conversations – Interview by Anika Meier – 06.04.2023
EMERGENT PROPERTIES: EXPERIMENTING WITH TECHNICAL INNOVATION
GENERATIVE ART AND ARTIFICIAL INTELLIGENCE
What actually is generative AI? Why is AI the future? And how do you select artists for a new platform? In conversation with Anika Meier, the artist Ira Greenberg and his two co-founders at Emergent Properties, Emil and Sandy Corsillo, talk about generative AI.
I remember well a dinner in 2022 during Art Basel. Present were art collectors who had not really come into contact with digital art before. So soon we were chatting excitedly and very soon agigated about NFTs. They said that everything that can be found there is bad. "Where?," I asked them. "On OpenSea," they said almost in unison. I then asked if they liked all the music on Spotify and how they would go about their search on OpenSea. "Click and scroll," they said.
If you don't spend your days on Twitter, you basically don't know what's going on in the NFT space. There, you hear about drops and new platforms. And if you are not on NFT Twitter constantly, you might feel like the group of art collectors: you don't know where to start to actually find what you might find good. I woke up one morning late last year and my Twitter timeline was full of THE ORACLES NFTs by Ira Greenberg. The platform was new: Emergent Properties. I DMed him on Twitter and that's how our conversation about Generative AI started. Emil and Sandy Castillo, co-founders of Emergent Properties, now joined the conversation. In the meantime 4 projects have been published.
Anika Meier: Emil and Sandy Corsillo, you co-founded Emergent Properties together with Ira Greenberg. Why was it important for you to get an artist on board?
Emil Corsillo: To answer your question of why it was important, Ira brings a history of working in the generative space long before blockchain, when artists knew with certainty that there was no money in it. So having him on the team really cements our commitment to placing generative and AI artwork in its proper historical context. In Israel, we have a primary source with a flawless pedigree, and we have great creative chemistry!
Sandy Corsillo: Well, technically, we already had an artist on board, with Emil, but I know what you mean. The relationship with Ira began with us chasing him to do a project with us. We loved his work, and we really viewed him as a pioneer in the space, so there was a ton of respect. Then, after working together, he and Emil really gelled. It was one of those things that just kind of made sense right away.
AM: Ira Greenberg, what excites you about working on Emergent Properties?
Ira Greenberg: At first, I was just excited to work with Emil and Sandy and also with our other partners, Victor and Luis, and their development teams. I really didn’t know what we were going to do other than build something in Web 3. As mentioned before, Emil and I just clicked right away. I just felt I could be myself around him and the team with this almost instant understanding. We actually grew up across the Long Island Sound (a decade apart) from one another. There seems to be some unspoken familiarity, or maybe this is all in my head.
I remember my wife saying one day, "This is what you always wanted—to work with a team of other super talented people with a shared vision." I think the other aspect of EmProps is that art really is king. We’re not just a company trying to make money in the art market, but we really want to promote great art and artists and move the needle, especially in regard to AI’s connection to the most serious creative practice.
AM: What does Emergent Properties stand for?
SC: Emergent Properties was a name I came up with when I was trying to describe a platform that would enable work that I couldn’t fully envision but knew would be transformative. The idea was: build the tool, and then the magic will emerge. True to this origin, we exist to experiment with technical and creative innovation and then bear witness to what emerges.
AM: Emergent Properties is the first generative AI platform. Why do you think is it important to combine generative art and artificial intelligence?
EC: The thing about AI image generation is that it’s never perfectly predictable, and its inner workings are mysterious. At a high level, we know how it works, but we still can’t perfectly predict the outcomes. So you end up with an art-making process that is very iterative and dependent on trial and error. This process can create extreme frustration, but it can also lead to magical discoveries.
So, in a sense, the process of creating art with AI tools is already long-form, but just not in a structured, controlled way. Marrying long-form generative processes with AI art tools allows the artist to capture that process and embrace the elements of discovery, chance, and nuanced iterative outputs. Each output becomes a moment in the creation process that is witnessed by the artist and collector / viewer together.
AM: What are some of the technical challenges and limitations?
SC: There are challenges and limitations from a platform standpoint—getting the thing to work as intended but also creatively. One of the biggest technical limitations is time. Most generative mints provide somewhat instant feedback to the collector. This is because the artwork’s code can run on the user’s browser or machine. For EmProps, we are using massive servers that are running software that isn’t available in a user’s browser, so instant feedback isn’t available.
But as we have moved forward with multiple collections under our belt, we have begun to see this as a benefit in some ways. The outputs are generated by our pipeline and revealed one at a time, so that everyone witnesses the reveal together. In this way, EmProps mints are closer to a communal event, like going to a movie or a show, in which the audience is all seeing the same thing at the same time. Going forward, we really want to play off this by creating better and better viewing experiences. The BEASTS exhibition at EXPANDED.ART in Berlin in March 2023 is an example of us moving in that direction.
AM: The first release on Emergent Properties was THE ORACLES by Ira Greenberg. Why did you decide to launch the platform with a PFP project that is based on the history of art?
SC: The plan from the very beginning was for Ira to launch the first collection on the platform and Emil the second. The funny thing is that Ira wasn’t thinking about it as a PFP. My recollection is that he was, somewhat ironically, trying to use AI to return to his roots in painting and to evoke Renaissance imagery and style. We call it a PFP now, but THE ORACELS was more of a portrait study than anything else. It was really Emil who pushed Ira to lean into the PFP thing. I guess it was a double nod to Renaissance portraiture and Web3 PFP mania.
AM: How do you choose the artist you work with? Meaning, what are criteria for good generative art and for good AI art?
EC: The process at the moment is pretty instinctual: we reach out to artists who we like and who we want to work with. If they share our vision, then the next step is to introduce them to our tools and try to figure out how we will achieve the vision they lay out. For some artists, this is going to require building new tools or changing the ones we have, but that is really the best way for us to decide what to build next.
IG: The one thing about AI artwork is that it can be a maddening process of trial and error. This isn’t that much different than traditional generative art; however, it is as if you have a partner you are working with, who has their own idea of what they want, and you just need to work with them, not against them. That’s all to say that the best artists for EmProps are people who love experimentation and don’t get discouraged by early failure. The biggest difference between that and traditional generative artists is that the tools are so new that there is really no expertise yet. As a result, we really choose people based on their past work and don’t really worry about their experience. We are lucky to have a great team that can work with them to fill in any technical gaps.
AM: Are we still at the beginning when it comes to artists working with AI to create art?
EC: I think we are at the beginning of the beginning. The reality is that we have no idea what is coming, either with the core technology or with artists learning to use it to its fullest potential. The one risk I see here is that the exponential nature of the technology may put artists in a perpetual state of playing catch up. That so much of their energy and focus will need to be spent on learning the tools that there won’t be much left for pure creativity. One goal we have is to build better and better tools for our artists to use as a way of helping them focus on the artwork itself.
AM: What does "curated platform" exactly mean? How do you work with artists on a drop?
EC: Curated means that we select the artists and work with them on their projects. We decided to go with a curated model for practical and qualitative reasons. Practically, it is expensive to run the machines we are using, and they have limited bandwidth. If we structured it as an open platform, it would be prohibitively expensive for us.
In the other sense, we feel that AI art in general is extremely well suited for curation. Basically, the barrier to entry is non-existent, and the process can be opaque. This means anyone can make artwork, but determining quality can be hard. When we take a subjective stance about what we are looking for and say this is our idea of great work, we make a small impact in contextualizing and validating the world of AI art.
But we don’t want to be a monolith or the only voices in the room. Our plans, which we plan to start implementing in the near future, include: establishing a curatorial committee with rotating members; setting up community curation initiatives; inviting guest curators to the platform; and implementing a “pay it forward” model where each artist who mints a project on the platform is asked to invite an artist whose work they love.
SC: The funny thing is that we are curating the collection both by selecting artists and working with them on their work. But in a sense, we are totally giving up curatorial control when the mint starts. At that point, it is all up to the AI. That can be a scary but amazingly fun experience.
AM: Do you plan to push what is technically possible one step further with each drop?
EC: We plan to continuously push the envelope and drive ambitious technical experimentation, but we won’t announce a totally new technical achievement with every single drop. And based on the way our technology works, many of these new technical achievements will build on what came before.
AM: Ira, after THE ORACLES it’s BEASTS. How do you start working on such a project when it comes to creating art with AI?
IG: I start it the same way as I do with all my other work, regardless of medium. I usually begin with a vague hunch or impulse to try something. I don’t have (nor do I want to have) too much of a formed idea when I begin. I need to find my way emergently as I start making. Almost any other approach tends to feel contrived and too constrictive. Because the AI is generative with, in a sense, parameterized randomization, it feels very similar to me to creative coding and also to drawing and painting. In many ways, AI is the perfect medium for a painter-coder, especially one who likes moving as quickly as I do
I just keep iterating, and eventually stuff happens. It’s often not directly related to the initial impulse, but I’ve learned as I’ve gotten older to remain somewhat detached from the output, almost as an observer. Occasionally, really cool shit happens, and I post it to Twitter. Anika, I know you’ve witnessed this manic posting.
AM: When do you know a project is finished and you need to let go?
IG: I let go when I start repeatedly screwing it up. I won’t end a project when I get bored or lose interest; I’ll end it when I can’t really push it any further. I’ve always been this way: push, push, push. Painting used to get me in a lot of trouble when I was younger. My determination to make a great painting often interfered with my ability to make even a good one. It took me a long time to learn to see what was happening in the work, independent of my conscious desires. This remains a constant ledge to navigate, a BEAST within dynamic.
Ira Greenberg has an eclectic background combining art and computer science. Well known for having written the first book on the Processing programming language, Processing: Creative Coding and Computational Art (Berkeley, CA: friends of ED, 2007), Greenberg has led the way for the present and future of creative computing and AI art.
Greenberg holds a B.F.A. from Cornell University and a M.F.A. from the University of Pennsylvania.
Ira Greenberg's solo show BEASTS is on view at EXPANDED.ART in Berlin until 16 April 2023. In collaboration with Emergent Properties. The BEASTS by Ira Greenberg are available on Emergent Properties. Edition: 1000, 48 Tez.