I’m in Dallas — or more accurately, Richardson, a silicon suburb north of the city — to meet with David Hanson, a maker of realistic (i.e. human) looking robots.
We’re collaborating on a piece that, if all goes well, will be part of a group show at The Museo Nacional Centro de Arte Reina Sofía in Madrid this summer. Some time ago, the curators invited me to be part of a tech-oriented art show, and I suggested approaching Hanson to make a singing robot for which I would write and record a song.
Hanson’s robots flirt with the uncanny and test our notions of what it means to be human. They have rubbery flesh made of what he calls frubber, with tiny wires on the inside that pull the “skin” to mimic human facial expressions (to an extent). Some of them can also make eye contact and some can carry on a weird dialogue, adding to their profoundly disturbing nature. Part of what makes this human likeness so creepy is our instinctive desire to empathize with the robots and to ascribe to their behavior human motivations and even emotions.
As a result, Hanson’s machines make us wonder how much of our interaction with our fellow humans (and animals) is based on instinctual empathy. We believe that behind the actions, words and facial expressions of the people and animals we encounter there is a life force and a consciousness. But the robots force us to ask how much of that is presumption on our part.
I was curious whether a singing robot might push these reactions even further. We often assume that singing is “from the heart” — or at least some part of it is. I myself believe that it is and it isn’t: it’s both a developed skill (to emote convincingly), and a true outpouring of emotion, as the physiological effect of singing is by nature more connected to the lizard brain than to the rationalizing frontal lobes. The fact that singing can engage both parts of the brain makes it maybe the least likely thing one would expect a robot to do.
There have been other singing robots. For instance, in 2001: A Space Odyssey, HAL sings “Daisy Bell (Bicycle Built for Two)” as he is powered down. The scene demonstrates HAL regressing from independent thought to mere parroting, and was not meant to be a kind of expression of HAL’s feelings.
Before leaving for Dallas, I wrote and recorded a short song in New York, something I believe is passionate, over the top, and extremely emotional sounding. The song is sung a cappella.
Getting down south had its share of setbacks. Due to some McDonald-Douglas planes having untreated technical problems, three successive American Airlines flights were cancelled. The reason was never given at the time. They would say things like “we can’t find a crew.”
I eventually arrived in Texas and drove from the massive Dallas-Fort Worth airport (it’s larger than Manhattan) across the flat plains of northern Texas. The gracefully curving highways were the color of the surrounding earth — a sort of warm beige. After about twenty miles, I turned north on Highway 75 on what might be the mightiest and most awe-inspiring interchange I’ve ever seen. At least five levels of roads are stacked up, all swooping over, under and around each other as if in some mighty concrete mating dance. It’s a truly incredible work, graceful, and of a scale so large that it is impossible to see the whole thing from any one vantage point.
When driving on the upper levels, you are almost completely unaware that you are arcing and swooping and curving in a ballet with all the other vehicles exiting and merging down below. You simply see the curve of the road ahead, and some signs alerting you of approaching merging lanes and future exits.
I ate dinner at the Renaissance Hotel. The restaurant’s only other diner sat off in the distance. I’m currently reading Temple Grandin’s book Animals in Translation, which seems appropriate for this project. She’s a highly functioning autistic person who claims her autism has helped her to better understand and empathize with the animal point of view.
From the door to my hotel room I could see across the atrium to the identical rooms on the other side. The building’s massive scale, warm moody lighting, and repetitive pattern of doors, plantings, and balconies felt more like some very peculiar temple than a place to sleep. Within this strange temple, all individuality is erased, all ego lost, and, as with many religious sites, one experiences transcendence, a sense of being part of something beyond and greater than oneself.
In religious practice, this glimpse of a profound truth would be channeled via word, sound and symbol to join a pre-established system and set of myths. In this case, one wonders where such channeling might lead? To the world of meetings, creative business exchanges and exciting capitalist enterprises?
The next morning I went to Hanson’s studio, located in an office/industrial park called The Telecom Corridor, where corporate headquarters for companies like Texas Instruments, Samsung, Ericsson and AT&T abound. One building — of a new type springing up here and there — houses data archives, so it has no windows. It’s beautiful in a strange way.
It’s like being in a new world where humans are merely visitors. We work to secure the data and take care of it, like worker bees for machines.
I arrive at the studio, which is on a street called West Executive Drive. The studio itself consists of a reception area, a conference room and few workshops. Some workshops are dedicated to hand-sculpting the heads and faces that will be cast in the fleshy frubber that Hanson has invented. Others are littered with servo motors and laptops that tell the partially assembled robots what facial expression (or Viseme, as Hanson says) to display and how to move their heads and arms. It’s a scene from a thousand science fiction movies, which is pretty exciting to actually walk into.
I proposed that I be videotaped singing the short song, enabling Hanson and his crew to study the series of head movements and facial expressions that I instinctively produce when performing. I do a number of takes, some with more movement, some with less. A few of the robots have mechanical arms, so I do a couple performances moving my arms like I normally would when singing. The robot — which doesn’t resemble me, by the way — won’t mimic my particular mannerisms, but will instead render the performance of a typical singer.
I also suggested that I write and record a version in Spanish since the audience in Madrid will be mainly Spanish-speaking. So, while the wire “tendons” are being attached to the inside of the skin, I hole up in the empty conference room and come up with a Spanish verse that might work.
Just as in the English language, some words and phrases can sound strange and awkward in Spanish. Moreover, the use of melody can place the non-native Spanish speaker at even greater risk. For instance, a melodically emphasized syllable effectively accents the vowel, which, in some cases, conjugates the word into the past tense, or changes its meaning entirely.
To be sure I hadn’t inadvertently done any of that, Thomas, one of Hanson’s collaborators, took me to meet a poetry teacher fluent in Spanish at the nearby UT Dallas (a school that originated as a Texas Instruments Research & Development facility). I sang what I had written for her and she helped with a few phrases. Sadly, I discovered that two of the lines didn’t work at all — though most of the others did — leaving me with a little homework to take back to New York. But, once complete, I’ll be able to record the Spanish version, too.
We returned to the studio, and eventually Hanson’s crew of assistants and collaborators drifted off to their homes and day jobs and I went back to the towering atrium by the side of the highway. That night, Kevin — a Hanson collaborator and production manager — and his boyfriend Carter hosted a BBQ at their traditional Dallas bungalow. (The bricks used to make Dallas bungalows and many newer houses as well are the same color as the highways.) The inside of their home is filled with their paintings and artwork, some of which leans against the walls or lies stacked in piles in the various rooms. One painting displays the Kool Aid pitcher man brandishing a bundle of dynamite. Others show a young Shirley Temple making a cute expression and wielding a butcher knife. These latter paintings have been popular so Kevin will paint more of them.
Hanson once did a stint at Disney in LA, which is no surprise since the Magic Kingdom calls on sculpting and molding skills like his for their theme park needs. While there, he began to develop Frubber, but didn’t work out all the problems and get a patent until after he left the Kingdom. I hope they don’t steal the formula; the Disney folks are rumored to be ruthless. Hanson also worked at the LA studio of artist Paul McCarthy. Although McCarthy’s work leans towards the obscene and scatological, the molding and sculpting techniques are probably not much different than those employed at the Disney parks. The crossover between McCarthy and Disney is, I think, significant.
Heather, a former Hanson employee, said over BBQ that she’s heading to LA this week to apply for similar work. There’s a circle of modelers, engineers and artists mixing technology with artier impulses. They float between the art world, theme parks, Hollywood, and high-tech AI conferences.
Hanson went to RISD, the same art school I briefly attended. Early on in his attendance, Hanson wanted to mix his tech and performance interests with more traditional art techniques and skills. But the school strongly discouraged this, since they lacked a department capable of overseeing that kind of work.
I had a similar experience. While working on a photo-based semi-conceptual art project, I was advised to try my luck in NYC instead of trying to fit in to the school programs. “We don’t teach this stuff,” they told me. At the time it was very frustrating, although in retrospect it turned out to be good advice. And at least I managed to acquire some valuable drawing skills while there.
Hanson and I decide that all programming and adjustments for the English song should be complete within two weeks. In the meantime, he’ll send me a video of the work-in-progress. Once I finish the Spanish song, I might revisit Dallas to facilitate programming Julio’s face for this alternate version. The robot’s body should be ready two weeks later, and then we’ll decide whether the arms will move, whether the head will bob in time with the song’s meter, and whether the torso will sway just a little to the (implied) beat. That leaves us a month to solve any additional problems and install in Madrid — the opening is in June.
I have my fingers crossed. If it all works — even forgoing some of the more ambitious ideas — it will be pretty astounding.
Some robot and artificial intelligence labs discourage and frown on work like Hanson’s. According to his detractors, building robots to mimic human appearance and expression is for the most part irrelevant. More significant investigations will explore what actions the bot can accomplish. Can it learn? How does it process sensory information? In what ways can it react to its environment? Hanson’s work may be less academically rigorous, but it does probe at some sensitive areas that traditionally focused engineers and theoreticians might prefer not to think about.