While some artists might use a paintbrush or turn to code to create their artworks, artist Pindar Van Arman combines the two through his use of robotics. For the last 15 years, Van Arman has been teaching robotic arms to become skilled creative painters. While his first robots could only complete simple tasks, like painting by numbers, over the years, Van Arman’s Painting Robots have become increasingly more sophisticated. Through the use of deep learning neural networks, artificial intelligence and feedback loops, the robots are even able to make their own independent aesthetic decisions.
Pindar Van Arman first used Verisart in 2018 to certify paintings by his robots after winning the prestigious Robot Art Award. Thinking he had lost the COAs (Certificates of Authenticity) for his early works, Van Arman reached out to Verisart in November 2021 for help. The certificates, permanently registered on the blockchain and stored by Verisart, were quickly found and Van Arman regained access to his Verisart account and the certificates. Van Arman explains on his blog, "imagine losing a treasure then finding them, and finding that they were preserved on the blockchain. That is really the power of NFTs and the Blockchain for us artists. It is a new kind of immutable provenance."
What started with using Verisart for his physical works now extends to using Verisart for NFT minting. Van Arman’s Certified NFT, Ghosts in the Machine - bitGANg 1001 is included in the curated NFT auction Artsy x Verisart: 20 for 2022, with bidding open through to February 8.
Vinciane Jones, Head of Community at Verisart, caught up with Van Arman to discuss his AI and robotic art and his journey into the NFT space.
Vinciane Jones: A question that comes up frequently in conversations about AI art is, ‘Who is the artist?’. You use robots and AI to create your works and you’ve previously spoken about the distinction between creativity and art in the context of your art. Could you elaborate on where you draw the line between these two concepts?
Pindar Van Arman: In the AI art field, there are a few AI creators who claim that the AI is the artist. To me, these are false claims because the AI doesn’t have the intent of making art. The AI doesn’t wake up in the morning thinking, I want to express myself today, make a piece of art.
Saying that the AI is the artist, is ultimately a misrepresentation of the capabilities of AI. AI is absolutely revolutionary and to exaggerate what it can do, does a disservice to the actual AI art that is being created.
However, on a creative level, I think that AI is very close to humans. If you draw a parallel with the stages an artist goes through in their training, I would say that AI is probably up at the level of a college student.
For example, I’ve taught my robots to use a creative feedback loop. They’ll paint a couple of brushstrokes, a camera will take an image of what they’ve painted and they make decisions on the placement of the next brushstrokes based on what’s there. I’m still part of that feedback loop, I might play with some of the levers, change out the color and then let it go. I would say only one in three times do I not touch it, do I just let it run from start to end. And when I do, I notice that the work was done all by the robot because it doesn't usually do well by itself.
The robots are not creating art. A piece of art is anything that communicates one thing from one person to another. AI, for the time being, fails in two regards; It does not intend the piece to be art and it's not a person.
Vinciane Jones: You’ve been working in the AI art space for over a decade. How has AI art evolved over the years?
Pindar Van Arman: As soon as AI was created and people started experimenting with the early low power neural networks, they very quickly saw the potential for it to be art. But no one has paid much attention. When I started, there were very few artists working with AI and robots, it was primarily academics. Early artists included Harold Cohen, Patrick Tresset, Mathew Stein and we would all talk to each other.
The AI Art at the time was very limited and it wasn't capturing anyone's imagination. I did a show in Washington, DC and the review begins, “this might look like it was made by a high school student but when you realize it was painted by a robot that really changes everything”. The writer might have thought I would be flattered but all I heard was “this is high school level art”.
Everything changed in 2014 and 2015 with deep learning. The neural networks are finally getting big enough to actually make amazing things happen. I discovered a style transfer where you can take a picture, compare the picture to a Van Gogh (for example), and ask the AI to replicate the image in the artist’s style.
Then eventually GANs, generative adversarial networks, came along. You show a neural network many examples of something, and it’s then able to spot the patterns and create versions of what it’s been shown. And that’s when I created the works I certified on Verisart.
Vinciane Jones: 2018 was also the year you started using Verisart, could you tell us a little more about the works you certified on Verisart in 2018?
Pindar Van Arman: For my Verisart works, I created an algorithm that would imagine faces. I was interested in the moment right when the faces came out of nothing. So I made a big series of hundreds of faces and I would have my robots paint them. That work won Robot Art 2018.
After they won the award, I put 128 paintings on the bitcoin blockchain with Verisart. I gave out the works at an event in New York to collectors including Kevin Abosch and Monica Lewinsky. I remember including the Verisart certificate which each painting I gave away. I'm always experimenting with blockchain and Verisart seemed like a cool application of the blockchain.
Vinciane Jones: It’s interesting to hear that these works were such an important part of your career.
Pindar Van Arman: Yes, these were my first use of GANs and it was the first time I’d received such an international prestigious award.
Vinciane Jones: We’ve worked with several other artists like Trevor Jones and Lucho Poletti who were drawn by the uses of blockchain and were certifying their works on Verisart even before they started creating NFTs. Now that we work across physical and digital certification & minting, it’s exciting for us to see the evolution of their physical practice into their digital and NFT practice. When did you first become interested in the possibilities offered by the blockchain?
Pindar Van Arman: My first dabble into the blockchain was an experiment where I had a robot that was painting online and I called the CloudPainter. Anyone could go to the website and anyone could take over my robot and paint. The issue was a lot of vandals were using the website to paint phalluses or hate speech.
I decided to add a paywall and ask people to donate BTC and depending on how much BTC was donated you own that percentage of the painting. It was a cool way to share ownership of a piece of art with an audience and encourage good behavior. The problem is I think only two people did it. At the time, my audience wasn’t used to transacting in crypto and there were too many hurdles, it was a little ahead of the mainstream.
I would call it a field experience. Everything was a failed experiment because so few people were using it half the time. But back then it was so fun.
Vinciane Jones: After your CloudPainter experiment and starting to use Verisart, you joined SuperRare. Could you tell us more about your NFT journey?
Pindar Van Arman: I think I used Verisart a little before I got onto SuperRare, in early 2018. I’ve often felt like an imposter in the digital art space. On SuperRare, everyone else was making these purely digital works but my robots were painting something, a physical piece.
I came to this agreement with SuperRare where I agreed to destroy the physical paintings. What’s funny is that I was the first photographer on SuperRare because I would photograph my physical works and then destroy them. Everyone likes the early, early work cause they're so raw.
Vinciane Jones: You’ve been incredibly active and successful in the NFT space, what is the greatest aspect of NFTs and blockchain?
Pindar Van Arman: The greatest thing about the blockchain is that artists and collectors are directly connected. No one in between us. The liquidity of this and the secondary market is fun for collectors, you can really root for an artist.
And if collectors want to start collecting works by a different artist, it's not hard to sell their work and put the money into a different artist. There are also royalties, which are amazing. I think it might be at the point where royalties can pay for my studio. The ultimate freedom for an artist is when you don't have to worry about sales anymore and I'm on the verge of that. Once an artist has broken that threshold, then I think their true art can come out because they're not worried about sales anymore.
I don't think I would be at this point of my career in a traditional art world.
Vinciane Jones: We’re currently working with you on the curated NFT auction Artsy x Verisart: 22 for 2022. You’ve included the work Ghosts in the Machine - bitGANg 1001 which was minted as a Certified NFT on Verisart. These bitGANs are part of a new series of works, what’s the story behind these?
Pindar Van Arman: I’ve started creating new series of works with bitGANs that are a culmination of my life’s work. As an AI artist, I’ve always been interested in the cutting edge of AI and the technology itself, but people often lose interest when they don’t understand what’s going on. So I started developing my bitGANs and podGANs which are low res pixelated works created from some of the best algorithms in the world.
These are influenced by Minecraft and Space Invaders so the works feel familiar and remind people of their childhood. It’s been fun getting the general public interested. I created 1024 podGANs and they sold out in minutes.
They can also take on a life of their own, I have this mechanism that if you buy two with special traits, you can combine them and they'll take over my robot to paint themselves. As the final iteration of the work, I mint a photograph of the painting, superimposed with neural networks that imagined the painting.
It took me three years to realize this, but the neural network thinking is beautiful, but I always just thought about it as the scrap paper that I had just left on the floor of the studio.
You know what? I have to think about what I need to do three years from now and start doing it right now. So after this talk, I'm going to take a break and say, “what will I be doing in three years?” And so I can just start doing it.