by Eileen Isagon Skyers
I caught up with Moscow-born, Germany-based artist Olia Lialina about interfaces and internet tendencies ahead of her upcoming project presentation with collaborator Mike Tyka, a Google Software Engineer. Lialina’s work has engaged with the web since the days of HTML frames. This conversation accompanies the 9th edition of Rhizome’s Seven On Seven conference, which convenes leading artists and technologists for high-level collaborations. View the full list of participants here.
Eileen Isagon Skyers: I’ve spoken with you at length about your practice before, but one thing I am still very curious about is this: what is it like to be an animated GIF model? What does that entail?
Olia Lialina: Apart from looking good and recognizable (as a human), even at a low resolution, apart from finding scenarios for the perfect loops, apart from creating them…being a GIF model is being an “ambassador of transparency.” I mean, a big part of my occupation is to remind others that the transparent background, not animation, is the real treasure of the format: to make the world aware of the historical importance of the transparent background that empowered early web users to create complex web pages out of simple graphics.
EIS: The free web hosting service GeoCities ceased to exist in 2009, but you’ve archived a collection of nearly a terabyte of its data, co-founding a research institute on the subject. I noticed that you’ve also interviewed one of the web page creators, and a Blingee aficionado. Would you say that you’re also weighing the cultural significance of user activity on the early web?
OL: I do. The problem is that there is a resistance. Sadly, it is coming, not only from the corporate and design world, but from the users themselves. People don’t believe that what they did, or sometimes still do, on the web is of any significance. There could be much more interviews, if people I want to talk to about their pages or collages would believe that I am truly interested.
EIS: There is a point of entry for any given interface. Once the user arrives, there are specific steps that the domain will want the user to take. What does it mean to eschew the seamless, corporate aesthetic being presented to us, in favor of a more conspicuous one?
OL: I think resistance can exist on many levels, from advocating for a “visible interface,” and developing them, to little, day to day activities that would contradict the logic of invisible routine. Like never clicking “heart” on Twitter, never saying “always” to any offer an app makes to you…
EIS: And do you think your Seven on Seven project will engage these modes of thinking?
OL: Our project is centered around a modern digital folklore legend—Blingee.com. It highlights the interface of that service, as well as role of the users in reshaping it; the means by which vernacular can be formed and perceived. And of course, the role of glitter, bling, and sparkle in the post-factual.
EIS: Our narcissism with respect to seeing ourselves, and our interests or hobbies, reflected online hasn’t really ceased to exist. It has only seen a migration from the personal home page to the social media profile, and perhaps, now, the multimedia photo-sharing application. What do you anticipate might be the next migration?
OL: Instead of answering your question, please allow me to comment on it. Migration from personal home pages to profiles, was not us moving somewhere with our narcissism. It was, rather, substituting the web user’s interest in building cyberspace, with comfortable channels for narcissism.
EIS: That’s an interesting way to look at it, as if it were handed to us as the only viable option, so of course it became pervasive. What are your thoughts about Google’s DeepDream—the artificial neural network that it uses to enhance image patterns? Do you think that it generates art?
OL: I love DeepDream as phenomena. It was a big moment for AI and ML, not because it is the most advanced algorithm, but mostly because it gave visibility to ML, it attracted attention aesthetically, and sort of self explained how it works. Well, it’s still not clear how, but what is clear is that ML eats data and it is hungry. So the question is, what data sets are fed to it? And who decides? DeepDream motivated me to organize the How Deep is Your Dream lecture series, where the excitement and critique of algorithmic power were discussed. And now, for Seven on Seven, I got the chance to collaborate with one of the DeepDream developers, Mike Tyka. I am so happy he has respect and interest in my type of data sets.
EIS: Did you ever think it might intersect with the net art vernacular?
OL: It already does, on different levels. AI and ML are at the center of net art’s attention at the moment, as a tool, as a concept, and as content. Networks of algorithmic power couldn’t be ignored by net artists.
by Eileen Isagon Skyers
This interview accompanies Rhizome’s 9th edition of its Seven On Seven conference, which invites pairs of leading artists and technologists to ideate collaborative projects and dialogues. I talked over artificial neural networks with Mike Tyka, who creates protein sculptures and DeepDream imagery, in addition to his work as a software engineer for Google’s Artists and Machine Intelligence program. View the full list of Seven On Seven participants here.
Eileen Isagon Skyers: I’d love for you to tell me a little bit about your background to start. How did you transition from your role as a biochemist, to now, working with Google’s AMI program? And how did your sculpture practice fit into that trajectory?
Mike Tyka: I always loved science, even as a kid. My parents were both doctors so, for example, I remember reading their anatomy books when I was little. I loved the beautiful hand drawn pictures of various anatomical dissections.
I taught myself to code when I was around twelve or so, after I finally convinced my dad to buy a PC. I started with QBasic, then discovered Pascal and Assembler, later C. I loved graphical effects and I got interested in the demo scene, a form of art, though I didn’t realize it at the time. I was mostly met with “cool, but what is it good for?” when showing my graphical effects to my parents.
I wasn’t sure what to study at undergrad, but I decided that I’d rather learn a classical science than computer science. I felt like the computer would always just be a tool to do something else, so it should be the something else that I study. So, I did undergrad in biochemistry in Bristol, UK and then a PhD, but by then my interest in computer science came back and I worked in computational biochemistry, specifically protein folding.
A postdoc position brought me to Seattle. Here, I met a bunch of people in the Burning Man scene who I ended up working with to create Groovik’s cube. Burning Man is accessible to anyone wanting to dabble in installation art, without the hurdle of gallery curation or anything, so that was a great way to try out if that was something for me. Turns out, I really enjoyed the process, so I started taking art making more seriously in my life. I founded a small art studio/maker space in Seattle and began making other work, such as the protein sculptures.
Having worked for more than a decade in protein folding, I was quite familiar of the beautiful folds that proteins have, and I wanted to make art that made this beauty accessible to others.
Now, I’m working at Google on a connectomics (neuroscience) project, and also working with artificial neural networks. Of course I see the artistic potential in everything, so I got interested in working with artificial neural networks to create art. DeepDream is an example of those efforts. After DeepDream, we started thinking about how rapidly advancing neural network technology could be used by artists, and that was the start of AMI.
EIS: Do you feel that artificial intelligence should be looked at from the point of view of biology, or the point of view of computer science? Do you have strong feelings about it one way or the other?
MT: Definitely computer science. To me, it has nothing to do with biology per se. Philosophically I’m a computationalist; I believe the brain is a kind of computer (albeit not a classical computer in the Turing machine sense), and that the hardware implementation is irrelevant. What matters is the information processing it does, so, in principle, it doesn’t matter whether that’s done using biological neurons, or silicon, or wooden tinker toys. Though some may be more practical than others, of course.
EIS: You’ve seen artistic potential through both the microscope, and the computer screen, using Google’s artificial neural network, Deep Dream, to generate new images. If a machine creates an image that looks like art, can we realistically call that creativity? Or is it just a switch between different modes of operation?
MT: People struggle to strictly define creativity even for humans, so obviously we have a hard time when it comes to machines. Humans are, apparently, incredibly anthropocentric and insecure in their outlook. We believed we were the center of the universe, in the only solar system with planets, etc., and one by one we had to cede the center-illusion to the truth, which is that we’re not all that special.
Creativity is no different, and many will be very reluctant to cede even the possibility that a machine could be creative. Though, on the other hand, people have been trying to build machines that can create original content for millennia, so we’re also kind of fascinated by this.
Recently, when AlphaGo played, and defeated, Lee Sedol in Go, many Go experts described some of the moves it came up with as beautiful, or ingenious. Those are words typically associated with creativity. This to me indicates that when the machine becomes complex enough that we struggle to understand why it made certain choices, but we find ourselves marvelling at, or resonating with, the results, we’ll begin the process of accepting that the machine was “creative.”
Now, I don’t think we’re quite there yet with art. Things are still very rudimentary, but much progress is being made I think, and this is a question we’ll increasingly have to grapple with.
EIS: So you used to operate a blog at the URL beautifulproteins.blogspot.com. The last entry, dated in 2012, begins with the phrase “I haven’t posted for a while.” Were you aware that your collaborator, Olia Lialina, actually collects screencaptures of blogs, just like these, where the owners of abandoned sites tend to apologize for their absence? The piece is called: Give me time/This page is no more (2015).
MT: That’s cool! I didn’t realize that. Yeah, that blog project eventually turned into my sculptural work.
EIS: Thinking along these lines, do you think that data would cease to exist, absent the curious individuals who inform it? In other words, does the network collect or produce any information without the presence of users?
MT: Which network do you mean, the internet, or a neural network ? There isn’t really “the” neural network. A neural network is just a program, operated by a human. There are many. None of the current algorithms have anything resembling internal motivation and it would not arise spontaneously. But of course people also build autonomous internet bots for example, which operate continuously on their own.
EIS: I suppose that answers my question either way.