As I walked to my art history class, a poster hanging in the Diag read, “What will you do when machines do everything?” Carrying in my mind all of the impressionist paintings I had to memorize for this exam, I thought surely the visual arts could never be done by a computer. The level of emotion and resolve that goes into a work of art seems incompatible with the idea of computing. Computers are best at doing rote tasks over and over again, freeing up people to leverage the creative ingenuity of the human mind. I thought there would be no way for a computer to develop its own works of art, since it required a level of thought and deliberation unknown to an everyday laptop computer.
Or so I thought. According to an article I read in The Atlantic recently, artificial intelligence is now capable of developing its own works of art after being fed thousands of images. These machines are able to detect patterns in the works and make a similar piece, albeit a blurry mass of color attempting to resemble a Renaissance portrait. That emotion and intelligence I believed exclusive to humans may no longer be ours exclusively, as computers are now capable of taking in information in a way that resembles human thinking. This may allow them to eventually develop works on their own and possibly become the new arbiters of art taste. If art, the beacon of creativity, could still fall prey to the unrelenting monster of computation, I feared nothing would be safe in its path.
But the knowledge I gained from taking that art class on French modernism in the 19th century stopped me from immediately deploring this new development. My professor reminded us all of the time that many of the artworks we studied took their inspiration from paintings that came before them. In fact, some painters, like Manet, literally copied elements from old paintings and pasted them into his own, something a computer could easily do.
For example, Manet, an admirer of the Spanish masters such as Velázquez and Goya, literally moved the crowd from one of Goya’s bullfighting scenes to his own Execution of Maximilien. But, instead of the crowd peering over the ring, they were peering over a wall while they watched the soldiers shoot the prisoners. The presence of the faceless soldiers standing in lockstep, away from the viewer, is essentially taken from Goya’s The Third of May 1808.
Many of the paintings we studied in class either copied or transposed elements from older paintings and made them into something of their own, but with a new style reflecting their own period. Put this way, it does not sound at all far-fetched that a computer could eventually do the same, given it can already recognize patterns and understand which paintings belong to which period of art.
Creating a new style from old elements, however, gives me pause as to how successful artificial intelligence can be in replicating new art. While it might be able to see trends in data, I don’t think AI will be able to predict the next trend in art. For example, I don’t see the technology eventually gaining enough human foresight to make a leap equivalent to the shift from realism to impressionism in the 19th century. Nor will I think a computer will be able to know which elements it should take from another painting in devising its own creation.
Manet, in painting his work Olympia, put his model on the same bed with the same shade of red as Titian’s Venus of Urbino, right down to the same untucked sheet in the lower left-hand corner. How would a computer know that it wants to put its art on the same plane as the renowned Renaissance painters and so remind the viewer of a painting from that period? If it somehow was sensitive to the opinions of its human critics, how would it know which elements to include and in what form so as to make that connection in the mind of the viewer? Given these difficulties, I don’t believe computers could make the same works of art that measure up to the standards of humans. So, rest assured, I don’t think machines are coming for the painters anytime soon.
But what if there is a third path — neither a complete replacement of artists by computers, or a Luddite-like destruction of technology that is capable of making art? What if art could instead be used to enhance the creative process, making the idea of looking to the past for inspiration a lot easier with the power of a computer?
A few days ago, my sister sent me a link for an art exhibition she was seeing titled “Machine Hallucination”, which uses machine learning to process thousands of images of New York City architecture and thereby make connections between different structures made in different time periods. Instead of destroying art, computers could instead enhance how we view art and allow us to make insights we could not make without it.
For me, someone who is looking to embrace both a creative side of writing and a technical field, this sounds like the perfect harmony. I hope, like many other tasks throughout human history, machines make the process of creating new content — such as these articles — easier and, possibly, even better.