ARTificial Intelligence

Machine learning is seeping into all aspect of our lives, including art. Sonia Klug delves into the world of AI-inspired creations and wonders what the future may hold.


We are on the brink of an artificial intelligence revolution. The signs are all around us. Not only are swathes of jobs being replaced by automation, but AI is also changing how we interact with the world and perceive ourselves. Even, plastic surgeons are now increasingly asked to alter their clients’ appearances to match their (AI-‘enhanced’) smartphone selfies. 

But as we enter this brave new world, we wonder – what are the implications for the future of creativity? 

Artists are often at the coalface of societal change, expressing the emotions and struggles through their work. How do they process and react to this enormous shift? And are they willingly making themselves obsolete by developing algorithms that are capable of creating art?

Mario Klingemann, an artist, who has been experimenting with machine creativity since the 1980s, considers AI-inspired art as a new and powerful movement. 

He says: “The art world (whatever that might entail) will not be able to escape or ignore AI, as it’s one of the major forces that will shape our society. 

“Right now, we are in the typical early phase of an art movement, where parts of the ‘art world’ are inquisitive, curious and welcoming and other parts are trying to barricade the doors or practice purposeful disregard.”

Neural Glitch 2018 by Mario Klingeman
This is one of Mario’s AI-generated images. He is fascinated by the possibilities of machine ‘accidents’ in the NN he uses, one of which can be seen in this image.


While attempts to build automatons are almost as old as civilisation itself, sophisticated artificial neural networks (NNs) have only very recently been used on a large scale. These are computing systems based on the human brain which can “learn” through training sets, rather than manual input. For example, this allows photo library software to group all images of cats, cars or your nan. Machine learning is already employed in medical diagnosis, and more sinister applications include deciding if you’re worthy of credit or a job interview and how to manipulate voting.


There have been many attempts to produce programmes that are creative in their own right. For example, The Artificial Intelligence Laboratory at Rutgers (AICAN) fed 80,000 images from the last 500 years into a computer system and trained it to produce and even title new works of art. Some of these were exhibited at Art Basel, where 75 percent  of visitors judged these to have been created by a human artist. 

Even more astonishingly, the first ever computer-generated image to be sold at Christie’s fetched $432,500. A spokesperson for Christie’s said: “The exceptional price realised for Edmond de Belamy reinforces that this is a significant moment for the art world. Christie’s continually stays attuned to changes in the art market and how technology can impact the creation and consumption of art.” 

However, it’s worth keeping in mind that the value of a piece is determined by its scarcity. It’s likely that this image sold for such a high price because there can only ever be one first AI-generated image auctioned by a leading auction house.

Portrait of Edmond Belamy by Obvious
This portrait is part of a series of AI-generated images, produced by Obvious, a collective of artists and AI researchers.


The capacity of machines to conduct semi-autonomous processes fascinates Memo Akten, one of the first artists to use NNs. His Deep Meditationseries uses photos on Flickr, which were labelled with words like god, world, ritual or universe. He then instructed an algorithm to morph these pictures into each other, resulting in a surreal display of abstract and figurative images in constant change.

He said: “As we look back upon the meditative, slowly evolving images conjured up by the neural network, bordering between abstract and representational, we project ourselves back onto them, we construct their meaning, we invent stories, because we see things not as they are, but as we are.”

He considers his work to be at the intersection between human and machine creativity that produces striking images and engages with the relationship of our and the machine’s vision. 

More and more artists and group exhibitions deal with artificial intelligence. One rising star is Anna Ridler, a London artist and researcher. In her work she has, for example, programmed the value of the BitCoin to change the appearance of blossoming tulips or created a voiceover for a nature documentary that is based on an algorithm, fed by the lines of female protagonists in romance novels, talking about female desire, trauma and fantasies. 


While there is great enthusiasm for the possibilities AI is opening up – from making our lives easier and safer to providing a fertile ground for creativity – there is also a wariness of how AI is seizing more and more areas of our lives with little transparency or regulation. There is concern about how NNs reproduce human bias, such as racism and sexism and the potential for abuse of AI mass surveillance in particular

Trevor Paglen tackles these topics in his work. With a PhD in geography and a fixation on hidden locations, such as military black sites, he is fascinated with the invisible in general. In his project A Study of Invisible Images, he wants to show what invisibility, such as data sets and algorithms, looks like. “If we want to understand the invisible world of machine-machine visual culture,” Paglen writes in an essay for The New Inquiry, “we need to unlearn how to see like humans.” To do this, he has developed algorithms to understand how AI computer system, such as laser-guided missiles or self-driving cars, ‘see’ the world and replicates this vision. 

He is also trying to deconstruct the Facebook algorithm that scans images for the things we buy so they can target their advertising more effectively. Instead of focussing on the articles relevant to corporations, though, he centres on objects, significant to Freudian psychoanalysis. 

He is deeply concerned about the potential abuse of AI by powerful institutions. He tells the Observer, “We live in a political moment where it seems reason has gone out the door. And at the same time we have these incredibly predatory institutions being created, whether it is white supremacy on one hand or Facebook on the other. It is kind of a surrealist moment. Everything is like Magritte’s Ceci n’est pas une pipe. Nothing is what it seems.”


Opinion is divided if these NN processes can be called truly creative. 

Mario Klingemann believes that machines are already being creative and that one-day algorithms will be able to create more exciting work than humans. He said: “I believe that it will become more and more difficult to decide who the author or artist behind a machine created work is, once we have AIs that generate other AIs and can optimise themselves based on the feedback of successes or failures they measure – at least for creative products that might not be masterpieces but rather entertaining pop culture for everyday consumption. I think this will happen within the next 10 years.”

However, most consider NNs as a tool, only capable of reproducing images based on their training sets. It’s argued that machines are incapable of artistic intent and cannot make meaningful social commentary or surprise – two deeply human skills, which are fundamental to culture and art.

Whether you believe that machines can or can’t create art is ultimately is down to whether you think that artistic merit depends on aesthetics and/or the power to engage or the message of the artist.

It’s often said that art – from first cave paintings to conceptual installations – is a form of human communication. While algorithms may be able to make a photo look like an impressionistic painting, compose a sad melody, or produce images based on training sets, they are no more capable of artistic intent, than an interactive voice response system can make an emotional connection (or make you feel like valued as a customer).

Uncanny Mirror, commissioned by Seoul Mediacity Biennale 2018 by MarioKlingeman
This installation consists of a screen and a camera and transforms the
spectators face in real time into an uncanny representation of themselves based on all the faces that previously looked into the mirror. 


Even though there is a lot of excitement about the potential of AI to create art, history suggests a different path. 

Like machine learning, photography was a ground-breaking technological invention that heralded a new era. By making realistic representations much easier to achieve, it pushed artists towards more expressive (and later conceptual) forms of art, like expressionism and postmodernism. Over centuries, the emphasis in art shifted from technical skill towards the intent and message of the artist.

Similarly, the easier technology made the creation of perfect photos, the more people craved the authenticity of candid shots, reflected in many photographers’ work.

Time will tell if AI will put the emphasis on human creativity once more. If history is something to go by, artists will do anything in their power to stay a step ahead of machines. The current trend towards process art, which focuses on the artists’ thinking processes and uniqueness, may already be part of a backlash against the dominance of technology. In popular culture, this is mirrored by the tremendous popularity of content by social media influencers – their personality is their main asset.

Nevertheless, machines imitating our traits, such as creativity and intelligence, questions what it means to be human. We need the critical engagement of creatives more than ever to help us to make sense of it all and find a way to use AI that values our humanity.

Mario Klingeman

The artist has been experimenting with machine creativity since the 1980s and says, ‘We are fascinated by non-human intelligence or at least something that seems to behave in intelligent ways, such as videos of pets playing music, and, increasingly, AI-generated text, images or art. Mechanical processes run in cycles or patterns, but intelligent behaviour is unpredictable and surprising, which is what makes it so interesting. I find AI as a tool fascinating because it combines the ability to control the process with the possibility of creating interesting accidents. Yet the balance of those factors is in my control, which is not always the case with physical media, like brushes and paint or a musical instrument. For more information on Mario’s work, visit

Sonia Klug is a tech writer, photography fan, scout leader and news-junkie, fascinated by the intersection of technology, society and culture.