“I want to paint with machine consciousness” — interview with Refik Anadol

Filippo Rosati interviews Refik Anadol on Zoom

Refik Anadol is a pioneering multimedia artist in data aesthetics and artificial intelligence. His work addresses the challenges and possibilities that new technologies have imposed on humanity and what it means to be human in the age of artificial intelligence.

Taking the data around us as primary material and the neural networks as a collaborator, Anadol and his team paint with an intelligent and artificial brush, offering us radical visualisations of our digitised dreams and expanding the possibilities of architecture, fiction and our own body.
For Machine Hallucination, 113 million publicly available images of New York City were used to imagine the next of a legendary city like NY; for WDCH Dreams (Walt Disney Concert Hall), 100 years of digital archives from the Los Angeles Philharmonic were used to inspire the images projected on Frank Gehry’s iconic building. For Archive Dreaming, Refik Anadol used machine learning algorithms to search and sort relationships among 1,700,000 documents. The interactions of the multidimensional data found in the archives are, in turn, translated into an immersive multimedia installation. Designed to interact with users, when it is active, it “dreams” of unexpected correlations between documents.

Can a building be able to dream? How can memories be viewed? How have technologies influenced our relationship with space and time? Filippo Rosati from Umanesimo Artificiale interviewed Refik Anadol on the themes of his artistic research and whether it is possible for a machine to dream.

Filippo Rosati: Your recent works are called “Archive dreaming”, “Machines Hallucination”; your research questions are “can a building dream?” or “can a machine dream?”. What fascinates you about a machine that can dream and where this fascination comes from?

Refik Anadol: I think it was my first encounter with the movie Blade Runner, when I was 8 years old. I clearly remember an incredible dialogue in the movie between two androids talking to each other and asking “what would a machine do with someone else memory?”.

Eventually, if machines have memory and consciousness, they will also have the capacity of dreaming and hallucinating; and this is a huge inspiration for me.

I couldn’t achieve a similar experiences in my first works. I’m using data to paint and sculpture since the beginning in 2011, but initially I couldn’t find a meaning from data. Then in 2016 I was one of the first artist in residence at Google at that was the first time I used AI and one of the reason me and my team are a pioneer in this field. Since then I couldn’t stop thinking what a machine do with human archives, with our collective memories. I’m not inspired by individual ideas but rather what are humanity’s important memories that can be critical information for a machine. Since then I’m constantly speculating machines hallucinations, dreams, because now AI allows machine to learn and give a context to data. Of course in a limited sense, but it’s an incredibly powerful tool to speculate the idea.

FR: Through the years, how has been your learning process? What humans can learn by interacting with a machine?

RA: First of all, let’s remember that AI do not forget (laugh). We as human beings, can learn, remember, but we can’t and we don’t have to remember everything. However, machines don’t forget and they can look at patterns of information in different ways. For example, when a machine learns it doesn’t store data in 3 dimensions; it stores data in “n” dimensions. If a machine learns in 20–25 dimensions, there is no way humans can do that, but there are incredible AI research allowing us to understand the pattern of information that machines extract and I’m very inspired by this idea of using machines decisions and machine consciousness to reconstruct the reality.

I believe archives are very inspiring places where information turns into knowledge, and I think it can be of some interest for a human to witness an AI learning in front of him/her. So in Archive Dreaming (a project developed for Galata tower) I started from information as an input and explored the culture archive letting AI to learn and hallucinate alternative realities. What I saw was that people are learning from this experience because it’s not just a bunch of points on a screen. It’s actually real information; when we go to a library we don’t know what exists there, we have to type in a boring search bar, expect machines to give us something and get inspired; but what happens when you go to a library and you see everything, every piece of information in front of you?

Archive Dreaming | photo courtesy of the artist

From that, my second question was “what happen if a library dreams, architecturally speaking? Or what happens if the building remembers us? What happens if machines become part of the architecture? So, this is how the story unfolded in the last 4–5 years in the studio.

FR: Very, very interesting!

RA: Thank you!

FR: If something, what would you like the audience to feel when they experience your work?

RA: We are constantly connected to a machine that tells our where to eat, what to listen, what to see… so with my work I would like to stimulate a reflection on what else can we do with these machines, with these systems? How can we be more creative? How can we learn better? How can we remember better? How can we dream better? Or how can we imagine and speculate much deeper, smarter and more meaningfully?

FR: This leads perfectly to my next question… we always use science and technology at the service of art or as an inspiration, but can art be at the service of science?

RA: Yes, 100%. I’m a huge fan of Leonardo Da Vinci and last year I got my first lifetime award at the Florence Biennale. I honoured Da Vinci’s soul and mind and I was so inspired by this award because I was researching his life, understanding the movement of Renaissance, try to understand the origins of it, what where the first steps that have been done. If you look at today, science inspires everything, not just art, but art is also very powerful because art for me is humanity capacity for imagination, and if you push this capacity eventually it touches science. For example, when I think about dreaming or a machine dreaming, I only use AI because AI is a scientific tool. It is not just random numbers. When I say that I want to paint with machine consciousness is no longer weird to say, because I can take my brush, dip it into the machines’ mind and paint with machine’s consciousness. It’s not wrong, it’s not weird or fake. It’s just truly the feeling. You really do that. You really take an AI algorithm, you train it and then you use light and paint a building.

Walt Disney Concert Hall — photo courtesy of the artist
WDCH Dreams — process

FR: wow. What an interesting way of saying it and a wonderful visualisation…

RA: Yeah, that’s exactly what it is. But if you said that a couple of years ago, critics where not taking you seriously. There are many people saying “NO” in life when you are pushing ideas, when you try to break the walls. You always feel that there are some people saying “No, it doesn’t work like that”.

FR: So true; in fact what I really like about your work is that you break the barrier of what’s art and what’s not art, showing that you can actually make art with a machine and with algorithms… So I’m curious to know what skills and competences are at play in projects like Archive Dreaming and Machine Hallucination?

RA: These projects are one of the most challenging projects of our studio. First of all, I’m not alone, this is not a one-man game. I have a wonderful team, I’m on the shoulders of giants. We are 12 people, we speak 12 languages, 25 years old is the medium age. We try to make art for anyone, any age and any background. Like we are trying to deconstruct the borders of humanity and create a language that unites everyone. This is a big challenge, because what is the one language that everyone can speak? These projects really use a lot of data, so in the studio I first start with the data, what data means, what it’s important and relevant; then I move to algorithms and think how can we use them; and then of course architecture is always very important for my work. I believe architecture should be beyond concrete, glass or steel. Architecture is unfortunately stuck in gravity and it has to survive in the modern nature, but light is not. Light is one of the most inspiring material in the world; scientifically light is particles and wave length and it can go away if you don’t like it. You can project an idea to anywhere and onto anything. So what we are inspired by the democratic use of light as a material, taking AI as a collaborator and find the collective memories that are important for humanities. Like the latest work “Renaissance Dreams” is inspired by the renaissance archives because renaissance is humanity’s most important movement. As an artist I want to learn renaissance, but I want to learn it with AI so I can understand it better.

Renaissance Dreams | photo courtesy of the artist

FR: I would like to conclude by asking what is the future you are dreaming of and where does artificial intelligence fit in the relationship with humans?

RA: I’ve been involved with AI for the past 9 years, but recently I’m also very involved with quantum mechanic and quantum computation. At the moment we are working at an installation of quantum memories for the National Gallery of Victoria in Melbourne, and we are creating an AI inspired by quantum mechanics that is creating an alternative nature for us. If we are not careful, soon we will lose our nature and then we will be at the mercy of machines to remember for us; so I’m saying “let’s have a look at what machines can remember and see what happens in case we are loosing”. This is our current extensive technology challenge.

FR: Fascinating, and we look forward to it. Thank you for the conversation!

RA: My pleasure.

What does it mean to be human in the era of artificial intelligence?