UNIT-ON-CHAIN: Sofia Crespo and Helena Sarin on GAN and the Role of Artists

Unit-on-Chain is a podcast series from Unit London offering a ground for critical discussions for artists and thought leaders from the Web3 ecosystem.

Season 1 of our podcast coincides with In Our Code, a highly-anticipated exhibition of generative and digital art in partnership with AOI.

For GAN and the Role of Artists, the fifth episode of the series, join us in conversation with GAN artists Sofia Crespo and Helena Sarin on creating art with AI, training datasets, and the role of an artist in the digital age.

LISTEN TO THE PODCAST HERE on Spotify and Apple.

What is the relationship between AI, GAN (Generative Adversarial Network) and generative art?

Helena: I always say that for me, Artificial Intelligence is not some kind of alien thing. I would say it’s something that I use as a tool and it’s just a bunch of softwares which I employ using my own art. Somebody who is not familiar can look at it as a kind of glorified Photoshop. So basically the tool that transforms certain input to certain output.

Sofia: Yeah, I totally agree with you. “The glorified Photoshop.” I think very often there’s this divide between AI art, in a capsule from the rest of generative art, and in a way that makes sense. Because a lot of AI art you see is mostly a top down approach because you decide a lot of what is gonna go in the dataset, what parameters could I use. But in generative art, it’s more of a bottom up approach where people usually build a whole system of rules to create an artwork.

How did you start using GAN?

Sofia: I started with Convolutional Neural Networks (CNN), it’s usually used for style transfer. That was my initial tool. I discovered GANs much later. There were people like Helena and Anna Ridler who had already been exploring GANs so I was really fascinated to see the work.

This is back in 2018, which is when I started and I was really interested in a tool that allowed me to automate the process of generating a texture. So I just wanted to extract textures from images and rearrange them. So that’s basically where I’m coming from in terms of what was the interesting thing behind it.

Helena: I started with Machine Learning as part of my consulting gig, so I never expected to use software as part of my artistic practice, because I was mostly an analogue artist. Then because I started using GANs at work, they worked so miraculously that I decided to try them on my own art.

Fortunately, I had tons of digital photographs from my food photography gigs. So I used them just out of curiosity, but the results were so fascinating. My mentor at the time, who had nothing to do with software or digital stuff told me, “You are onto something.” That was the encouragement, as if I needed any encouragement. I’m a very curious person by nature. So long story short, I started at the end of 2017. And here I am. 

How much “control” do you have in creating AI art? Does “control” play an important role for you?

Helena: Gerhard Richter once said that, “I need to create the conditions for randomness to do its work.” So, being a control freak I let the network give me something that I can work with. Maybe I’m guiding it a little bit, but in this case it’s a lot of fun to let the model go where it wants to go.

Sofia: At first when I started, I really felt like I couldn’t control anything of what I was doing because I came from doing photography and 3D rendering. In 3D rendering you can decide what happens at every part of the image. 

Switching to these tools and suddenly I felt like I couldn’t decide where the jellyfish is going to be, in which part of the picture. In a way, it was a bit unsettling, but in a good way. After a while I gained more control because I learned more parameters which I can use. 

Like “If I do this, this is gonna happen,” or “I think this is going to happen.” I feel like the tool became more naturalised just because I used it so many times. Then I started collating my own dataset so that I could create something more interesting.

In a way it feels like now I have control, but I still really enjoy this aspect of randomness and I prefer generating hundreds of variations and having to pick from one rather than controlling what happens at every part of one image. I think it’s a balance. I guess for every artist there are some things that are fun to completely automate and not have any relinquish control, but there are things that are good to be able to decide on.

What does “training” in AI mean? How does that impact the output?

Sofia: The idea of training is that it’s the loading or compressing of data in the model. The model can start from random noise and zero knowledge of your dataset. But with time as it trains, it starts to gain more of an idea of what the main or the core patterns in your dataset are. You could ask “the longer I train a model, the better it’s gonna be?” Not necessarily, sometimes you can let a model train too long and something called “overfitting” happens, which means your model starts to deviate again from the knowledge of your data set.

Helena: From an artistic perspective, it’s like Sofia said, you need to do a lot of monitoring. It’s not like you dropped your dataset, even if it’s well-cleaned and well-curated. You need to watch for what the network is doing. Maybe stop it, give it a jolt and let it continue because it might start hallucinating not in the direction that you wanted. It’s an interesting process that requires a lot of dashboard watching and a lot of monitoring.

(Scroll to bottom for full transcript)

Sofia Crespo (b.1991), an artist based in Lisbon, seeks to capture the way in which generative lifeforms manifest in the digital art space. She focuses on the portrayal of biologically inspired technologies in particular, showcasing how the artificial mimics the natural, through creative expression and the consideration of both organic and man-made spheres of life. Crespo advocates for artists to become more adept with machine-learning technologies, as these provide encouraging tools for creation, while enriching and transmitting culture in a universal language.

Born in Moscow, Helena Sarin lives and works in New Jersey. With a background in art as well as software engineering, she uses cutting-edge technology to produce complex, yet refined works of digital art. Working with watercolour and pastels, she draws inspiration from fashion and photography, as well as Russian Constructivism, to create her unique AI-meets-art aesthetic. The elegant nature of her work is visually pronounced, based on its attention to detail, rhythm, repetition and texture. Sarin served as a guest speaker for a ML/AI conference and, most recently, spoke at TEDx, MIT, the Library of Congress and at Capitol One. She has exhibited in collections in Zurich, Dubai, Oxford and Shanghai, among other locations worldwide. Her work was featured twice in the publication Art in America (January 2020 and May 2021) and in a 2018 BBC publication about AI-created artwork.

Abigail: Welcome to Unit-on-Chain, both of you.

Helena: Thank you. 

Sofia: Hello. Thank you.

Abigail: To both of you, our first question is about the relationship between AI GAN and generative art. And how would you define these terminologies to an audience that might not know much about them?

Sofia: Helena, would you like to start?

Helena: Yeah, I can start. I always say that for me, Artificial Intelligence is not some kind of alien thing. I would say it’s something that I use as a tool and it’s just a bunch of softwares which I employ using my own art. Somebody who is not familiar can look at it as a kind of glorified Photoshop. So basically the tool that transforms certain input to certain output. 

I don’t consider myself to be a pure generative artist. Because they work with other tools like Processing language to start from the idea and go straight into generating mostly abstract things. But in my case, I work similar to Sofia with something that produces organic (images) or something that is familiar to people. Even if we work with alien-sounding tools.

Sofia: Absolutely. Yeah, I totally agree with you. “The glorified Photoshop.” I think very often there’s this divide between AI art, in a capsule from the rest of generative art, and in a way that makes sense. Because a lot of AI art you see is mostly a top down approach because you decide a lot of what is gonna go in the dataset, what parameters could I use. But in generative art, it’s more of a bottom up approach where people usually build a whole system of rules to create an artwork. And I think it comes more from smaller elements to something larger, which is the final output. That’s the way that I think about it. In my case, I think generative art can be considered both things, so I call myself a generative artist as well.

I think in a way, they’re just different tools and I’m really fascinated by AI tools in particular. Yeah, I agree with Helena that the name is a little bit misleading sometimes in a way. 

Abigail: What brought you both to GAN initially, over other practices? I know I’ve talked to you guys individually about this, but I would love for the audience to hear you guys’ initial stories.

Sofia: In my case, I started with Convolutional Neural Networks, it’s usually used for style transfer. That was my initial tool. I discovered GANs much later. There were people like Helena and Anna Ridler who had already been exploring GANs so I was really fascinated to see the work.

This is back in 2018, which is when I started and I was really interested in a tool that allowed me to automate the process of generating a texture. So I just wanted to extract textures from images and rearrange them. So that’s basically where I’m coming from in terms of what was the interesting thing behind it.

Abigail: What about you, Helena?

Helena: I started with Machine Learning as part of my consulting gig, so I never expected to use software as part of my artistic practice, because I was mostly an analogue artist. Then because I started using GANs at work, they worked so miraculously that I decided to try them on my own art.

Fortunately, I had tons of digital photographs from my food photography gigs. So I used them just out of curiosity, but the results were so fascinating. My mentor at the time, who had nothing to do with software or digital stuff told me, “You are onto something.” That was the encouragement, as if I needed any encouragement. I’m a very curious person by nature. So long story short, I started at the end of 2017. And here I am. 

Abigail: Amazing. The one thing about GAN and AI, I do think there’s so many different practices within it, and two of you are at the top of it. I would love to go more into your individual practice and talk about the specific things: about ideation, data sets, inputs. Sofia, specifically with you, neural networks, and then Helena with your work with the patterns you bring out of your analogue work.

Sofia, if you wanted to start about your practice specifically?

Sofia: Okay. So my practice, first of all, has changed throughout the years. I feel like now I think a lot more before [I create]. I like to play, but I also plan more about what I’m going to do, whereas at the beginning I didn’t plan my datasets so much. I was constantly just playing and throwing things without thinking too much. 

Basically what I would do is I would create a setup, like a virtual environment to run a code repository. And then from there I would create a dataset and run the code on that dataset, and I would generate an amount of variations.

From there I look at the outputs and I think, “What do I like? What do I not like here?” And sometimes it’s a feedback loop. Some of that gets fed back into the dataset. It’s a kind of self feeding system and I start to see what I want to explore at that moment.

For some projects, we trained much larger models because we wanted to represent more of the larger scale point of view of nature. So we wanted to have a lot of species. So for that, it took several weeks to train and we had to do a lot more planning on how to run it.

There were a lot of technical things to take care of there. Like distributing the computational workload across several GPUs. 

So a lot of my work has those technical things, but because I don’t have a Computer Science education, I have to learn on the run how to solve it. So a lot of it is like learning on the run, asking for help from developers.

But that’s what I like. I really like learning. I like playing with these kinds of thoughts because I like learning about it. I hope that’s a good description of my practice. 

Abigail: I think it’s great. I really think you see the research and inquisitive mindset you have. It comes through in your work as well, especially the natural history element, specifically with the piece in the show, In Our Code.

But to Helena, can you talk about your artistic practice and how you combine the analogue side of your photography, sketches and drawings with GAN and where did that come from. 

Helena: Basically, it just happened by chance because somehow I became known as a person who trains all of her own datasets, and this is what I stuck with.

Essentially that was a way to differentiate from all this. Like what Sofia mentioned, like style transfer, where we had these horrible study nights which still give me nightmares. And what happened with this personal dataset? They’re so small. At the beginning I used things like cycle GAN and other research GANs. But that was exactly the point. A lot of digital artists work with older technologies because they give them more interesting results. Instead conceptual artists working with Blender or Unity these days. Essentially gives a certain homogeneity while when you use these old tools, they allow you for better control because of their imperfection.

In a sense, I try to feed my datasets in the new type of GANs, basically you get either complete production of what you fed into it. They just learn your data set and start giving it back, or they can clearly collapse and give you really ugly garbage.

So the cycle begins again, they produce this semi-abstract result, especially with my sketches. This is something that I’ve been training for more than a year. Because it can’t decide ways to be done, it continues to hallucinate and basically every snapshot can give you some results.

I like this quote by Linda Berry who said, I’m paraphrasing, “Rely on the human inclination to find the useful patterns in random information.” Basically this is what I play around. That’s why I said titles are so important to me because I stare a lot into this garbage or something nicely looking but has meaning. I come with the meaning, I give you some meaning and that’s how the work is born in a sense. 

Abigail: Amazing. So how much control do you have in creating these artworks? You raised an interesting aspect that you just talked about, control. So does that play an important role for you?

Helena: Like a cultured Russian person, I like quoting. So Gerhard Richter once said that, again I’m paraphrasing, “I need to create the conditions for randomness to do its work.” So, being a control freak I let the network give me something that I can work with. Maybe I’m guiding it a little bit, but in this case it’s a lot of fun to let the model go where it wants to go. 

Abigail: What about you, Sofia?

Sofia: At first when I started, I thought, “Wow, this is amazing.” I really felt like I couldn’t control anything of what I was doing because I came from doing photography and 3D rendering. In 3D rendering you can decide what happens at every part of the image. 

Switching to these tools and suddenly I felt like I couldn’t decide where the jellyfish is going to be, in which part of the picture. In a way, it was a bit unsettling, but in a good way. After a while I gained more control because I learned more parameters which I can use. 

Like “If I do this, this is gonna happen,” or “I think this is going to happen.” I feel like the tool became more naturalised just because I used it so many times. Then I started collating my own dataset so that I could create something more interesting.

In a way it feels like now I have control, but I still really enjoy this aspect of randomness and I prefer generating hundreds of variations and having to pick from one rather than controlling what happens at every part of one image. I think it’s a balance. I guess for every artist there are some things that are fun to completely automate and not have any relinquish control, but there are things that are good to be able to decide on. 

Abigail: Super interesting. And then Helena, with your process, do you create a hundred outputs as well and then curate it down, or do you have a singular output that you’re constantly adjusting?

Helena: There are two aspects to this. Curation is a huge part of any artistic process, of course. But with GANs it becomes really part of your process. You either make it like Mario Klingeman. He did a lot of work around automating the output, like building the agents that are capable of finding the right things.

But I’m a manual person in this case. I do a lot of curation, and I also like to employ the so-called chaining a lot, which is basically not a single output, but maybe a bunch of output and to stream them through another. Get another model and build a pipeline of models and as a cherry on top, have some programmatic way for post-processing to do what Sofia mentioned, like collaging.

I mostly use grades and the work I’m presenting in this exhibition is a fine example of this. So I mean curation is a huge part, but processing and automating the curation and maybe playing yet another kind of process with curated material is part of the process.

Abigail: Another word that I feel like is often used in conversation with GAN is the word “training”. What does that word mean to you both? Can you explain to the audience what that means when creating a work.

Sofia: Yeah, so basically the idea of training is that it’s the loading or compressing of data in the model.

So the model can start from random noise and zero knowledge of your dataset. But with time as it trains, it starts to gain more of an idea of what the main or the core patterns in your dataset are. You could ask “The longer I train a model, the better it’s gonna be?” Not necessarily, sometimes you can let a model train too long and something called “overfitting” happens, which means your model starts to deviate again from the knowledge of your data set.

Abigail: Yeah definitely. Helena, do you have anything to add to that about training? 

Helena: Yeah, I mean from an artistic perspective, it’s like Sofia said, I mean, you need to do a lot of monitoring. It’s not like you dropped your dataset, even if it’s well-cleaned and well curated. You need to watch for what the network is doing. Maybe stop it, give it a jolt and let it continue because it might start hallucinating not in the direction that you wanted. It’s an interesting process that requires a lot of dashboard watching and a lot of monitoring.

Abigail: Interesting. This relationship with machines, do you think they can really understand in that way? 

Helena: At some point they will, because again, we won’t be too futuristic, but let’s stay with the current reality. It can learn certain patterns. So you can tell it “this is the style of Van Gogh” or you can go further. You can define what’s placing, what’s not. Guided by this knowledge, it can adhere to the principle and can start giving you out the placing results. But again it’s not what I’m after.

There are some approaches, maybe Mario with his Botto project where you can automate as much as possible in terms of these autonomous agents. But I’m still working in this semi manual approach, like doing a lot of hand holding. It’s me who decides what’s interesting aesthetically, versus the network. 

Abigail: Super interesting. How have you seen the GAN and generative art as a whole really progress over the years?

Helena: I think right now the technology I’ve seen by other models because the researchers can’t sit still that they need to make breakthroughs after breakthroughs. So GANs are now shaded by these diffusion models. It’s hard not to notice DALL-E and Midjourney taking the central stage. Because it’s so widely covered by the media that AI art is basically, that’s what it is. If we go back to 2018, GANs were in the media, Sotheby’s and Christie’s were selling those art.

Then the media lost interest and for a few years we were just working. Now there is a new wave of interest. For me that’s a bit of a problem because I feel like for me personally, I exhausted the tool in a sense. I can definitely create yet another dataset and train yet another thing. But that’s the problem working with technology where novelty is maybe the main attraction. So there are definitely artists who will continue working with GANs, but I would say it is an art movement.

Abigail: Super interesting. 

Sofia: Yeah, I totally agree with what Helena was saying. In my case, I had been doing art for a long time before I discovered AI. It was only when I started to work with AI that people were interested in hearing what I had to say.

Very often when I presented my work, AI was the most important thing that people were most interested in, not so much what I was trying to do with the work. So I noticed people are genuinely really curious about the tool, but I think that there’s a lot more than the tool itself.

There’s also the whole spectrum of things that the tool can be used for that kind of space of possibility for telling a story or for your self-expression. A lot of it is like, I guess that kind of shiny new aspect that the technologies have. I really like looking at Helena’s work, and I’m not saying this just because she’s here in the podcast, but it’s that very often it grounds me.

It reminds me that it doesn’t matter where the currents are going, one can stay true to what one considers to be important to use the tool for. I think that’s something that’s very necessary in the space of art and tech in general. Very often the conversation is like, “so what’s the next big thing?”

Abigail: That’s really interesting. Going into that, I wanna shift the conversation into the community side of GAN and how important has that been in your artistic practice? And this new wave of Web3 community. 

Helena: I like to joke that there’s not much of a community in AI art. There are fractions. For a long time there were groups around particular curators: Luba Elliott was our favourite curator. We tend to gravitate towards people who understand us, who we understand and with whom I have no competition because we work in our respective areas.

So I would say that we developed a small community including people like Sofia and myself, Mario, Robbie and a few other people. Unfortunately that’s gone. I would blame crypto for this because they have a huge sense of community. Sometimes it’s fake, sometimes it’s not. AI art was drowned in it.

Right now we became part of the generative art movement. But I would like to talk more, not so much about community but about collaboration. For me that was a huge part. Because we started by discussing our plans with Sofia, which was very exciting and hopefully something comes out of it. There were two collabs I went through, both of which I consider something which have opened new venues for people. One of them was with Dmitri Cherniak called Gen2Gan, we are publishing the book that should come out pretty soon. It was very exciting for the community because in a sense it united two different strains of generative art, GANs and generative art. Honestly, I myself was surprised by the results and by the feedback that we got from people. It was really an exciting project and I would attribute the success to the fact that we have such different styles. With Dmitri, it was like a pipeline, not a usual collaboration in a sense that we work together on singular output, but I used his works to train my GANs. There were 17 models and there were tons of works that we produced as a result. 

The other thing is completely different. I collaborated on a book of recipes which we just published, The Book of veGan. (Pun was totally intended). Kate Ray wrote the recipes inspired by my generative vegetables and lettuces. I think collaboration of this nature is definitely fruitful, it’s an interesting angle and it’s definitely synergetic, I would say.

Abigail: What do you think, Sofia?

Sofia: I agree. I remember when Luba Elliot was organising exhibitions and back then NFTs weren’t really a thing. There was no money involved. We weren’t earning much from what we were doing but we were really passionate about it. There wasn’t a community, there was no media or Discord or something particular where we were logging in and talking. But we got to connect through these exhibitions and we knew each other and we followed each other’s works.

Then a lot of communities started growing around particular tools. When Runway ML grew, a lot of people joined the Runway community and they talked more about what’s possible within Runway. In a way it became fragmented and in the end what to me would mean community is something that I don’t fully follow. I think I would find it really overwhelming to be on Discord. Personally, I need time to step away from the crowds and really take time to think about what I’m interested in or what I want to explore with my work and to make it have meaning for me. 

I still think it’s really good to have been there all these years and have followed the work of Mario, Helena, and all these people who have been there doing it for several years now until the evolution of the work. That’s something that’s really good to have. 

Abigail: That’s great. I think it’s so important within art to have those conversations, the reflections, as well as to work together, because I think that’s where innovation comes from over a long period of time. We’re so happy to have you both on the show, and I wanted to take a moment for both of you to speak about your works in the show, In Our Code here at Unit London. Helena, we can start with you if you wanna speak about the work and also more about the input as well.

Helena: I don’t usually do commissions, let alone getting an idea of what I should be doing. That’s why essentially I had this first moment of resistance. Then I was challenged and spent time thinking about it. I did some work generated with John Zorn in mind. This is the avant-garde jazzman who really was an inspiration for me, for his wilderness, for his free use of material from all over the plate. People around him also were inspirations. I wanted to do blue note kind of thing, like playing out of pitch. It was nice that you mentioned the idea of sheet music. In a sense there are thesis and grids which remind people about sheet music.

There’s a moment of “you know you got it right.” The only thing that I had in mind was blue stratagem. The word stratagem for me was deploying the idea of subversion that runs through all my work, because you want to play against the digital correctness of your models. 

It’s a lot to unravel. But again, for me, it all makes sense. The blue notes, the sheet music, the Zorn inspiration, not so much in this particular piece of music, but more in the title. From the technical perspective, it was a chain of models of this grand super model that trained on my sketches. But because I use very small squares of the output and the model created for me, a few gig of checkpoints. I put them in a certain pipeline with glue code in between to get to the results that I need. Then I put them in grids and compose the collage. That’s pretty much from the technical perspective and from the inspirational perspective. 

Abigail: And it’s amazing. We’ll be linking all the works in the descriptions for our audience if they wanna click on them while listening. I just think it’s absolutely brilliant how you have incorporated that inspiration of jazz. I was just thinking about it the other day, there’s so much randomness in jazz, like when you improv. And I think it really highlights that note within your overall practice of GAN as well. I was looking at it and I was like, “Oh, this is just so beautiful, the sheet music.” This is very like holistic. 

Helena: Thank you so much. Two things to pick up on. The temporal aspect, which is very hard to depict in a static image, which is important in music. Also I think working with structural and pattern elements allowed for this temporal aspect. Also what you said about randomness and jazz, is what I call the “deliberate wonk”, because it makes all the sense I hope it makes sense to viewers as well. The whole thing is about jazz, free play, resistance, the freedom and improvisation. 

Abigail: Yeah. I really hope when a viewer comes into the exhibition the music at play brings out those elements and makes the audience who might not be as familiar with GAN think of it in a different way, rather than thinking of it just as Computer Science or Computer Engineering. It’s beautiful and I’m very excited for the audience to see it. It’s very deep.

Moving to you Sofia, your work as well has so many layers to it. If you wanna go into them. 

Sofia: Yeah, so first of all, when I first went to Unit London in person, when you first approached me for this show, I didn’t think at all that I would end up producing a physical piece.

Usually when I’m invited to a show that involves NFTs in some form, I always leave out automatically the idea that I’ll get to produce a physical work. When I mentioned to you that this was really special for me, the practice with cyanotype, you said, “How about doing a physical work?” And I was so excited and I still am.  For me there is one aspect of this work called essential_protozoa_1862, is being able to combine a digital technique like using neural networks. But at the same time, combining it with a physical process like cyanotype printing. I’m obsessed with this technique. I’ve experimented with a few printing techniques, but this is the only one that has stuck for me. I love working with it on fabric. It’s really a lot to do with stepping out of the computer, and I’m sure that you understand this, Helena, as well because you work a lot with physical media. It’s really good to step out of the computer and to actually do something with your hands or change your focus from the keyboard and the files.

Sometimes when you leave the computer and focus on something physical, the world feels a lot more chaotic and difficult to control because there’s so many unknowns. It’s kind of like the computer is its own capsule and things can be ordered quickly if you want them to be ordered. In the real world, it takes a lot more effort. Maybe I’m going to a side topic. The piece itself is about protozoa which are microorganisms. To me they’re really special microorganisms because they have an external skeleton made out of silica. I find that absolutely fascinating because very often when we think about microorganisms, we don’t think of beings that actually have some sort of skeleton.

Some years ago I was watching a documentary about Ernst Haeckel’s work, and I found that he talked about how hard it was to illustrate them because this skeleton would dissolve as soon as it was in contact with the air. He had to absorb the image of this radial area that he saw through the microscope, and then he had to try to draw it as fast as possible before he forgot about it.

That really struck me, this kind of organic thing of having to look at something through a microscope. I’m fascinated by microscopy. I have several microscopes at home. It’s something that I’ve been into for a while, even before discovering AI. 

I had never done work about that until now, but I felt like it was a good opportunity to bring this up. And I also really like thinking about the microorganisms that exist, embodies of water that we don’t necessarily see with naked eye. All these tiny life forms that just exist there and that are really hard to look at.

And in a way the microscope is a technology of its own because it allows us to look at something that clearly we cannot see in the day to day. I like this poetic layer as well, or the use of another technology like optics that helps us see the world in a different way and create a different relationship with it.

I think that’s the main inspiration for it. I’m also really, really inspired by the practice of Anna Atkins and this is more about the layer of the physical work itself and how she was experimenting with tools at that time in the 19th century that nobody had been using to document nature in that way, particularly like botanical specimens.

Abigail: We’re really happy that you also get to have the physical work and I think it adds a very deep layer of almost paying tribute to the women of natural history like Anna Atkins that we talked about before, and inverting the traditional cyanotype colours. To the audience, a traditional cyanotype is usually a deep blue background with white figures. Sofia’s work inverted the colour palette which I think is just beautiful. And you actually make these physical works by hand, which I just wanna highlight as well, which is incredible that you are an artist that’s so multidisciplinary too. It’s incredible.

We’re coming up to the end here and we do something fun that we’re doing with all of our guests. So we have a list of questions that we ask every single person who comes on here. So we just ask to make it a little trickier for you both, to answer the questions either in one word or one sentence. We’ll get started with the first one and then we can start with you Helena, and then go to Sofia. Our first question is, what does In Our Code mean to you?

Helena: Maybe if I hear Sofia’s answer it’ll trigger something. Code for me is in our genes in a sense, being an engineer. 

Abigail: What about you, Sofia? 

Sofia: I would just say perspectives. 

Abigail: I think that’s actually a really good answer.

Helena: For me it would be engineered art.

Abigail: These are very good answers and we haven’t had anything even similar before. Very deep. So our second question is, what inspires you most in the space? And the word space is purposely vague. 

Helena: Freedom to create.

Sofia: I think, slowness. 

Abigail: I like that. So our third question is, what is one artist you’d love to own a work by, or NFT by?

Sofia: I think I would like to have one by Kevin Abosch. What about you, Helena? 

Helena: Mario Klingemann.

Abigail: Great. And our last question is what technological advancement do you think this space needs?

Helena: Learning from the past. 

Sofia: That’s very good. Maybe less peer pressure. 

Abigail: Very good answers. Well, thank you both so much for being on our first season of Unit-on-Chain, Thank you both. 

Sofia: Thank you. 

Helena: Thank you. Bye.