ART + AI: CUEVA GALLERY
Turquoise, from the ‘Ixchel’ collection by diavlex
Artificial Intelligence in the realm of Art may be free from apocalyptic visions of sentient machines subjugating and enslaving humanity, but it is no less devoid of the existential ambiguities that seem to arise everywhere in the wake of this technology. If you haven’t got past ‘what is art?’ yet, buckle your seatbelt …
AI is only slightly easier to define. Start describing what this umbrella term encompasses and it's easy to disappear down the rabbit hole into a fascinating world of semantic segmentations, backpropagation and neural networks. But this is not simply a jargon-laden branch of computer science, it is a far-reaching technology underpinned by some serious and beautiful mathematics (think partial derivatives and multivariate calculus) and it has come a long way in the sixty-five years since the term was coined.
Marcel Duchamp once said: ‘not all artists are chess players, but all chess players are artists' (D'Harnoncourt and McShine, 1989). I wonder, if he had lived to see IBM’s Deep Blue beat Gary Kasparov in 1997, whether he would have conceded that a machine could be an artist? Twenty years on the question remains unanswered and no less contentious: roboticist Hod Lipson reported near-violent scenes erupting at his suggestion of just that notion to an MIT audience containing art students (Lipson, 2016).
Artist Mario Klingemann sees the AI he uses as no more an ‘artist’ than one might think of a piano as an artist(Sotheby's, 2019). But there are myriad ways AI can be incorporated into the process of art creation, which leaves ajar this idea of AI as creator or apparatus.
It is a difficult question because artificial intelligence, specifically Machine Learning (ML) is no longer merely an increasing power of analytical decision-making. In the era of cognitive computing, machine learning and the fully connected deep networks that improve experientially are vastly changing not just 'the' game but almost every game and the art world is no exception. It wasn't long ago that you would be scoffed at for asking if a machine could produce 'art'. Now with a combination of robotics, webcams, convolutional neural networks (CNN) and stochastic algorithms, a machine can create the face of someone who doesn’t exist and render it in paint with greater accomplishment than many people could. The resulting painting would be unprogrammable and unique, not dissimilar to the human creative process. These days the questions are whether a machine can be truly creative; if there exists imaginative capability; and whether it and its output have individuality, identity.
Chucho, from the ‘Dia de los Muertos’ collection by diavlex, 2019
We might find these questions resonate in our technological age where the great concern is loss of individual identity and, certainly, they are as pertinent to the art as to our understanding of our relationship with technology. Whatever answers you arrive at, perhaps the point is that it is not clear cut. The rapidly diminishing gulf between technological and human capability seems to reduce us to semantic distinction to defend our ownership of such essentially human traits as ‘creativity,’ ‘learning,’ and ‘artistry.’
Machine learning has come an incredible distance in a short space of time and its extensive applications span computer technology, farming, the automotive industry, education, healthcare, right down to mobile apps embellishing your latest upload with style transfer filters. The only real question is where the technology goes from here. In the realm of creative arts, I am given to think of the conviction during its popularization in the early 20th century that photography could never be 'art'.
AI art may develop into an artistic subgenre, may attract its own new breed of art buyer, or develop a cult following. One thing is for sure, the technology is not slowing down to ponder its own future. It continues apace with no signs of relenting and mercifully, so far, without signs of (art) world domination.
Beth Jochim is the Director and a co-founder of Cueva Gallery based in Dublin. Along with co-founders Dr Claudia Orellana-Rodriguez, Dr Ernesto Diaz-Aviles, both of whom have Creative AI and Machine Learning expertise, Jochim brings a background in the Humanities and professional experience in project management, acting as a bridge between the technical and the artistic communities. They are a diverse team passionate about Artificial Intelligence and its impact on the art world and society. We spoke to Jochim to discuss the ways in which AI may enhance and disrupt human creativity.
How did Cueva Gallery come about?
The gallery is an initiative of Libre AI (www.libreai.com), an AI company whose mission is to democratize the advances in Artificial Intelligence (AI) and Machine Learning (ML) and make them accessible to broader audiences. Through our work in Libre AI, we noticed that there is a lack of communication between AI experts and people in creative fields such as Art. Setting up Cueva Gallery, an online gallery that presents unique works generated or inspired by Artificial Intelligence (AI), is our contribution to bridging this gap. At Cueva, we provide a space for new ideas, where seasoned artists can gather to reach new markets, and where non-tech-savvy artists can take their first steps into the world of AI. Cueva, which is the first gallery of its kind in Ireland, also aims to implement cultural and commercial partnerships at local and international levels to promote its disruptive potential.
Big Flock of Electric Sheep, from the ‘Dreaming of Electric Sheep’ collection by diavlex
How does the gallery promote and develop the use of AI in Art?
Technology is part of the life of each of us at a certain level, whether we like it or not. It modifies our behaviour and way of thinking, experiencing and interacting. Our society has turned digital at high speed, so we could say that nothing better than generative Art captures this transitional period. We should remember we are in a favourable situation to make the public understand a genre that is a child of its time, and the gallery embraces this challenge on three different levels: with artists, private collectors, and through service to businesses. Cueva works with both tech-savvy and traditional artists in ways that aim to open doors to new markets and creative augmentation. AI and ML allow artists to explore original compositions, colours, and styles that offer new sources of inspiration and learning. This can enhance the creative process as a whole and help new AI algorithms to become better innovative tools and partners.
Cueva also works to bring private collectors into the process of co-creating limited editions, one of a kind, and bespoke artworks. The difference with traditional commissioned work is the fact that Artificial Intelligence shows itself to be versatile, highly creative and contains infinite possibilities of realization that can be adjusted according to specific needs with a quick turnaround.
Finally, the gallery provides services to businesses, such as property developers, interior designers, hotel owners, and corporate buyers who want a unique piece that represents their style and business identity. We offer also, and this is not a negligible aspect, a certificate of Authenticity in Blockchain for each artwork.
Do we need AI in Art?
I would not use the word "need", but I would prefer to speak of opportunity. AI in Art has a high potential for pushing the boundaries of creativity and reflecting the interaction between human and machine. Technical people appreciate it because they understand the challenges of making it and see the beauty in maths and algorithms, other audiences may find it stimulates a new way of thinking and relating to ourselves. AI art also shows modern aesthetics that can be uncanny and challenging, or more conventional and traditional. It is up to the artist (and the viewer) to tell a story and to perceive through the artwork.
In an age of blurring boundaries between the physical and the digital I think people are becoming less interested in the medium and tend to focus more on the emotions that artworks communicate to them. So not engaging in AI art could be a lost opportunity, as it can tell new stories and offer modern aesthetics.
Can AI enhance human creativity?
I See You, from the ‘Residual’ collection by diavlex, 2019
This is an interesting question and an ongoing discussion in the community. I would like to answer shortly and say yes, it does. At Cueva, over the summer, we conducted an experiment with a traditional artist who works on ceramic and porcelain. We asked Mas, the artist, to paint inspired by the AI pieces of the collection "Residual" (https://cuevagallery.com/collections/residual-collection-by-diavlex) that we were going to launch. The results have skirted the edges of collaboration and confrontation. You can read the full article on our blog on our website (https://cuevagallery.com/blogs/explorations-of-ai-art/when-a-traditional-artist-meets-ai-for-the-first-time-an-experiment). Although at the beginning Mas was disoriented because she did not understand how to use the AI-produced images, in the end she found her own way to incorporate them as an inspiration. Her work has shifted from copying to metabolizing an almost abstract image and returning it with a twist in palette and composition. This experiment has been particularly interesting because it shows AI Art need not be hallucinogenic and distorted nor rely heavily on technological aspects (such as programming). We have also seen that non-tech-savvy artists can experiment with it at different levels.
Can Traditional Art be useful in producing Generative Art?
Yes, I think it is useful for several reasons. The most important being that traditional Art can always be a source of inspiration. I do not believe that there is contraposition between tradition and technology, but a fluid path where the two parts can find new meaningful points of discussion. For example, AI artists like Robbie Barrat (https://twitter.com/VIDEODROME) are devoted to coming up with something never seen before. Still, they recognize the influence of classical painters over the work of their fellow colleagues and the importance of traditional skills.
Does the growing presence of AI in art threaten the continued importance of the human creative?
I think there is a genuine and general fear that at a certain point AI could take over humanity. This has been endorsed by writers on AI that offer a dystopian view of the future, and this is perhaps why we should talk of Augmented Intelligence. Following an exhilarating essay of British Computer Scientist Simon Colton (http://eecs.qmul.ac.uk/profiles/coltonsimon.html), "From Computational Creativity to Creative AI and Back Again", in the last decade, the discussion has shifted from whether the software can be independently creative to whether it should be allowed to be. On the one hand there is an active community, called Creative AI, that embraces AI/ML as a tool or work collaborator but that refuses the idea of software being independently creative. On the other hand, there is the research field of Computational Creativity that supports software to be, one day, individually creative. My answer to the question is that we, as humans, will keep our importance as a whole and find new ways of existing side by side with machines.
I think AI/ML and art have a symbiotic and mutualistic relationship. Art and technology can positively have an influence on each other, bringing a more mature development of both parts. From one side, we do have a technology that is programmed by us, supposedly based on our needs. From the other, we have two kinds of Art: one that has become part of our cultural heritage and the other that is in fieri. The common point is that the human component is always present and reflects into the technology that acts as a mirror. It is up to us if the reflected image is something we like or not.
Happyness, from the ‘Ixchel’ collection by diavlex
How does AI work in the realm of Art?
AI is shaking the art world in different ways, involving several professionals working on several aspects such as authenticity, authorship, legality and validity. From an artistic point of view, AI is bringing together two essential elements: the infinite possibilities of image generation (and consequently the stimulation of imagination and creative augmentation), and the chance of getting a machine viewpoint of artwork. This happens using Generative Adversarial Networks (GANs) where two neural networks contest each other to generate something that looks authentic to the human eye: the so-called generatorcreates new images by trying to fool the discriminator into thinking the generated images are real. The ability of AI to reach novelty and results that a human could not (for example, in the latent space that emerges after the training of the GAN there is the realm of possible images with which to feed the generator. It is not possible to program the GAN to paint a specific painting. Still, it is possible to adjust the dimensions in the latent space to reach desired results and this has generated a lively discussion about who the artist is and what creativity means. This ongoing discussion affects the concept of creativity and pushes further the one about the relationship between human and machine.
What considerations are there, ethical or otherwise, regarding the curation of datasets from which AI/ML art is produced?
AI ethics has become a hot topic and involves all the fields touched by this technology. The dataset used to train the machine is made by humans and can contain biases. This is a very delicate point to consider before starting work as the labelling of the data could be misleading. An exciting exhibition about this topic has been organized by Fondazione Prada (http://www.fondazioneprada.org/visit/milano-osservatorio) in Milan called "Training Humans". It looks at the evolution of datasets over the life of AI with some sinister overtones relating to the politics of state surveillance, the harvest of our personal data and the codification of people. Lately, the interest of researchers such as Joy Buolamwini (https://www.media.mit.edu/people/joyab/overview/), has shifted from discovering the biases contained in the training datasets and eliminating them from AI systems, to aiming to avoid them in the first place. AI is not good or bad per se, but it is fed with data that can potentially contain any sort of bias. This is another reason AI art can be exciting to explore: it is a mirror of our times.
References:
D'Harnoncourt, A. and McShine, K., 1989. Marcel Duchamp. Munich: Prestel.
Lipson, H., 2016. Hod Lipson “Can A Robot Turn A Canvas Into A Masterpiece?” Talk June 1 2016 San Francisco. [video] Available at: <http://https:Hod//www.youtube.com/watch?v=YKDr6CQsb-0> [Accessed 25 May 2020].
Sotheby's, 2019. Artificial Intelligence And The Art Of Mario Klingemann. [video] Available at: <https://www.sothebys.com/en/articles/artificial-intelligence-and-the-art-of-mario-klingemann> [Accessed 25 May 2020].
All images used with kind permission by diavlex.
All image rights reserved by diavlex.