A new horizon, or the end of the road?
In October of last year, a painting titled “Portrait of Edward Bellamy” sold for $432,500 at Christie’s auction house in New York, N.Y. In ordinary circumstances, such a price tag in the exorbitant New York art world might not fetch any noticeable attention. “Portrait of Edward Bellamy,” however, is no ordinary work of art. The painting by Obvious, a French art collective, was created with GANs, or generative neural networks. In layman’s terms, GANs are machine-learning systems that synthesize data sets containing thousands of data points and repackage that information into a new product.
The art world took notice of this landmark sale, as AI technology has been frightening markets with its autonomous capabilities. Artificial Intelligence has made its presence known in the creative realm of art long before the sale of “Edward Bellamy.” Its capabilities and potential may be limitless as it operates more efficiently than any human artist ever could.
As scientists and artists explore the possibilities of AI technology and its works, poignant questions arise over the purposes of art and its relation to human expression. Self expression has been inextricably linked to the art we create. So can AI truly make art as we imagine it, and if so, should we allow it to?
Since the “Portrait of Edward Bellamy,” AI has progressed to offer more human-like and diverse pieces of art. The GANs used by Obvious is limited to the art in the dataset that it uses and it cannot add any original artistic flair to the paintings it creates. It picks up patterns from previous works and reshapes them into something different, but not entirely original. Dr. Ahmed Elgammal, a professor of Computer Science at Rutgers, sees GANs as a nascent and rudimentary stage of what artificial intelligence is capable.
Elgammal described his vision for the future of AI art with AICAN, his Creative Adversarial Network in The Atlantic. AICAN is GANs with a major twist. Instead of drawing from art solely based on its neural network, it also adds a completely new touch or style distinct from what it has learned from the neural network. This added dimension shifts the artwork slightly more into the realm of novel creation. Painting isn’t the only genre of art in which AI is making headwaves. It’s making leaps and bounds of progress in music as well.
AI has become an increasingly important part of music since its introduction in 1963 when 17-year-old computer scientist Ray Kurzweil displayed a computer program and machine-learning system on the TV program “I’ve Got a Secret” that recognized patterns in classical works and created new original works.
Since then, AI has developed to instantaneously create instrumentals based off the neural networks and datasets of music. It swaps art data sets for instrumental datasets and analyzes the patterns of these instrumentals to interpret musical patterns with music theory, forming entirely new beats.
Currently, the state of musical AI isn’t to the point of creating Grammy-worthy work, but it is capable of creating catchy and soothing beats, suitable for lounging or as background music. AI’s improvements in this field have startled some who are worried that its abilities will eventually surpass and supplant human artists. Other AI optimists have taken its qualities in full stride, using it to complement their work, making the creative process more efficient.
AI allows artists to explore different pathways and ideas in their instrumentals at a faster rate than engineering them manually. Holly Hendron, an artist and a pioneer in the AI music world summed up the aspirations of AI music in an article with Spin magazine.
“I’m more interested in not how we can write ourselves out of the creative process, but rather expand the creative process by augmenting what we’re doing with this technology,” Hendron said.
Many artists like Hendron believe that AI holds the key to unlocking new musical possibilities. But others are cautious of allowing AI too much traction in music. One of the disconcerting elements of AI in music is its ability to replicate human musician’s voices and styles creating original works that play off already established artists. This can open the door to many legal battles, as the copyright world tries to discern whether a replication of style that doesn’t use an original “sample” holds the same consequences as intellectual property theft.
“There’s nothing legally requiring you to give her any profits from it unless you’re directly sampling,” said Meredith Rose, policy counsel at Public Knowledge, in an article in Verge.
These issues are still being mulled over as AI is in its infancy with no true formidable rules or regulations to govern its domain.
There is no telling how far the AI artistic capacity will grow in the coming years. It remains to be seen whether it will end up being an incredible tool for human artists, drastically elevating their abilities or an autonomous steamroller, demolishing everything in its path. This question may boil down to whether we determine the artist’s emotional and creative struggle as a vital component of art.
Written By: Andrew Williams — firstname.lastname@example.org