Personified data in a post-human utopia with Mario Klingemann

Interview by Ally Ferraro and Guilherme Ferrari

Introduction and Writing by Matthew Burgos

The fertility of the portraits painted by the art masters in Western and European history supplied the faces Mario Klingemann fed into his machine for Memories of Passersby I, one that generated images of non-existing profiles and would continue to do so for a lifetime. Based on the collection of neuro-networks, the machine projects a portrait that eventually shifts, fades, and disappears to usher in a new one, a testament to Mario’s reverence for artificial intelligence, deep learning, generative and evolutionary art, glitch art, data classification, and visualization or robotic installations.

Personified data in a post-human utopia. The generation of artificial intelligence as the harbinger of the future. Not a precise fortune to tell upon the purity of the artificiality, but the guardians of its intelligence trespass the research to uncover its potential as a personal and cultural - and perhaps, spiritual - booster. With French Fries Magazine, Mario powers through the works he has done so far, his perception about accumulating data for artificial intelligence, the potentials and threats the future of technology may bring, and his desire to understand, question, and subvert the inner workings of systems of any kind.

Mario Klingemann: Thanks for wondering what I do.

French Fries Magazine: Your art and messages are amazing! We love your work, and we’ve been following you for a long time now. Your works are fascinating and very futuristic, and your ideas just blow our minds! I think you are very passionate about new things in the future.

Mario: I try my best, thank you!

FFM: We saw your Instagram and wondered what a neurographer does. We found out that it's putting art into artificial intelligence! What did you do before that led you into doing this?

Mario: In a way, I'm always trying to find new things, particularly visuals, audio, and texts, and I have always used computers for that since I was a teenager. I still have one of my first computers from the 80s, the Commodore 64. I haven't run it for a long time, but I can't throw anything away. So, I've always used computers as my creative tool and learned to program quite early, and I was always trying to find ways to be creative with coding and then, later on, with artificial intelligence. It was my attempt to build a system and free myself from limitations.

I'd also done other things before I called myself an artist. I was working at an advertising agency, I did motion graphics for TV, I laid out newspapers and magazines. All the things you have to do to make a living with your art, which is always the tricky part. Ideally, you just want to be wanted for the things you want to do, but then for a long time, you have to do something to pay the rent. In the 90s and early 2000s, I did a lot of flash - I don't know if you still remember flash, but now everybody hates it - and you could be very creative there too. Then, lots of the things I did then were for commercial projects, trying to break boundaries, and let’s remember the times when websites were totally unusable because you had all the wonderful animations, buttons, and other elements, but now everything is very clean, but also a little bit boring.

FFM: That’s right. I feel that you are also one of those who are predicting our future.

Mario: That's always the thing. By predicting it, you also have to make it, so it's a tricky thing. You open the door and build things, and then they sneak their way into our daily lives and inspire people to do that too. From there, it spreads and that's the fun part. The frustrating part is because the moment you make something, untested and new, it's yours, but after a few years, everybody is doing it and it becomes normal. When that happens, I have to move on and find something else to do because I always try to escape what's normal. I'm always a fugitive from things that are easy to do, so I like it hard. I like difficult, complicated things.

FFM: We’re wondering about what you think of AI as a potential threat to us? Considering the data that we give out as users.

Mario: For me, that’s not a threat. Maybe with AI you can never capture your full true self, but you can get to the point where something of you is not just a static package. Generally, the question of whether AI is a threat or not depends on where you derive happiness from. For instance, AI will be able to do a lot that so far only humans can do, or where we also pull out joy from being creative which is a joyful experience. So now, if you spend months or years on something and you are happy with it, then somebody else has just to press a button and in the eye of the public that’s good, that can be frustrating, right?

Sometimes you might want to do something, but you are lacking certain talents or capabilities and just need a step forward to get to the next step, and there, the AI can help you. For example, if you want to write a book but feel it's a bit frightening to, then maybe you can write a book together with an AI and develop it together. In this way, you have this partner where you still lay the path and say what you want to do, but the AI fills in the gaps for you. You can still decide ‘I don't like that style, AI. Can you give me another option?’ as if you were working with some assistant. I see it as you working with humans and they might not do exactly what you want to do, but you try to explain, then they do something and you react and in the end, you make something. So, there's the AI which has certain capabilities or knowledge that we as a person might not have but then again it has no motivation of doing anything or it’s sitting there, so you have to make it move; you have to tell it what you want it to do. Then, it reacts and gives you something back that might not be exactly what you want, so it could be better or it could be worse. You have to adapt and give better instructions or work with it and recombine it.

I like this process a lot as it gives us more time to focus on the core parts that are still important in creativity, which is also a process that takes a lot of time for things that you wish you could just fast forward. Again, sometimes that process might also be good because, in the process, you may discover something. AI is not the only option, but an option, and you have to decide if you want to pick it or if you have other methods to work with that are better and more helpful for you.

FFM: It's definitely helpful. We remember Time Magazine printed the cover Jesus Online in the 90s, which for us resembles cloning ourselves in AI, in data. Isn’t that a premonition of the possibility of AI killing a ‘real’ part of our lives, or this becoming a religion?

Mario: It could be. At the moment, it's not good enough yet. For example, it would’ve been a very lossy medium if you’d recorded some gramophone record of your voice 100 years ago, and listening to it today would just be muffled and scratchy voice. Even though you don’t get the full picture of that person, AI would probably be able to get at least a pretty good impression of who you are. It becoming a cult is another question. Some people might want to turn it into one, imposing an entity over it to put their hopes and fears in even though we don't fully understand it. My piece Appropriate Response - the one with the prayer stool where viewers kneel in front of a display - is going for that notion that AI is very powerful which may also talk in mysteries, for example in oracles. We’re still interpreting what the AI is giving us, and everybody might interpret it in different ways. I still think it’s based on technology and does not have all the hallmarks of a potential religion, but then we never know, so I'm not that worried it will turn into a cult because there are still too many intelligent people with the right ideas but, of course, it becomes more accessible, so some people might want to transform it into one. I’m sure I won’t be the cult leader for that.

FFM: When we saw your project Memories of Passersby I - the machine you’d used created human-like faces - we felt that those faces may have been derived from data, a symbol that machines may create us, who we are, our culture, and even our personalities. What information did you put in there to generate such work?

Mario: Well, Memories of Passersby I is a closed system and will never grow. After I put it into GAN, into the computer, it's static there. It can still freely move and will never repeat, but it’s actually the opposite of me. What you can expect from me is that next year, I will do something different than this year, and this piece in particular will never be the same since it is limited; it's a snapshot of time. So, this piece, for example, is a marker of AI, of 2018, and once we look back in 20 years or so, we will clearly revisit this again because it's 20 years of technology will be so much further than what it looks now. This will almost be nostalgic, the same way we feel about vinyl records or cassette tapes.

Once we have more defined faces, we can do much more since the deep learning space and its developments move so fast. It's hard to try and make something that can keep up with time. It’s also nice if people remember me for this. If it's something recognizable, then I have done a part of my work as an artist. Hundreds of thousands of artists do faces and portraits. It's quite hard to find something that has not been done or does not remind you of twenty other artists, so that's always a tricky part.

FFM: We’ve also read about your work for Art Basel, where your AI-based installation is going to do the portraits of the people attending the exhibit…

Mario: Unfortunately, I can't be there, but this piece that I will show in Basel will be interestingly a combination of several other pieces I've done before because of the way I built it. I have these certain models which create faces and another one interacts with cameras, and so there are several pieces I made in the past which I almost took parts out and recombine them. I also retrained the model with something new, so in the end, what we get is the feedback loop that is similar to the way Memories of Passersby I works. In this case, the feedback loop is not closed, but through the camera image, which is a similar part I used in Uncanny Mirror where I already have this mirror piece.

I made six different models that are trained, and the faces in this case are synthetic. What it does is it looks at your face but reduces it to very simple markers. It sees where your nose and eyes are, then feeds them into the loop and combines them there. We have streams where sometimes your face pops in, but it looks different. You can see when you move that it’s actually you. It's quite interesting that you sometimes recognize parts of yourself through the proportions. You are influencing the system the same way as, for example, there’s a river and when you put your hand in there, the water flows around your hand and there's a bit of spilling. In a way, that's what you do. There's this constant stream of data flowing, but you interrupt it by stepping in and changing it, and when you step away, the water first has to calm down again to see the faces, but the information that you injected into the flow is there now.

Of course, over time it will dilute, and then maybe after 10 minutes, not much of you is in there anymore. It's the same way with everybody who's writing a tweet on social media might have a chance to be that butterfly, that one who causes a storm, so I guess that's the idea of it. Also, it’s for people to just enjoy themselves and have a bit of a playful time, and maybe then can get an NFT of the portrait - that’s the other new thing now, right? - and they can mint their portrait, take it home, and sell it even!

FFM: That’s another interesting thing that we've noticed, that it might be scary that machines can copy the reality - this concept of deep fake, for example - and you don't know what's real and what’s not. We were thinking that maybe machines can also fake emotions that are based on data as well, allowing the machine to have this sense of understanding.

Mario: Right! Now, imagine you’re reading a romantic novel and you feel empathy for the characters at some point, if it's well-written. I mean, the emotion transported by the texts, by telling a good story, or maybe by watching a movie, you definitely will feel emotions. So, all that is just arranging data in the right way, and why couldn't the machine copy those approaches? Knowing the ways how to trigger our certain reactions is a type of information that goes into the machine that’s already happening now. People might have a lot of fun with it, or they will be shocked, so it's already creating emotions.

I don't think there is any magic component to that. In the end, we all interact through our senses with the world. There are some senses that AI is not good at yet meaning touch or smell, but this is mostly technical limitations. In the end, what counts is the essence of what is being said or being expressed, and machines are already pretty good at detecting those. The machines can look at an image now and tell you pretty accurately what emotions are being evoked, empathy, for example. Let's say some nasty guy killing a puppy. That will cause to radiate strong emotions, and machines can already tell the emotions that are very likely to happen.

Maybe right now, there is still not enough fine-tuning, which is what humans are good at, but it isn’t a point where we can fully drop our beliefs. It’s also a matter of improving the models, which isn’t really a technical hurdle except for memory space or computing power.

FFM: I wanted to talk to you about your collaboration with Nick Knight for Harrods.

Mario: We’ve collaborated twice. He came across my early GAN works and then got in touch and said ‘I like what you are doing; do you want to do something together?’ After that, we talked about what we could do together. So, he did a fashion shooting, sent me the images, and then I started training on his material. It's funny because, at that time, I was trying to get more realism and not get certain artifacts, so at that moment, my aim was to work on totally different things, but looking back now, of course, the artifacts and the weird AI stuff is exactly what makes them interesting

Recently, we did another collaboration where he had these Harrods windows and he asked me if I could contribute something. I did neural transformations on his collection of runway photography. For fashion aficionados, I think it was great for them, but for me, it was sort of runway model audience right having a strict format, so I used the same model I had trained for a totally different thing that is now running in a piece for Bosch, so in the end, that model can look at anything the results were quite interesting since it misinterprets the content. Don’t get me wrong, I’ve always liked that! The machine assumes there's something coming in, then I give it something else and it still does something with it even though that’s not what I expect.

FFM: It sounds very spontaneous.

Mario: Yes, and it’s playful! I always also like just to see what happens if I feed the machine something, right here and now, and the beauty is that these models are very forgiving so they don't crash. You can fit something in, and you will get something out. I mean, most of the time it's noise and that’s not very interesting, but sometimes… it's a misunderstanding, right? Good jokes you mishear what somebody says, but you have a different interpretation of it and give it a new meaning. It’s the same for these models: you give them something they mishear, and then they give you something back, one that’s not what you expected but still interesting.

FFM: And there, you have this element of fascination that is so omnipresent there. We’re looking forward to more!

FF MagazineFebruary 22, 2022