A/D/O by MINI | AI Generates New Era for Art

Journal

Graphic/Visual

AI Generates New Era for Art

Are computers the next Impressionists? Artificial intelligence has infiltrated the art world, but the rise of “GAN-ism” raises questions about authorship and accessibility to the tech.

Art and technology have a long and tangled history. In the 16th century, the camera obscura introduced perspective to European painting, while the invention of the humble paint tube in the 1840s allowed artists to go en plein air and explore landscapes in new ways. Much hand-wringing accompanied the birth of photography in the 19th century, with commentators fearing it would lead to the debasement of visual art, but what followed was in fact an unprecedented flourishing of creativity as artists were freed from the tyranny of representation and delved deeper into the unique messiness of their mediums.

Taking the long view, artificial intelligence can be placed in an endless march of technological innovations in art. One key difference, of course, is that no one was worried the camera would one day enslave humanity to its mechanical will. Artificial intelligence (AI) is a uniquely controversial form of tech because it presents an alternative form of intelligence to our own.

The Lumen Prize honors artworks created by AI, including Lovebirds, Nightbirds, Devil Birds.

This has been portrayed as an existential threat in fiction – from The Matrix to Terminator and 2001: A Space Odyssey – but the very real risks have also been highlighted by academics and technologists the world over. The physicist Stephen Hawkins famously pronounced that the development of full AI could “spell the end of the human race,” whilst arch-technologist Elon Musk has warned that we’re “summoning the demon” by developing sentient technology. 

Demons aside, the current advances in AI will undoubtedly transform our lives in the very near future, impacting everything from how we shop to how we date and work (forecasts show that anything up to 40% of jobs will be replaced by AI in the next couple of decades). The question now is whether art should remain uniquely insulated from AI as it cuts swathes across the rest of our culture.

Other Lumen Prize winners include Melting Memories (this image) and Drawing Operations (main image).

“Art needs to reflect society and artists have always been curious people who want to cross-boundaries and breakdown borders,” explained Carla Rapoport founder of Lumen Art Projects, the organization behind the specialist art-tech Lumen Prize. “In a world that is driven by digital technology – and increasingly by AI – it’s essential that art explores the implications and possibilities here.”

When the Lumen Prize 2019 took place at London's Barbican Center in October, AI-inflected work was represented strongly across all categories, despite having a specialist category of its own. “When we launched the AI Award three years ago we really had to explain ourselves, but this year I’d say at least a third of the entries – from stills to moving image and interactive – use AI in some way,” Rapoport said. “The flourishing we’ve seen in recent months has taken everyone by surprise.”

Mario Klingemann's The Butcher's Son was created using a Generative Adversarial Network (GAN).

One person who has played a formative role in this flourishing is Mario Klingemann, last year’s Lumen Gold Award winner who has been working with AI for over a decade. His winning work The Butcher’s Son has drawn obvious comparisons to Francis Bacon’s distorted figures, the dominant aesthetic that comes from using a type of AI called Generative Adversarial Networks (GANs).

Without getting too mired in the technical details, GANs are fast becoming the darlings of creative technologists, allowing a remarkable level of collaboration through “training” with reams of imagery. GANs then interpret this imagery, creating an eerie, dreamlike (or nightmarish) translation that is both alike and uncannily different to the source material.

Klingemann trains GANs using existing imagery, to create pieces like this called Careless Whisper.

“Artists have always been searching for some external impulse, something unexpected, something not entirely controlled by the conscious mind,” said Klingemann. “Working with GANs is a way to achieve this state; it reveals another way of seeing.” For Klingemann, this does not diminish the role of the human artist (whose intentionality is still felt in the GAN training and in the assessment of what works have been successful) but just extends the creative act to a network of human and non-human agents. 

It’s this extended definition of the creative process that partly explains the controversy surrounding AI-art, however. When Christie’s auctioned off a work by Paris-based collective Obvious in October 2018 for $432,500, it was signed by the algorithm used to create it rather than by a named human artist. What’s more, this algorithm wasn’t even created by Obvious but was adapted from an existing code sourced online, raising questions about authorship that the industry is still puzzling out today.

Klingemann's AI work has been likened to the distorted portraits of artist Francis Bacon.

This partly explains the prickliness of art world gatekeepers to AI-art, but there seems to be a deeper concern about somehow diminishing the creative act, as if art and science/technology are opposing spheres that should not mix. “It’s not supposed to be the case,” said professor Ahmed Elgammal, director of the Art & Artificial Intelligence Lab at Rutgers University.

“If you look at the likes of Da Vinci or Galileo, there was no barrier between disciplines for them. They were great artists, scientists, writers, philosophers – the whole spectrum of human experience. There has been a shift since where art has somehow become seen as subjective experience, and science as somehow objective and in opposition to art. This does not represent reality.”

Because AI still requires speciality knowledge, most artists working in the area today could perhaps be better described as creative technologists and a vast majority of the art world remains locked out. This is changing however, with intuitive platforms such as RunwayML and Processing.org democratizing access. Elgammal and his team have recently launched Playform.io to allow artists to train their own GAN from scratch using an intuitive interface, no coding required.

Artist Yinka Ilori also experimented with AI to create works for Frieze London 2019.

“I have to confess I’d never even heard of AI before starting on this project,” said artist and designer Yinka Ilori, who was commissioned by Bombay Sapphire to create a series of AI works for Frieze London 2019. “I was actually very skeptical, because my work is so much about my heritage, and traditional craft and memory. It’s very personal and I couldn’t see how AI would fit in here.”

Meeting with the Old Street-based studio Happy Finish to better understand the AI aspect, Ilori soon began to see the potential however. “I explore ideas of memory in my work, and AI works in the same way: you can feed in imagery and it starts to create histories of its own. The final works feel very recognizably my style because of the source material, but the AI adds an extra layer that gives the viewer something to decode.”

As a new digital-native generation of artists, curators and collectors come of age it seems inevitable that we’ll be seeing more AI work on our gallery walls. “The establishment will die at some point,” concluded Klingemann, “the traditional gatekeepers just fade away, like an amorphous block that shifts and absorbs a different way of thinking.” What exactly this new way of thinking will be is still up for debate, but it’s success will hinge on as wide a range of creative talent as possible getting their hands on these fast-evolving tools.

A global community of creators empowered by MINI to boldly explore the future of design.