TIME 100 Talks: Merging Art, Data, and AI with Refik Anadol | Dreamforce 2025

You.

We're going to talk about a lot of things. But before we start. Well, where I wanted to start was with where we are at Dreamforce and Salesforce and that's because you told me earlier that you have a history with Salesforce going back more than 10 years. And so I'd like to start with first just that story because I think it'd be great for everybody to hear that story. But what I'd also like you to talk a little bit about is how those early projects because 10 years in normal human time is 10 years in Silicon Valley it's like 300 years. So in that time how your work has evolved and how does early projects helped you evolve?

Absolutely. So maybe to start with this was like a probably some of you maybe from San Francisco can remember this piece of ride in 350 mission almost 11 years ago. I'm so grateful that when the world was pretty much not aware of AI side of the world, but a data science was just starting. I coined the term data painting 2008 almost when I was an undergrad and I was dreaming one day to be able to find a reason of embedding media arts into architecture. So this project to me so important to our studio in Los Angeles and this was as far as I know the very first world's three dimensional data sculpture.

It's.

It was real time using the wind data and the social media data and the building data almost 11 years ago. So I'm so grateful for Salesforce team of course Mark and everyone and the Kilroy team of the building and the city of San Francisco. I believe it's the first public art program that truly use data here in San Francisco. And that project really opened up so many dialogues. I'm so grateful that I'm just guessing people with coffee walking in the street are the people probably right now writing the next large language models and all the ideas. So I'm so grateful that this project not only let our studio go beyond what we can do, but really promise a project that has a longevity and when the world was questioning if an artist can use data algorithms, machine learning. So this project is very important and literally this project I think also triggered our friends at Google when they were having really time in Downtown and CDPs and invited for the first Arts and AI Conference at Gray Area one year later and I became the first arts in residence at Google AMI program. So that nine years feels like a 90 years, 10 years.

So which is very cool. But so to that evolution and how you use data There's a way in which we're kind of connected because you just helped. We have an annual AI issue at Time and Rafiq very kindly designed the COVID Yes. And so I'd love for you to talk everyone here through the thinking behind it and the very. I'm not going to give it away. The very specific data that you used to create that cover.

Yes. So if you go back like 10 years ago, this project started lots of amazing ideas and so grateful that me and my team in Los Angeles now we are 20 people with my partner Evson also here we are running our studio. We have 15 languages in the studio, 10 countries and try to create art by just using art, science and technology. And last 10 years with AI, we only use our own data and we only train our own models. And it was very different than using just available tools and services. It was about understanding how machine learning algorithms can use from GAN division models and written many models. So we push ourselves to break those tools and understand again. But we always, but always we are in love and respecting archives so much. When I learned that as a possibility. And by the way, dw, one of the wonderful creative director at times who.

Just did his thousandth Time magazine cover.

An incredible like a living archive. And then he was super happy to share the archive. And we were training an AI model called large nature model. Maybe I can show a little bit the video here. So you can see here a couple of things. These are incredible 5000 every single time cover in the history. And what we as we know Time is one of the most incredible archive of humanity from anxieties, happiness, concerns, success, breakthroughs. It's like last hundred years captured. We thought that what will happen if a machine learning algorithm look at all these hundred years and come up with amazingly positive and rightfully we needed concepts, ideas, jobs, solutions to many problems. So the challenge was we have a large nature model which is called LNM, trained on 40 million nature articles. And thanks to Jensen Huang of Nvidia team, last year we trained our first image model from scratch, which is a half billion image archives ethically collected. And we let these two AI models basically have conversation. And what was fascinating is a cover that is transforming humanity's many, many, many important moments. How can we use this moment to bring inspiration, joy and hope? By using the technology that we, I think will all need every single life of our life. So super grateful for the response last night we also showed this on a beautiful media screen. It was fascinating to see people and their responses. And we call it living Cover, by the way, not like a frozen time, but like our future, which is all the time shaping. And we have a chance to shape our future so it's not frozen and we can shape ourselves all together.

And like we were so blown away by it. And we're so grateful that you did this.

Thanks so much.

But that then takes me to where we're going next. Right? And the question of where we're going next, how do you see this kind of art evolving? What's next on your agenda? I know there's a museum in the works, so. So I'd love for you to talk a little bit about that. Like, where are we going? We've come from 10 years ago with Salesforce to that cover 10 years from now. What are you going to be talking about when we look back at now?

Yes. So one more time. So grateful for timecover, because I also remember how many other incredible artists had a chance to touch this data. So it's a really honor and. But with Epson and I have been talking so much about. So I'm so grateful as an artist had a chance to work with moma, Guggenheim, Pompeiu. And we learned that it's actually super hard for museums to be able to implement complex technologies. Projections, media, walls. You just think about complex networks and machines to compute large data. That is very, very rare chance for museums to really both understanding it and also applying it. So we learned that if this medium doesn't go further than we are just pushing forward. We didn't feel right. We said, how can we use our force and energy and friends and families? Imagine the world's first museum of AI Arts. So we are calling it Dataland. And I know the name is kind of little bit funny, but it's actually the idea of a data which I believe the most important language of right now, humanity between machines, humans and environments. We thought that we can maybe create the world's first space by using technology in the architecture, not necessarily adding a beautiful picture or a painting or a sculpture. Imagine the building from the scratch has the engineering infrastructure, media walls and screens and projectors that can allow us to dream from the scratch. So it was a very exciting start point. And we announced in a Frank Gehry building in downtown Los Angeles. So if you're in la, hopefully early next year, we'd love to host you all. What makes airline very different is we thought that this is the building. By the way, this is the new Frank Gehry building next to Disney hall in downtown la, a part of the Cultural corridor. We have moca, Broad Museum and other amazing museums of Los Angeles. So we thought that the museum cannot just start as a normal museum. We thought that we can start the museum with its own AI model. So last two years we worked ethically with half billion images from Smithsonian Natural History Museum and many others. And both Nvidia and Google friends allowed us to train our models on Assassin and Energy on Oregon Cloud Compute. And we were able to create an incredibly exciting system. By the way, we are 20 people, just to let you know, not like a product, service company, we are art and design studio. But we love to code, we love to learn. And because with AI, learning to learn is much easy. It's a perfect time to be by the way artist and a creator and a creative. So we create our own system focused on research, academics and of course for anyone nerding with nature and just want to understand nature much better. And that's the service at the up and running.

So I, I mean this is super fascinating. And so I'm wondering for everybody who visits this museum, and hopefully everybody here will visit the museum, what do you want them to take away from that visit? Like what's it, you're the creator, but from the point of view of the spectators coming in, what do you want them to feel? What do you want them to take away?

So the first thing is as an artist, of course I was so in love with the idea of collaboration. I believe AI is not just a tool, not just another amazing, like you know, a force for us. But I think AI can be an incredible human machine collaboration symbol. And I know that there are many artists and creators want to collaborate with machine. And we are hoping that we would love to bring together people who want to see what could be the ultimate imagination of a machine human collaboration. But I'm not talking about just prompt engineering or context engineering. I'm not talking about just ready tools that can do something. I'm truly thinking about going much deeper, training models together, finding data together, like pushing forward these mediums together. Sounds, sound, image, even smell. By the way, four years ago we started training AI models on scent molecules. And we are also experimenting with what will happen if you not only see and hear, but be able to smell the machine dreams. And by the way, ultimately this is very funny, maybe it's part of this research. So I don't believe artists are inspired by reality. I think like myself, I'm asking always what is beyond reality? And when we ask this question, I think AI is a little bit sometimes limited because all AI services right now is so about reality being like factual, punctual and never want to hallucinate. Everyone is pushing forward the reality. But to me, art happens on the other side when we break the tools and reconstruct them, let them hallucinate dream and fantasy. There's a world there. And I think what Daylan can do maybe entering those dreams and. And hopefully.

So that makes me wonder then on that other side, what does creativity mean today?

I think to me art is humanity is capacity of imagination. And to just close that gap is the creativity is the force of cognitive skills we have. And I believe that doesn't need to be just at conventional art making. Any solution to any problem requires the same creative mindset. And for me, what I found so much fun and inspiring is like working with machines. I can't draw properly, but I learned how to compute drawing like as a, as a, as a programmer and loving the idea of computing. But now with AI like this gets much, much more interesting. I mean even an eye gesture can be a meaning heartbeat, skin conductance, body temperature, weather data, brain signals. We can think anything and everything we can quantify can become a pigment. And that is to me and infinite possibilities.

So you use, I mean this is a perfect segue into. You call AI your collaborator.

Yes.

Instead of a tool. Right. Tell us why, what does that mean to you?

I think the best way for me at least 8 years old get my first computer Commodore. I don't know how much you have also a chance to play with the Commodore age, but it was an amazing day at home getting this early computer, being able to use Basic, like a very basic basic and understanding how to little bit manipulate the game levels and understand the possibility of a change in the mind of a machine. And then based on my mother's memory, apparently after I got the machine the same summer I watched Blade Runner the movie completely changed the perception of media architecture. And an Android gives another Android saying that it's not your memories, it's someone else's memories. I mean what could a machine can do with someone else's memory? Was a very early question. But what I felt so exciting about AI was Player two in the machine as a child. There was someone, something was playing, there was already someone or something was responding. And I think for many people playing very early age games, I believe that has the same feeling what is that playing with me and can I just be a friend with that? And of course sounds funny as a child, but actually that hopeful and a positive response is now with AI not so Different.

So you're playing with it?

I think so. It's a kind of a playful thinking and imagination. But I will say the best example I will give maybe a project that we are so grateful two years ago. This is a piece at the Museum of Modern Arts in New York. It's an honor to imagine very similar time covers as well. We were so fortunate to be able to work with 138,000 artworks of 200 years of every single artwork at MoMA Archive. It was a fascinating place. I know how much you love about MoMA, but as an artist, I think they say you have two options. If you're not alive, the best option is lure Musim in Paris. If you're alive, the best option in life is momo. So I'm so grateful the work was one of those. So what you see here is imagine two machine learning algorithms. One knows exactly every single artwork in the history of MoMA. And one is me as an artist agency taking that data and reconstructing that realities. We use a camera and a microphone and weather data station. The artwork was everyday changing based on the sound of the space and all the movements or the weather conditions. Snowy, windy and rainy days was making the artwork ever changing. V received 3 million people in one year. It was the largest audience at the MoMA history. But what is more meaningful is the work became the first generative AI piece in the MoMA archive. And what was so meaningful was the artwork is going back to the archive. You still keep dreaming. So thanks to Janssen, he donated two DG stations. It's a very heavy compute machine and it's a more than 8K screen. It's a really high end thinking required for compute. But it was fascinating to see with a 38 minutes average viewing experience. So what I see here is not just an artwork or an AI does something. It's true human machine collaboration. And this is, I think a good example of showing other artists that is a possibility of pushing forward the medium. So I. I saw it.

It's. It's extraordinary. Again, thanks very much. But it also raises a question to me at least, which is with something like this, when do you know that a work is finished?

So this is an incredible question because to me it's a living painting. I don't think there's anymore the beginning or end of a decisive moment. Like in photography, we remember that the moment is the decisive moment of the artist or the photographer or in the painting. There's a moment of a frozen pigment. But I believe the future we are Going needs a new kind of a name for it. Maybe this is what I'm calling it, generative reality. It is all the time in change, in flux. It's all the time like life ever changing, adapting and responding. I know there's like a protest in the museum. It was a different day in the early morning when the students are coming, a different day. I believe it's a new kind of a communication with an alien language. I think that is creating a new type of understanding of a new world.

And then what about questions of authenticity? Because they must come up, right?

Yes. And what was more amazing to me. So if you think about a museum archive, we have photography, videography, painting, sculpture, even this archive has a Pacman and Mario, by the way, the games in this amazing archive. But what will happen was unsupervised the name of the exhibition. We asked the question what happens if AI doesn't need to worry our ways of categorization of art? What will happen if there is no those categories and the machines start dreaming? And that was, I think, much inspiring. Then at the current like division models, we had to type something. Let's find something, let's finding becomes something versus what happened. It doesn't exist. And let machine to generate these latent space outputs. I believe authenticity there is almost infinite.

So since we're talking about a museum, before we move on to something else, I want to talk to you a little bit about your view on the accessibility of all of these tools. Right. And it goes to the larger question of just people being able to access the tools they need to create great works of art, to get them into these kinds of spaces. How. How do you think about that? And you've been advocating a lot for it as well, so.

So first of all, one of the biggest challenge of the future is how to create ticking brushes for artists. When I say artists, I mean of course the current AI models are exciting, midjourney stability, AI and so forth. But. But what I think is realistically very interesting is they're amazing, but they are the breakthroughs. The breakthroughs is very challenge their self to make another breakthroughs. But there's a very low chance that someone can create a breakthrough by using what is commercially available. I do believe the laboratory, the kitchen, the place where the artists can train their models and work with them, is where the magic may happen. But at the moment, we don't have any place for it. We don't know where to go for this idea. We don't know where an artist can bring a million images and where do we go? Where do we train these models? And that's what I think where we want to solve the problem through the dataland, our AI museum. What will happen if a person come with a an incredible data sets anything and everything and then how we can use it. And the other thing is I'm so excited, which I hope we can achieve in near future, is if AI means anything and everything, I think it has to be for anyone and everyone. And when I say it, sometimes people think we here in the United States sometimes privilege and sometimes in Europe privilege. But we are working very deeply in different parts of the world from Amazonia in rainforests to the deep Africa and many places where and one AI model subscription is beyond someone's accessibility. What we are planning with our living encyclopedia, Live Nature model is a place for people who can still have an access and have a way of like interacting with it. And I think if we can think like this, I hope it will make this much more exciting platform without like being really too much worried about this, this currency. We can definitely predict that there will be very expensive AIs coming soon. We can predict that maybe more than a car price AI cybers will happen most likely in a couple of even I heard that less than 1000 days there is the AGI possibilities. Everyone talking just here, I mean that's so fast. So we as artists, we are the alarm mechanism of humanity. We always ask these questions much deeper and sometimes much earlier. And I believe as an artist I felt that responsibility to be sure that I can give my thinking brush to someone else. And I think that's an important feeling for many creatives to understand, to teach, to share, to collect and also I hope bring everyone together.

Do you think people are receptive to that message?

I'm happy to say at United nations, at Davos last three times, we had a beautiful response. I had a chance to sit with one of the most exciting leaders of Amazonia. For example, all the presence of the countries around Amazonia, we were in a one room discussing why AI is important to protect nature. And by the way, let's just remember that I know hopefully we will solve this energy problem. But let's not underestimate the energy problem. It's incredible to have gigawatts of whatever's in every single month and even announcement. But it's also possible that that idea will eventually make nature somehow broken. There is no single way nature is ready for it. But on the other hand, I believe nature is the most inspiring thing we have. Nature is the most Important thing we have for some reason we forget that distance. And I believe that these artworks, these ideas, these concepts, these models may just remind us how beautiful nature is. But not just hiking in a beautiful park, not just having a beautiful time with a family, but deeply and understand the idea of nature. Perhaps we already have an AI called Nature, which we forget to maybe understand what will happen if we are already living inside another AI and we somehow forget that. And our egocentric desires of innovation and discoveries somehow left it behind. So I just want to be sure that as many artists, we don't want to like lose this connection.

So you talked a little bit about what you feel sort of your responsibility as an artist passing the brush along. But of course when it comes to using technology such as this, there are all kinds of other costs. And yes, how do you see your responsibility as a pioneer in this world who is in a way propagating a certain kind of use, a certain kind of new way of doing things with this.

It's incredible question. So again we decided to open an AI foundation. We have a nonprofit AI foundation specifically trying to focus on education, learning and practicing these disciplines and try to share this. And when I. I mean I am so grateful. I have multimillion social media followers. And as soon as AI takes different shapes, believe me, the inbox is very interesting place. It can be half very, very, very bright. It can be very, very, very dark thoughts at the same time. So you can wake up with a beautiful message. You can wake up with a very very concerning mind. So what this teach me that there is no way, there will be only one way of doing that. But what is the right way? That from our hearts that we believe that we should be contributing. And I hope that this type of actions in the next months and years, the more artists creative initiatives take attention, I do believe we'll have a lot of success stories behind the scenes. Two months ago with football player Messi, we were very fortunate in New York auction his best memory in his life. So he donated his data, his heartbeat and his body signals and we turned it into a UNICEF auction and it was sold for almost $2 million. And that's the type of support we try to give back to UNICEF to places put our work in places where we can bring value back. So we are finding those type of moments.

So we're in the last five, six minutes. I'd love to ask you to kind of pick up your crystal ball. And you've talked about how you think of AI as a collaborator and where this art has Come. Where's it going? Like, how do you see this evolving? What are you seeing that we're not seeing?

I think there's three exciting parts at least in the studio. As an artist, like working with these tools. Maybe I can show another project right here. This literally two days ago, it was a beautiful building in Spain, in Bilbao. It's Guggenheim Bilbao museum. So what we are seeing here, the architect Frank Gehry, I think a living legend, a truly a pioneer of an architectural forms. He donated all his life data to our AI model. So why this project is to me so important? Thirty years ago, when Frank Gehry invented this beautiful architecture, Guggenheim Bilbao, it transformed the entire city. And we were able to work with him and with with his incredible data and let AI dream a new world based on what you imagine. So because we have this nature model, ethically trained now, we are bringing other incredible minds and souls like Frank Gehry, like others. And what was amazing, the Museum received almost 1 million people in just nine months, which is an incredibly exciting improvement of the society and the museum and bringing new art forms into like, by the way, this was the first digital artwork in 30 years in Guggenheim Bilbao. So in 30 years. So imagine how exciting the change may happen. And on the other hand, the AI should not be about deleting history or mushing history. It should be about defining our heroes, pioneers, teachers such as Frank Gehry, Gaudi, Rumi, Mozart, depending on people we work with. We just want to be sure that our work is highlighting these incredible people and highlighting the archives and importance and make them remembered through as high end technology can be.

So you used the phrase just now, which you've used a number of times today, ethically trained.

Yes.

What does that mean to you?

Permission. And permission is a very challenging thing. Field of ethics. And it's very, very hard for many giants to go back and find, opt in, opt out, completely complex. But for artists, I know that we don't want to touch a tool that is unknown what is inside. We want to work with a tool that we know exactly what is inside. And that's one of the reasons we work so hard that maybe nature, the most beautiful, the most important thing we have, maybe we have it there as a base, no other artist but nature, and try to blend the world together. So I felt that this is a very important, important topic for many artists, many creatives, just not artists, any creatives. And I think this brings you a safe, secure place. By the way, in all our exhibitions we openly share where data comes from, what Algorithms we use why and how, by the way, this exhibition, for example, is using only four iPhone charging energy in nine months to compute AI in Netherlands, we found a renewable energy cloud compute and we were able to run the exhibition last nine months and every day hundreds of images computed in real time. So there are lots of ways of doing it. If someone really deep dives and asks the pioneers to write like a right questions, it is very possible to change this dialogue to somewhere else. And ethics also to me that how artists should be bravely, honestly, openly, from heart can tell this and make them happen.

So just as a final question then, about the next generation of artists, how do you see them engaging in this world? And actually also how would you wish for them to engage? So if you could set a rule.

So 10 years I am teaching at UCLA Design Media Arts Department. A fascinating journey. I can clearly see what happened first five years, what happened? Like, you know, just AI coming pandemic, unfortunately. And the large language models like these, all incredible to witness. I mean, I'm so grateful that I witnessed the birth of Internet Web one, web two, Web three, AI, quantum computation. It's fascinating. Time to be, first of all, a creative. The good news is many of the artists that are just practicing want to learn so quickly. But AI has a tendency of being lazy. And the more information comes to us instantly and constantly, there is a high chance of a form of fatigue. So what we are trying in our UCLA classes, being sure that the right question, sure, something important, but accessing the answer and applying the answer to life and practice is still being human. And I believe that is the part that many young artists right now find out. How can I go beyond prompting? How can I go beyond? Because we know that the same question may be answered everyone in this room. The same question may be applied to everyone here. And then where is the authenticity? Where is the creativity? Where is the inspiration with the tools that are locked down, Right? So there's a lot of excitement about how to ask the question. But still, I think learning to learn is the fundamental life challenge.

So like a professor, work harder is your message. Well, on that note, we're out of time.

Thank you so much.

Thank you so much.

Thank you very much.