On AiArt

You are alone in a dimly lit room. A screen flickers in front of you. Then, two images appear. Relying solely on your eyes and your experiences, you must decide which one is generated by a machine and which one is made by a human. If you cannot tell one from the other, your failure proves that machines can create art.

I see you struggling out there, scratching your head, focusing on traces of reality in the blue of the sea. Just stop, there is no point in trying. The riddle was worded wrong all along, and it was not me who made it. We are in the year 1950 when an English mathematician imagines a similar scenario. There is a human interrogator in a room, facing two computer terminals. He interacts with two entities on the outside through written messages alone. After a while, he is asked to guess which one of them is human. If he fails, that implies machines can think. That man, who inadvertently set the agenda for artificial intelligence for the next decade, was Alan Turing, and by asking us to distinguish a man from a machine, he ended up labelling what it is to be human as a floating box of circuits. He was so fascinated by brains, that he forgot they were contained in bodies, and by doing so, he set the stage for the erasure of embodiment at the basis of our current definitions of intelligence, and therefore of humans. By labelling intelligence as the manipulation of informational patterns, Turing contributed to the formulation of information as a bodiless fluid, and therefore of human identity as a mere informational pattern. An idea that inspired some radical propositions, like Hans Moravec’s belief that the assumption could be demonstrated by downloading consciousness into computers, without the loss of any meaning or form.

In the years that followed, many logicians and philosophers proposed counter-arguments to Turing’s test, with the most famous being the so-called Chinese Room by John Searle. In his thought experiment, he argued that imitating intelligence would not imply possessing it, and by distinguishing mimicking from being, Searle delineated two different forms of artificial intelligence, and he called them ‘strong’ and ‘weak’. The first one implies that the machine is equal to a human mind, and the latter, that it is a mere tool in its understanding. Following Searle’s categorisation, most of the progress we have achieved since the 1950s in terms of artificial intelligence belongs to the second category, which is now labeled ‘narrow AI’, alluding to the computer’s ability to only perform one task at a time. I will argue that this same counter-claim is a helpful tool to solve my initial riddle, and therefore modern controversies on the so-called AiArt. But let’s go back a bit. What is AI after all?

Although artificial intelligence is now at the basis of most of our daily processes, attempting a comprehensive definition is a hopeless task. If we have to try, then we can describe it as an expanding scientific field aimed at the development of adaptive and autonomous machines emulating forms of human intelligence. AI is what powers self-driving cars, or your favourite social media platform; even your laundry machine at this point. It is the brick at the base of reality, the new hidden bit behind the veil of appearances, the one that no one fully understands. Within the past twenty years, it has made breakthroughs in various technologies, such as computer vision, speech recognition, and machine language processing. With the advance of AI applied to most industries, many have begun creatively exploring its possibilities by engaging with the latest techniques, such as deep learning and neural networks. In the last decade, several AI image generators, like DALL-E or Stable Diffusion, have been released; and with the advent of these open platforms, code-powered images can be generated by the click of a button, or the prompt of a sentence, by literally anyone, at any time. Here is where the controversies begin, and many ask whether we are at a point of rupture, or just witnessing the development of a brand new artistic tool.

Some argue that every time a new technology is invented, it is integrated into the art world in progressive waves, take photography as an example. First, the medium threatens old media and the artist, then it becomes the subject of the artworks, and finally, it is either integrated or discarded. When photography was born, many thought it was the end of painting, some screamed in despair and labeled it the end of art itself. And yet, we came to the conclusion it was just another tool, like pigments or the printing press. Many see even more parallels with the early days of photography in AI’s vain attempt at mimicry. They argue there is a difference in scope between human and machine creativity. Computers can create images, even in ways we don’t yet understand, but there will never be any soul in them unless there is a human-machine symbiosis. They label this a new form of ‘statistical creativity’, limited to data sets and restricted by the lack of human intent. Not a step-change into making art, but just a new technology that will bring us closer to human- machine complementarity. They say the only real impact of AI applied to the arts is making us question what creativity is after all. I argue that’s the case. And yet, others are questioning the impact of artificial intelligence on the art world and believe that it is still unclear. Although there are similarities between current ML art and the computer art of the ‘60s and ‘70s, there is something radically new in it. It is not just a new tool sparking creativity, or the latest form of a century-long trend in automation. It is instead the start of a fresh new form of art, AiArt, far superior to all that came before. Let’s now delve into a short journey into the development of this revolutionary art form.

It all began in the ‘60s and ‘70s when computer art first bloomed. Its pioneers, artists like Vera Molnar and Frieder Nake, set out to harness the potential of early computers against the very notion of control to produce unexpected results, making glitches and misunderstandings a new form of art. Then, from the 20th century to the early days of the 21st century, AiArt truly commenced. The date is May 11, 1997. A big blue box called Deep Blue beats the Russian champion, Kasparov, in a game of chess, the true symbol of human intelligence. The Computer Age has begun. New human-computer technologies bloom and some sense of a shift with the previous trends in automation. From then on, AiArt takes centre stage. In 2016 Google develops a new kind of robot, AlphaGo, that defeats the Go champion Shishi Li, showcasing the real potential of artificial intelligence in a real-life scenario. Then, in 2016, following a new trend in ‘Deep Leaning’ — a new kind of AI algorithms — Google develops a new neural network, ‘Deep Dream’, the first Generative Adversarial Network (GAN) able to mimic traditional paintings. Only one year after, the first Creative Adversarial Network (CAN) is produced, a program not only able to imitate artworks, but to create them. And here we are now, in 2022, in the age of creative machines and AI-generated images.

Now, you might be wondering, how is this supposed to convince me that a computer is as an artist as I am? Well, it is not. Yet, some may argue that not only a robot is now a creative subject, but it is also a far better one than you are. Not only it can replace your labor, but it can do so by breaking through the constraints of time and place. It may not understand what art is, and yet, here it goes and makes it. They say we are living in a whole new age of art, where creativity is in the hands of everyone, not just artists. The scientist living next door is as much of a Picasso as you are, and your computer can make a better picture than you could in half the time. They proclaim it a new age of creativity, where all will be transformed under the influence of technology and science, and new art will emerge to bridge the gaps between the fields and break the last remaining boundaries. Maybe, we are at the precipice of something new or at the end of something old. Maybe, AI is a revolutionary threat and our relationship with art will be never be the same. I doubt it.

When Turing set out to define intelligence as the manipulation of informational patterns half a century ago, he was thinking of the human brain alone. He discarded our skin and labeled our flesh a cage. I argue there is art in the movement of our cages of meat. A computer will not be able to replicate it, or even understand it. In his 2017 novel, ’To be a Machine’, the Irish writer Mark O’Connell describes the meeting of a four-year-old girl with a four-foot humanoid called Pepper. The robot is a customer service machine, designed to feel emotions by receiving data through touch sensors. Pepper is asked to hug the little girl. Pepper does not understand.“You would be surprised how difficult it is to solve the problem of hugging.” What a machine can do is what we scripted of us into it. Yet, most of what we do is out of our control. There is art in the nonsense of our instinct, in the meaningless of actions. I argue you could hardly explain a machine that art is putting a urinal into a museum. I would like to see you try it.

Reference List

Chen, Weiwen and Shidujaman, Mohammad and Xuelin, Tang. AiArt: Towards Artificial Intelligence Art. ThinkMind, MMEDIA, The Twelfth International Conference on Advances in Multimedia, Lisbon, Portugal, 2020.

Field, Sarah-Jane, The end of something… but I’m not sure if it’s art just yet. Wordpress, 2022. Retrieved on 23rd November 2022 <https://sarahjanefieldblog.wordpress.com/2022/10/03/the-end-of-something-but-im- not-sure-its-art-just-yet/?like_comment=697&_wpnonce=d1335dfcbe>

Harrison, Anya. All systems go. Flash Art, 2017. Retrieved 25 November 2022 <https://flash—art.com/ article/lawrence-lek/ >

Hayles, Katherine N. How we become Post-human. The university of Chicago Press, 1999.
O’Connell, Mark. To be a machine: Adventures among cyborgs, utopians, hackers, and the futurists solving the modest problem of death. Anchor, 2018.

Ploin, A., Eynon, R., Hjorth I. & Osborne, M.A. AI and the Arts: How Machine Learning is Changing Artistic Work. Report from the Creative Algorithmic Intelligence Research Project, Oxford Internet Institute, University of Oxford, UK, 2022.

Using Format