Cool, huh? A computer with the cognitive ability of a pre-schooler!

… But what about the art? This is where it gets fascinating. And scary. This artificial brain is so good at recognizing and understanding images, it can also generate them. When this artificially creative computer is given a blank canvas of RGB digital noise, it can be instructed to “make this look like that” - and effectively tweak the noise into different shapes and colours based on what it “knows” different objects look like. 

The AI brain’s “dream mode” is where the neural network can run rampant interpreting shapes, transforming them on its own volition, essentially creating an original work based on past experience. It’s not just a glimpse into the future of machine vision, but an unsettling insight into how creativity arises in the first place, and maybe what “being creative” as a human even means.

In Kazuo Ishiguro’s novel Never Let Me Go (also a quasi-decent film starring Keira Knightly), a group of clones spend their days making and enjoying art at a boarding school. These clones exist so that the counterpart human could harvest their organs, with each clone lasting for only four “donations” before they’re disposed of. It is revealed in the denouement (SPOILER ALERT!) that the boarding school was not a sinister faux-utopia distracting the clones from their inevitable death with finger painting and poetry, but actually the exception. A rogue experiment trying to prove to society the clones were in fact human because of the deep and emotional art they produced. That if creativity is an innately human experience, then these clone lives must be spared.

I bring this all up to help answer the question: Is Google’s machine brain thing making art? Can machines be creative?

I suppose when it comes to Google’s art-producing brain the true artists would be Alexander Mordvintsev, Christopher Olah and Mike Tyka, software engineers at Google that programmed this being. Maybe they are the human heart in a creative process that otherwise lacks any soul. I believe that what separates man and machine is our ability to suffer. Until a computer can become so artificially intelligent it creates its own software to make it process with crippling self-doubt, corrupting existential stalls, and become so overwhelmed by the pain and beauty of the world that it purposefully fries its own chip, then it will never be able to create true art. Because true art mimics life not just in image, but in spirit.

Vincent Van Gogh is a revered artist because his (torturous) creative process is almost as famous as his work. You can see his pain through every brushstroke. What gives his work value is our attachment to the man. An emotional connection that seems to span through time. I look forward to the first depressed robot, I will take its art much more seriously.
 

Written by James Alexander Dunphy