Working at the intersection of scientific research and craftsmanship, aurèce vettier is an artistic collective founded by Paul Mouginot and Anis Gandoura in 2019 which aims at understanding how relevant and meaningful interactions with machines and algorithms can be achieved, in order to push the boundaries of creative processes. Experimentation is crucial in their practice; their works are often the results of a creative and rigorous approach at selecting, remixing and generating new forms. Looking at their works, the viewer can’t help but recall the precision and imagination of medieval herbalists who catalogued plants and flowers in manuscripts that were hand copied across the known world at the time.
aurèce vettier is composed by Anis Gandoura and Paul Mouginot, both engineers whose expertise in AI technology provide a unique perspective on the limits and perks of working with AI to make art. They both live and work in Paris, France.
Filippo Lorenzin: Paul, it is rather unique to find collectives of creatives working together nowadays, especially in the field of so-called “new media art”. I wonder if you could tell me about how you started your collaboration and what it means for each of you to work as a collective; for example, how does it shape your creative process?
Paul Mouginot: aurèce vettier is indeed an artistic collective I founded in 2019 with Anis Gandoura. The format of a collective helps us to invite experts and craftsmen depending on the complexity of the project. In an age where collaboration is more and more crucial, we see it as an organic experiment. We’ve known each other for a long time and followed each other in our early professional adventures. In 2016, we co-founded an AI company deploying image recognition technology in the fashion sector. By training, tweaking and implementing many AI models, we started collaborating with a few experts and creatives in this field at that time.
Filippo Lorenzin: Your approach to the development and making of artworks reminds me of how art workshops worked for centuries; under a single name one could find many experts in their own fields who worked together on many various projects.
Paul Mouginot: Yes indeed, this is what we are trying to reach at aurèce vettier. It’s not a studio where as an artistic director, I decide everything. It is rather an experimental group. For our upcoming exhibition in Paris, to be held in December 2021, we invited a cutting edge sound artist, Varhat, to create an interactive generative installation with us. Our poetry book was designed by a super talented graphist called Damas Froissart. And in the exhibition, we are collaborating with art historians such as Zsofi Valy-Nagy, who is currently writing her PhD thesis about the work of Vera Molnar. It’s smoother for me to work with such diverse and talented artists by creating a workshop-like atmosphere. It helps create well-balanced collaborations. Thus, deeper ideas emerge.
Filippo Lorenzin: Given the importance that the crypto art market puts on authoriality and uniqueness, I’d like to know what you think about the role played by cult personalities in digital art.
Paul Mouginot: Are authoriality and uniqueness really specific values of the crypto-art market? I’d say that due to decentralization, the very concept behind crypto-art, three key phenomena emerge. First, the effect of global attention scarcity. Since the barriers to entry are getting super low, a significant number of players are arriving in the space -artists, collectors, curators. However, you can only spend a limited amount of time on Twitter or reading art publications, to listen and discover new artists – even if it’s 24h a day: the global amount of attention is limited. So within this network, there is a stronger share of attention naturally given to the “pioneers”, the prominent collectors, the stars arriving in space or the people with exceptional stories. We’d call them the “nodes of the network”, and we believe their influencing power will grow in the future, reproducing the mechanics of the traditional art world. Given the amount of people involved, we actually believe it might be more difficult in the future to “make it” into the crypto art space than in the real one.
Secondly, the power of network discovery. Most of the current superstars of the NFT space were almost unknown a year ago. Thanks to the network effect, they were discovered, promoted and their career is skyrocketing faster than many artists in history. Some of them are now called “OGs” or pioneers and have garnered a lot of attention. The auction houses contribute to the starification, as was the case a few years ago with the emergence of “AI art”. It’s wonderful to see how crypto art has changed the life of many overlooked creatives and created a positive dialog about environmental impact or the representation of minorities within the art space. But algorithms, AI, and NFTs are tools, or at best, media. The pioneers in photography such as Nicéphore Niépce or William Henry Fox Talbot did not make the best photographs even though they made history. So we believe that in this field, the best is yet to come, the medium will probably be investigated and questioned more, this is just the beginning. Lastly, the need for auto-promotion. Without any intermediaries, artists have to promote their own work, one way or another to achieve sales. This is the part we are the most uncomfortable with. As far as we are concerned, we can have many roles or interests, but it stops at being a merchant. Not because of the monetary aspect of a sale, but because mediation is crucial in art – to give context, to tell stories, and to find the right work. Whatever our present or future ratings will be, we want to be detached from them. We hope that initiatives like DiverseNFTArtists will flourish more in the future, to give more curatorial angles to the space.
Filippo Lorenzin: How do you see the role played by AI in your practice? Is it like a a sophisticated but ultimately “dumb” artificial assistant helping you to achieve what you wouldn’t be able to do by yourselves or it is more like the third active member of aurèce vettier?
Paul Mouginot: What we do is research. As such, using AI, we try to not recreate what already exists, from a theoretical, conceptual and aesthetic standpoint. As great admirers of pioneers like Eliane Radigue, Vera Molnar or Peter Türk -among many others, we measure what their generative research brought to us back in the 1960s and afterwards. To propose new paradigms, we try to reach maximum hybridity between human curation and technological expertise.
I wouldn’t say AI is a “dumb” assistant, or at least it is as dumb as we are. It depends on the quality of our dataset and our models- but it’s absolutely not the nth member of aurèce vettier. AI, again, is a tool. In our case, the tools and process are secondary concerns. Our first concern is to remain constantly in a precarious equilibrium, on the frontier between an imaginary, multi-dimensional data space and our beautiful, analogical nature.
Filippo Lorenzin: I believe that very often the general public is interested in AI-generated art in the same way as they would be with a magic trick; something happens behind the scenes and we, the viewers, are left to enjoy only what the artist/performer lets us to see. If we think about AI as any other tool, I’d suggest there is one main difference between it and, say, a hammer or a brush; these are tools that have been used for many centuries and everybody knows how they work, at least at a basic level. AI is relatively new and only experts know exactly how it is developed and used to generate contents, whether they are pictures, texts or music. In other words, I wonder whether AI-generated art would be intriguing in the same way as it often is now if the public would know more about it in practical terms. What do you think?
Paul Mouginot: Art is not engineering, so we would tend to disagree. Of course, it might be of interest sometimes to reveal which algorithms were used – be it StyleGAN, Clip or anything else. It would be awesome to describe the significant amounts of work needed to create our own datasets as it often represents 95% of the workload. It could be interesting to show how we interacted with craftsmen and métiers d’art to create pieces that are out of this world. But in the end, what matters is the outcome. We want the audience to visit our shows, not necessarily our atelier. We want the audience to meet our pieces, not necessarily our process, even though much of our work is conceptual.
Filippo Lorenzin: Elegia Machina(2019) is a book of poems written in collaboration with a Markov chain, a mathematical process that has been used many times in the past, notably in music. You rightly mention the influence of Paul Verlaine, Tristan Tzara and Guillaume Apollinaire, among many others, poets that lived through what can be seen as the initial stage of modernity. Could you tell me more about this work and your point of view on the legacy of certain modernist experiments in art?
Paul Mouginot: It’s more than an influence, we actually trained the Markov chains on the works of these poets among a few more. But we are very inspired by movements like Fluxus,OuLiPo, pieces playing with typography like those of Carl Andre, Guy de Cointet or more modern generative poetry. Damas Froissart, the graphist behind Elegia Machina also introduced us to Typoésie, a sublime book by Jerôme Peignot which is now constantly in our mind. We spend a lot of time reading and discovering new propositions in the field, not for benchmarking purposes but rather because it’s sometimes exciting as a research to partly build on the propositions of others. However, we haven’t seen many artists doing numerous forth and back trips between real and data spaces to create super hybrid texts and visuals, so in this field it looks like we are clearing new territory.
Filippo Lorenzin: In Latent Botanist Writings (2020) you trained your GANs with old herbaria and generated a new alphabet. The result is beautiful: the letters look like long-forgotten handwriting styles left to intertwine with plants and branches in an organic way. Can you elaborate on the technical process behind this work?
Paul Mouginot: When we programmed the GANs on herbaria, it generated not only impossible flowers and distorted colorimeters, but also strange attempts at reproducing what was supposed to be notes written by botanists. But it was in a new kind of alphabet, and it was wonderful to see.
In the exhibition, there will actually be reproductions of these raw “impossible writings” as oil paintings. So I gathered many GAN-generated notes and manually extracted a significant amount of letters, based on purely aesthetic considerations. Then, I created a new font that I called “Latent Botanist” and that is now part of my practice. I try to only have it on an off-line computer as I don’t want this font to spread. It’s kind of an industrial secret for me.
Recently, I was invited to take part in Twelve, a crypto-litterature project initiated by Kalen Iwamoto. For this project, I proposed a poem written in collaboration with Markov chains, the same process I used for my poetry book Elegia Machina but written in Latent Botanist. The piece was introduced on Twitter with not much information and a few minutes later, Rishabh Chakrabarty, a researcher at Refik Anadol studio was already trying to decode it. He mobilized many expert friends around him and spent almost a day on the enigma. Sometimes, I bury parts of a Rosetta stone, either in the blockchain or on physical works. Many of my pieces or texts are deeply intertwined, as you very rightfully say.
Filippo Lorenzin: The need to translate and transfer data from the physical world to AI and back again seems to play a central role in your research; for example, in We Will Make You Bow To The Delicate (2020) you used the alphabet generated by GANs to write a poem on small sandstone tablets. When did you start developing this physical approach?
Paul Mouginot: The need to project virtual calculations, modeling or generations into physical objects is a structural aspect of my practice. However, when I work with technologies, I don’t feel the need to turn everything into a glitchy, screaming, post-Internet circus. It’s actually quite the opposite. I want to be in a monastery, in a garden, in a silent room somewhere in a ryokan. I want to be in a mountain, with birds flying high, and as I look up, I want to be fascinated by the strange shape of an impossible tree. On one of the sandstone tablets, I wrote a poem, indeed. But on the other hand, I wrote a medical diagnostic corresponding to a super hard time I had in a hospital a few months ago. So I can say two things: these pieces are so beloved and sensual that it took me a while to let them go, and the contemporary collectors that got them are fortunately good friends. And without the use of sandstone and the elegant secrecy offered by the Latent Botanist writings, I am not sure the balance could have been reached. In the exhibition, I’m not just working on physical pieces. The paintings were created in collaboration with a Chinese craftsman called Jack Lee. The five large-size bronze sculptures are made in collaboration with Fonderie Fusions, an award-winning métier d’art located in the Auvergne region in France. So these objects will have outstanding quality and it is crucial for me. Lastly, the works we produce are never “medium-native” but rather kinds of hybridization. For instance, our oil paintings are augmented. The wooden frame holds a lot of additional information and potential adventures.
Filippo Lorenzin: To what degree do you believe the use of AI in art is tied to its technical development? In other words, would new engineering discoveries lead to works and art projects unthinkable with the current technology or it doesn’t matter as much as theoretical and philosophical research applied to it?
Paul Mouginot: Usually I dislike answering in this way but I strongly believe the use of AI in art is and is not, at the same time, tied to its technical development. On the one hand, AI research is not a new field and generative artists didn’t wait for GANs to emerge, to research such technologies. Sometimes, artists even had ideas and only could implement them a few years later, creating a “double datation” on the artworks. This is for instance the case for Vera Molnar. On the other hand, I could witness the emergence of new types of processes, new types of aesthetics and new types of experiences around AI. Recently, I’ve been fascinated by this piece from James Yu, where two AIs try to schedule an appointment together and always come up with excuses. I believe this could never have been created in a relevant artistic way without the emergence of GPT-3 for instance. As far as I’m concerned, the starting point of a work can be either technical, because I try to play with sophisticated tools or intimate, when I play with an idea and build on it. In any case, this is the start of numerous back and forth trips. For instance right now, I’m playing with text to image generation algorithms such as CLIP, or Phil Wang’s “deep daze”, trying to adapt them to my taste and visual ecosystem. My poetry book Elegia Machina is, in a way, an imaginary journey. I selected fourteen important steps, like a chemin de croix. I’m generating images to depict it, and will then turn them into small oil paintings, like precious post-surrealist icons.
Filippo Lorenzin: What are you working on right now?
Paul Mouginot: Apart from the exhibition in December, I’ve been working on various collaborations, and I am finishing a very ambitious project with Galerie Gismondi in Paris. I try to follow my own rhythm, read a lot, follow what’s happening of course but without being too embarked in the current frenzy. The last piece I worked on is a personal trial, using an implementation of the “deep daze” algorithm. It is called “a bearded man working on a completely beige painting (AV-2021-U-120)”. I loved the outcome of the algorithm. I’ll turn it into a painting with the help of Jack Lee, and might actually keep it around for a while.
