GL: I’ve had some discussions about The Snout (pictured above) which is simply a robot that looks at you. Obviously it’s a computer and the computer’s not living, so I’ve done a whole bunch of work to create the illusion of life. People often make guesses about how sophisticated the model is that I’m using to create the illusion of life. Multi-level emotional state system, machine learning, and things like that. I tell them that with the very explicit admonition that it’s not just that the piece is simple, it’s that the observer is very complex and that the observer is going to impute all this intelligence to the thing that really isn’t there. So, in the case of The Snout, for example, the one thing that it does that is actually sort of tricky is a the geometry that it does so that it always orients toward you. Then, it’s just doing Perlin noise. That’s it. But when you watch it, what happens is that you’ll do do something abrupt, it reacts with a little more speed, it accelerates or decelerates the Perlin noise in response and people say, "Wow! It understands what I’m doing and it’s reacting!" No, you just really feel like it’s alive because you have big chunks of your brain that can’t help but see it as alive. You can’t help but to put emotional weight on it pulling slightly back of coming slightly forward, we are really built to put that weight on everything because the things that matter the very most to us, parents, the tribe, reading intention in movement and faces, are so important that it’s easier to just attach them everywhere rather than miss them in a vital instance. I think that’s interesting because it brings us into the territory of the robotic body, the robot that we know is a robot, but that we ascribe humanity to simply because we can’t help but do that.
I think we really have to talk as much about the human perceptual apparatus as we do about the aesthetics and machine form when we talk about the robotic body. The reason I say that is because if we’re going to see a thing in the world move, big chunks of our reptilian brain are going to say, "That’s alive." Or, the first thing they’re going to say is: "That’s alive," and then it’s only when the higher order processes kick in and say, "Ew! That’s in the uncanny valley, I don’t like that," that then we say, "Ok, well, then what is it? Oh, it’s a robot." So, getting out of the uncanny valley is a really important place to be. We understand what the uncanny valley is, it’s been identified, we agree that it’s there. So, then it’s a matter of taste as to whether we want to be to the left or right side of it. I’m personally interested in being on the left side. I think that’s where you have the realm of cute abstraction, it’s where you have the realm of cartoons, it’s where you have the realm of things that are alive but don’t have to be understood as quadrupeds or bipeds or even as any kind of phylum we know of. It’s where we can invent new species. Making The Snout allowed me to design a character animation system, a procedural one, that was freed from the limitations of having to imitate what a human being might do. On the right side of the uncanny valley so much other stuff comes into play about expectations about cultural modes of gestural communication, body language, appearance. You’re not just dealing with how things move but with the gendered body and the racialized body and the socialized body and the dressed body and with the culturized body that reacts to you with certain degrees of interpersonal space and other kinds of things.
JN: It’s interesting though, that we still expect that a machine senses the same things that we sense though, that it conform to our sense of what it’s supposed to ‘see’ in the world, no matter how unlike us it might be. GL: I was just in London this weekend with the Opto Isolator (pictured above) which is a little eye robot that I made which has a very noisy face-detector that’s kind of unreliable, and it suddenly looks off to the side when it imagines a face. When it does that, it’s really disconcerting because people are like, "There’s nobody there…?" They don’t understand what it’s looking at. So, I think that it’s very important that to the extent that we perceive robots as sensate, we are also able to sense the things that they’re sensing. In terms of our ability to understand what robots are sensing, it’s very important that we have an agreed upon reality. That might just be from human conversation that we inherit that, right? You know, if I’m conversing with you, and you start looking around — JN: you’ll start looking around. GL: … or I’ll think you’re crazy. JN: Right, because I’m a system just like a machine and from a certain abstract, rather cold, point of view, steering a conversation and steering a ship are more or less the same thing. To learn more about Golan Levin's work please visit flong.com. |
|||
