Artificial Intelligence has progressed exponentially since its inception in 1956, from systems that can play chess to complex algorithms which can now create expensive art. Obvious’ The Portrait of Edmund Belamy is the first AI-generated piece of art that was sold at a major auction house in the fall of 2018, establishing Obvious as a major leader in the community.
While the first estimate that Christie’s put on the painting was in the range of $7000 – $10,000, it was sold for a whopping amount of $432,500, more than 40 times the former. This sale led to the creation of a unique marketplace altogether, and now we have platforms like aiartists.org and artaigallery.com established just for AI artists.
How was it created?
The algorithm used to build this AI masterpiece is called a Generative Adversarial Network (GAN). First developed by Ian Goodfellow, the concept behind this technique started with a simple idea: “what if two neural networks are pitted against each other?”.
One network (called the generator) is trained to look for patterns in a pre-determined dataset, and then use those patterns to create replicas. At the same time, another network (also known as discriminator) assesses these replicas and compares them to the originals in the dataset.
If the discriminator is successful at finding differences between the replicas and the originals, the replicas are sent back to be tweaked. This process is repeated the generator is able to create replicas that have minimal differences from the originals. The Verge’s Vincent explains it best as a bouncer at a club, not allowing a drunk inside till they act sober enough.
The GAN used by Obvious works in a similar manner. They’ve used a dataset of 15,000 paintings created between the 14th and 20th Centuries to train both the generator and the discriminator. The result is a set of 11 portraits of the fictional Belamy family (a subtle tribute to Ian Goodfellow, “bel ami” is French for “Good Friend”), one of which is “The Portrait of Edmund Belamy”.
Since then, the community of AI artists has grown exponentially, establishing not just separate auctions and platforms, but also unique techniques and courses available for anyone with a keen interest to join the community.
What does this mean for the future of art?
Art is a complex amalgamation of inspiration, emotion, and experiences that are unique to human existence. This is what gives art its intangible value. More importantly, art is what helps individuals express themselves without the fear of judgment. It is a result of human suffering, curiosity, affection and hence fundamental to our existence.
A very intelligent AI could possibly have an advantage in its objective evaluation of art. However, the experiences, self-doubt, and sentiments that compel an artist to work on a piece is hard for even the most intelligent AI to replicate.
At the birth of any new creation or idea, there exists a subjectivity bias of the creator. Even the creation of objectively true subjects like physics at its core has human curiosity and a thirst for answers driving it. Therefore, although the creation of AI, or the choice of art pieces, has a sense of subjectivity, the AI itself exists and works objectively. As long as the subjective curious human nature is driving AI, it has the potential to exist as a collaborator instead of a competitor.
We are now progressing into a time where such intelligent systems will become the norm. A new generation that aims at providing people with the free time they need to ponder over ideas and ways of self-expression by leaving the monotonous and mechanical jobs to intelligent machines. In this world, the biggest challenge is not creating an intelligent program built to replicate human consciousness. The biggest challenge is protecting a very essential human purpose.
Leave a Reply