Artist-Engineer Marc Bohlen uses some fairly advanced technology to express his artistic visions. It’s not often you find an artist with a degree from CMU in robotics, or an engineer with an Masters in Art History. Bohlen’s projects explore how people and technology interact, ranging from the bickering robots Amy and Klara, to his latest project, the Glass Bottom Float. In advance of his appearance at the Emerging Technology Conference in March, Bohlen talked to us about how he approaches art, and just what art is.
James Turner: This is James Turner for O’Reily Media. I am speaking today with Marc Bohlen, who seems to collect degrees like some people collect comic books. He has a Bachelors in Electrical Engineering from the University of Colorado, a Masters in Art History from the University of Zürich, a Masters in Robotics from CMU, and a MFA, also from CMU. He’s been a visiting professor in universities from Zürich to California. His work explores the boundaries between Machine Intelligence, technology, art and society. He will be speaking at O’Reily’s Emerging Technology Conference in March. Thank you for taking the time to talk to us.
Marc Bohlen: My pleasure.
JT: So let me begin by asking: do you consider yourself an artist, an engineer, a social commentator or a melange of all of them?
MB: A melange of all of them, but I think artist-engineer is quite precise actually.
JT: What led you to that fusion of art and technology?
MB: Well, I was working in Art History, on Marcel Duchan and Joseph Beuys at the time, trying to figure out how the materials that they used in their work generated meaning. So the traditional art historian methodology just didn’t work anymore. I was forced to start to look into domains of knowledge that were not part of artist textbooks or repertoire. So I wandered off into engineering, trying to solve those problems, and in the process of doing that I jumped into this field which, at the time of the late 80’s and early 90’s, started to formulate itself as an art technology complex, art technology endeavors, and I never looked back since then.
JT: One of your projects, Amy and Klara, that’s very well known, involves two computers who, in your words, have a ‘hissy fit’ with each other. Can you talk a little bit about the technology behind that project and what you wanted to explore?
MB: Okay, well let me start with what I wanted to explore, then I will get onto the technical issues. It’s a inquiry into the humanoid robot design paradigms, and the assumptions we make when systems mimic humans, and there seems to me a bias that we think that we will be able to replicate only the good sides of humanity, in our artificial systems. So I wanted to correct that assumption with a system that does the opposite, that actually imitates some of the worse things that people do. I thought an argument or a fight between robot-like entities would be a good way to warn us, in a way, that the future might actually be nasty as opposed to bright and sunny, by our own design of course.
Now technically what I did was, I built two systems that are exactly the same, equivalent, that have speech recognition and speech from text generation. So what these robots do is go online and they search texts. In particular they read Salon.com, which I find to be a site for a lot of interesting trash. They do small analysis of the text they find there. They go to the same site, they go to the main page and then they compare their results. They do that with statistical processes and they can compare those notes, and while they should be the same they’re not quite because I made small perturbances in the system and so that starts them in sharing some of this data and they do that by verbalizing some of the text segments, so they speak and understand each other. They are speaking as text to speech and this is known technology and the other one is the speech recognition part, which is much less developed, unfortunately, you might know it from dictation systems that really don’t work. Well the interesting thing is that, in my case, I used the failings of speech recognition as a way to jump start the mis-behavior in the sense that a mis-understanding, not understanding what the other robot said actually, generates a kind of bad notion or discord instead of asking what’s said, a curse word is uttered, which leads to another one on the other side. Back and forth and pretty soon they jump over into a pretty nasty rant. So it starts with an actual, real technical problem. It is tweaked in such a way that it ends up something that we might call an argument or robot rant if you want. Then again, the whole idea was to shine a light on these assumptions that we have about where humanoid design will go and make us reconsider those assumptions.
JT: How much of the banter has intelligence behind it, and how much of it is a glorified Eliza level?
MB: There is a part that is Eliza-like in a sense of putting the text segments together, that is fairly simplistic routine, you can compare to things like Eliza. But the overall system, getting the components, was something of a systems engineering challenge. By no means do I claim this to be synthetic or artificial intelligence. I am using a whole bunch of different tools, cobbled together at times, in order to bring something forth that is conceptually part of artificial intelligence, meaning generating human-like intelligence, but bringing it to a different place. So the contribution is not that I am outdoing AI in any way, nothing like that, but rather that I am working on a much lower level, but asking ourselves what are we constructing here. As opposed to just writing a text about it or doing a presentation and wanting to “get the discussion going” I actually build systems that do this because I am an artist-engineer, I guess.
JT: So one of the points you made is that we’ve grown used to nice computers. Do you think overly nice computers lead people to expect nice behaviors, and maybe if our computers weren’t quite so nice, if they were a little ruder, maybe we wouldn’t get so upset when they go wrong?
MB: Thats an interesting thought. I don’t know if the reciprocity can work like that, but I do find it strange that the assumption is basically, if it looks like us, we’re going to feel comfortable with it. Like the vegetarian that says, if it has eyes don’t eat it, I just find that a very curious and shallow way of interfacing with the world. The anthropocentric and anthropomorphic notion of only using what we know from ourselves as the guide line from judging good for bad, and pleasant from unpleasant. It carries over into how we design the artifacts around us. I think computers are a good place to watch this happen as we speak because they are so malleable, they’re universal machines, they can do many things. So these biases creep into how we design things.
JT: We’ve grown used to the use of computers in arts, with CGI for movies and sound editing for records, but normally the goal is to pretend the computer isn’t there at all. To use it as a tool behind the scenes. Your work places the computer right out in the open in the center of the piece. Do you think people are ready to accept technology as a visible component in art?
MB: Sometimes yes and sometimes no. I think when art becomes, let’s say, the canonization of the computer and the art will put it into the background. But the inquires into what the computer, as such, can do for art brings it out into the front a little bit like the famous dictum by Mark Weiser, 1991, that said all technology that is significant disappears into the background. From that we have all these systems that are supposed to generate ambientness for us, computers off the desktop and in the wall, and around you, and so on, and it disappears. When it disappears, you can’t argue with it or negotiate things, you can not turn off the systems anymore. In other words, it becomes a piece of plumbing that you have little rapport with. I think that it’s maybe OK for plumbing as such that doesn’t have the same complexity degree as information systems, but the information systems, I will claim, should be up front because the paradigms they create for us are still so novel and we have not yet come to terms with how to integrate them into our lives. We just know that they change them, and so to have them go into the background, even for the arts, I find that maybe fifty years or so we can discuss that, it’s too early.
JT: In computer graphics theres this point where visualization of people gets real enough that it gets disturbing. I’m wondering for example in The Polar Express, that was one of criticisms, that it had gotten to that point where it was almost creepy. I’m wondering if the work you do, you don’t really try to make them very anthropomorphic. Do you think that you intentionally do that because you don’t want to to approach the human-like look?
MB: That is correct. The term you are referring to is the uncanny valley by Mori, the Japanese roboticist of significant fame. So what he described is that when systems that mimic humans become all too real, but, fail to reach real mimicry, they fail all the more, because the small differences are perceived as significant. You see that in current humanoid design again and again. Yes, I respond to that by trying not to make the systems more perfect, to get on the other side of the valley, which is probably more like a canyon than a valley, rather to question the assumption that lead us to want that. The whole notion of making systems human-like is based on the assumption that human intelligence is the highest form of intelligence, and that the systems should mimic us, in order that in the end they can out-perform us such that they can reach super human intelligence. And that might be true but it is not the only path. I just don’t enjoy the assumption that we can not question that or move around the field. So I try to perturb those assumptions somewhat, with some of these works and its to be seen in the light. So, no, I don’t have intentions towards perfection. I purposely skid away from that, it makes them look or brings them closer to that uncanny valley, and thats why I don’t make them look like people in any way. I mean they are pink boxes, that particular piece, so there is no way that anyone is going to say these are humans. They’re symbolically acting like it, they have a little pink purse, so on and so forth. There’s a tongue and cheek component there, the moment you engage with it, its clear this is going somewhere else and not trying to make you think it is a purse.
JT: A lot of your projects have what I would call a whimsical flavor, such as counting stars out loud, but there’s obviously a lot sophistication and engineering behind them. How much of your work would you classify as computer science or electrical engineering as oppose to artistic expression?
MB: Well that’s hard. I actually trained as a stone mason before I really went to university. There the formula is 10% inspiration, and 90% transpiration or sweat. So with the stones, you work quite hard till you get what you want. In this particular field of art engineering, I think the formula is turning out to be quite similar, that you can sit down and have this amazing idea, and then its gonna take you two to three years until you can make it happen. So yes there is quite a bit of effort involved in transforming these ideas into reality and I don’t see that changing in the foreseeable future. At least not in the kind of work I do, unfortunately, to some degree.
JT: Have you found offshoots or have people come to you with practical applications? Not to say some of your stuff is unpractical, but you know more commercial or mainstream applications of things you’ve done?
MB: Yeah, every now and then it happens. I mean, I work at a university, so our patent office comes by every now and then. I developed a computer vision application that counts the number of people who go into a building or room, by looking at their feet instead of their faces. Because people like to show off their shoes, as opposed to being watched by cameras. So there was a discussion: can we use that, is that going to be commercially viable, and so on and so forth and it turns out then as things get complicated with patent offices, often times they find alternatives that do the job just as well, but of course have a different spirit, I mean, responding to surveillance issues, such that you take people’s desires of not being seen into account does not necessarily have a monetary corresponding value. So you can’t say it’s worth something, in that sense. I think a lot of my work is still too marginal to be of general economic interest, but other work has been seen differently. The one I am working on currently and that I will be presenting at E-Tech is the Glass Bottom Float, which is a robot-like system that tries to evaluate water according to new criteria. And there I am actually working with a beach that is hosting the prototype, and I’ve been approached by others to see if that it would be possible in their locations. So sometimes it happens, but it’s not something I directly seek. It’s more the projects define those opportunities. I mean I am open to them, but I don’t try to actually let them dictate where the work goes.
JT: So I want to end with the classic question that get asked of anybody who works outside what people consider mainstream art, for better or worse. One of the criticisms that you hear about especially modern art, is that sometimes it almost seems like the artist is trying to see how far they can go and still be classified as art. Things like having an installation, which is people walking through a room. Do you think there is a reasonable boundary of what art is?
MB: Well Duchamp said that the audience tells us what art is, in the sense that there’s a consensus that builds itself over time, whether something is on this side of the fence or that side of the fence. That said, I don’t think that all of the experiments are equally interesting. I have seen and heard of some that push that boundary, but don’t push in an interesting way, whereas others push it and do so in an interesting way. So I think we just need additional criteria, is it not only art or not art, but is it good, is it going to persist in such a way that 20 years from now are we’ll still want to worry about it in a particular way. If you think early Duchamp work, or Beuys’ or Acconci’s or Conrad’s early work, these things still function as radical pieces, because they were really good, and that’s certainly not the case with a lot of things that are happening now. You can’t generalize this and you have to look at that on a case by case basis. I hope the experiments I’m making will have that kind of power, but that is not up to me to define.
JT: We have been speaking today with Marc Bohlen, Artist Engineer or Engineer Artist who will be speaking at O’Reily’s Emerging Technology Conference in March, describing his work with his project, The Glass Bottom Float. We’d like to thank you for taking the time to speak to us.
MB: My pleasure.