One of the headlines that caught my eye a few weeks ago was from the Smithsonian: “Artificial Intelligence Can Now Decipher Medieval Graffiti (Cat Sketches and All)”.
A nerd and aspiring tech-lover, my interest was immediately piqued. I opened the article in a new tab in my browser where it waited alongside other tech-y and culture-y content for me to read it. Sadly, I never did find the time, and the article got flushed a few days later when I cleaned out browser clutter.
It didn’t think of it again until earlier this week, when I came across a Buzzfeed article titled “51 Hilarious Art Memes”. This one I did read immediately (we’ll cover what that implies about my intellectual sophistication later) along with two of the related articles listed at the bottom of the page (“Art History Snapchats Are The Best Snapchats” and “44 Medieval Beasts That Cannot Even Handle It Right Now”). I was in the middle of “23 Ugly Medieval Cat Paintings That Will Speak To Your Soul” when the Smithsonian article popped into my head.
So I pulled up the article in a new tab (thank god for browser history) before returning to my ugly medieval cat paintings. Once again, I was entertained by the combination of the medieval cats that ranged from cute to bizarre and the Buzzfeed-provided captions that accompanied each one. The captions were mostly just creative ways of calling the cats ugly or expressing surprise at the cats’ ugliness, but some interpreted the paintings for me too. One cat was apparently staring into the abyss looking for meaning in its existence while another was uncomfortable in its shell of a body. Overall, the humor came less from the ugly cats themselves and more from the way that Buzzfeed had pulled them out of the context of Art History to roast them through a modern lens. Its a pretty ridiculous way to look at art, but it’s also infinitely entertaining.
When I finally shifted my focus from ancient cat paintings to ancient cat graffiti, I found a very different approach to the study medieval art. Researchers from the National Technical University of Ukraine and Huizhou University’s School of Information Science and Technology have apparently developed a machine learning model that “detects, isolates and classifies” letters from about 300 pieces of medieval graffiti left on the outer-walls of a cathedral in Kiev, Ukraine.
What exactly is medieval graffiti, you ask?
Well, people started carving messages into the walls of the cathedral as soon as it was built in the 11th century. They wrote out prayers about love, complained about their shitty neighbors and, you guessed it, drew cats (sadly, I couldn’t find pictures of these particular medieval kitties). The cool thing about the markings is that they reflect popular culture, and they give us a lot of insight into how normal people lived in medieval Ukraine. Research on them has already suggested that lower class residents of the city were more literate than previously believed.
The study, then, wants to classify each carving so they can be studied more thoroughly. The technology behind it is really interesting since natural language processing and image recognition are hard things to get right. Human communication is subjective and machines by nature don’t do well with subjectivity. Add in terrible graffiti handwriting, layers of markings from different time periods, and a mixture of languages and you have a computational nightmare.
That’s where machine learning comes in. The researchers created a neural network, an AI tool that can be trained to classify complex images, and trained it against a database of Glagolitic and Cyrillic glyphs and an image recognition dataset of publicly available fonts. Moving forward, they want to improve the model’s understanding of things like date, authorship, meaning, and genuineness of the graffiti.
The thing that struck me about the study was the idea that machine learning could help us decipher the past. It’s a pretty exciting concept. AI has already been used to help decode other ancient texts, and this project builds on that field by moving into the realm of popular culture. In the world of Siri and Amazon Echo, it’s good to know that beyond imitating us, AI can apply a systematic approach to helping us better understand ourselves.
It’s that idea, the idea of using a modern technology to interpret the past, that brought me back to the art memes. They don’t have anything to do with AI or any other field in computer science, but they’re posted on the internet and do often showcase medieval art. Like the graffiti, memes are also heavily rooted in popular culture. They’re free to distribute, and they get passed around the internet to become a sort of digital graffiti themselves. We love them because they’re weird and relatable, but it’s also really cool to see old forms of artistic expression incorporated into this new form of expression that’s risen in the digital age.
The medieval cat paintings, and the rest of the works I came across in the four separate lists of art memes I looked at, weren’t posted for their composition. They were posted because someone saw them and found something in them that they could relate to their daily experience. Attributing an existential crisis to an ugly cat painted in the middle ages becomes a weird sort of ode to the way we see our ancestors reflected in ourselves. Memes like these point out the common expressions and body language that prove that we’ve always been just as petty and irrational as we are today.
The study, then, follows this same vein. In a much more scientific way, it takes a mode of ancient expression and examines it through a modern lens. What it reveals is “a community’s daydreams and worries”, as the article puts it. It’s writing about love and feuds and drawings of cats.
The bottom line is that we relate to our past through a sense of common human experience. Whether through machine learning models or art recycled across the internet, we seem to like to use technology to reinforce all the ways are ancestors were relatable. It’s how we preserve them, and it’s one way that their legacy lives on.