One of the most divisive debates in tech in the last few years is whether AI ought to be considered as a person or not. On the face of it, it might seem absurd. Only people can be considered people, right? AI is merely a tool we’ve engineered to make our lives easier—albeit an incredibly complex one—just like any other techno- logical advancement. So why is this even a question people are asking? Perhaps the answer lies in the relationship between art and AI, what we can learn about the machines we ourselves create. Perhaps—and this is even more interesting—it can tell us about the nature of creativity itself.

 

AI as the Artist’s Tool

Whether you look at art as an industry or a creative outlet, the medium has become utterly transformed over the years as technology has gotten more advanced. After computers went mainstream, the creative landscape changed entirely, with digital art opening up a plethora of new options, techniques and styles for artists. Beyond even this, the internet has spurred an absolute explosion of artworks that are made, shared and consumed by billions of people around the world.

As technology grew more complex, so did the art we created with it. Our inter- actions with software and hardware became more abstract, with people taking a more supervisory role and the machine performing the grunt work. There’s no better example for this than animation and special effects. Back when animators had to hand-draw every frame, the entire process of creating a cartoon or fantastical VFX shots was labour-intensive, limited by the physical tools we had, and narrowed the scope of what kind of stories they could be used to tell.

 

Once computers were powerful enough, we could do all kinds of things with them, like creating models for characters, rigging them for movement, and having the software ‘fill in’ the motions between different poses. Or having the software simulate lighting, texture or physics through an algorithm. It still is incredibly painstaking work, but it means less time for the artists to waste on the mechanical aspects of the task. It also dramatically changed how we created new art — and consequently, our capacity for bigger, better storytelling. Computer animation has even been making use of AI for years now, with programs used to simulate behaviour patterns for massive armies during big battle scenes in the Lord of the Rings or Marvel films.

 

AI is incredibly adept at tasks that involve collating and processing large amounts of data and outputting the—sometimes unpredictable—results. Many engineers today don’t even understand exactly how some AI come up with the results that they do.

Merging Contemporary Art and AI Tech

It’s this unpredictability that inspired the art of Refik Anadol, a Turkish media artist, who specialises in creating large, immersive art installations. Over the last decade, he’s been using complex AI programs to generate evocative digital works of abstract art on a massive scale. In 2018, collaborated with neuroscientists at the University of California, San Francisco, to create an art project unlike anything that had been created before. He asked volunteers to think of their most cherished childhood memories, perhaps “the most precious and private information we hold as humanity.” This neural activity he captured in the form of EEG scans, which he fed into a series of machine learning algorithms.

 

Anadol was inspired to create this work when he discovered his uncle could no longer recognise him due to Alzheimer’s disease. In an effort to visually recreate the activity of a brain recalling cherished memories, he created a series of artworks titled Melting Memories. Unveiled at the Pilevneli Gallery in Istanbul, Turkey, his ‘data paintings’ were shown off on massive 65-inch OLED displays. Slow-moving images washed across the screen like fine sand across the seabed or clouds in a weather forecast, swirling in the currents. Shapes would form and disappear, morphing into yet new shapes, like a mass of…something, melting and unmelting. It was a breathtaking synergy between contemporary art and advanced technology.

Anadol’s oeuvre consists of many such data paintings, augmented data sculptures and light projections, all of which use massive datasets of images, video and audio to create evocative works of art that resemble almost nothing a human could ordinarily make. One of his most famous pieces, depicting New York City in a surreal metamorphosis, was created using millions of images of the city found on search engine results.

 

“[The] machine looks at this information like a human being,” he said about the project, “but [it’s] more like collective memories than personal memories because a building in New York can be explored by thousands of perspectives. It’s a more honest memory [because] it feels everything and everyone than just one person.”

Like Nothing We’ve Seen Before

It’s fascinating to see how an artist is able to use AI to create visual experiences that are unlike anything seen in the history of art. The internal mechanisms of deep neural networks and machine learning algorithms are far abstracted from the processes of the human mind, which makes it infinitely more fascinating to see AI present these strange, distorted visions of reality.

But that’s the thing — we’re not entirely sure why these artworks look the way they do. When machine learning algorithms make mistakes, we have to constantly ‘coach’ them into getting the right answer, sort of like teaching a child. But just like we rarely ’teach’ a child how or what to draw—often to bizarre results—the freedom we give AI to produce art leads to some very weird, very interesting end products.

 

Hold on a second. Doesn’t that sound vaguely familiar? By not strictly imposing rules on the AI’s processing and output of data, we’re kind of letting it do its own thing. It’s almost as if we’re letting the algorithm be…creative.

Join us on the second part of this journey as we explore the realms of machine creativity and what AI could achieve if they were to be the artists themselves.