2.23.05 -- For an academic, Natalie Jeremijenko wears an unusual mantle: “artist-experimenter.” Formerly a member of the engineering faculty at Yale, she was drawn to the California Institute for Telecommunications and Information Technology and an assistant professorship in the Visual Arts department at UCSD last fall because of the opportunity to conduct research where computer science and engineering meet art and social practice.
“This is an area where Calit2 expects to make significant contributions,” says Calit2 director Larry Smarr, “and Natalie is the sort of visionary and innovator to show us the way.”
Jeremijenko is intrigued by, and studies, new technologies because they represent the opportunity for social change. “But is that the same thing as social progress?” she asks, citing the fact that micromechanical processes are driving costs down but adding toxic gases and other pollutants to the environment. “What does it mean when you claim that ‘cost is reduced’?” she asks, underscoring her view of the market as an imperfect representation of the effects of social and technological systems.
In many of her projects, Jeremijenko focuses on toys as the “lingua franca” of society. Toys are great investigative territory, she says, “because they come with the expectation that we learn from them. But what do we really learn?”
She cites the new generation of robotic dog toys powered by artificial intelligence systems: “People think they’re fun because they’re a concrete manifestation of science fiction. Popular thinking says they’re something new. They’re no longer toys with on/off switches but rather intelligent, autonomous agents with ‘behaviors.’”
But how do such gadgets see us? No one had ever asked until Jeremijenko undertook a study of people interacting with various types of gadgets. She recorded hours and hours of video and says, “The diversity of reaction is extraordinary.”
In an empirical study of the Furby manufactured in the late 1990s, for example, she embedded cameras in them to see how people would react to them. This study led to some conclusions that confound conventional wisdom.
Furbies were promoted by the manufacturer as companions for the elderly. The notion was that the toy would turn on when a person interacted with it. Jeremijenko studied this interaction over weeks and months, “which is more like a real timeframe for learning to take place,” she says. What she saw is that most of the time people didn’t interact with the Furby: People in fact largely ignored them. However, people would interact with the toy in the presence of other people.
Manufacturers of such toys promote them by creating the intrigue that it is the toy that scripts the interaction with the human being. But Jeremijenko’s research showed that people retain the script for the interaction: They don’t interact with, but rather through, the toys for the benefit of other people.
One of Jeremijenko’s upcoming projects is based on what she describes as “feral robotic dogs.” The idea is to explore how the toys can be used in a socially significant way.
She and her students plan to take existing commercial robotic dogs, which she describes as the “cheapest, most widely distributed robotics platform,” then mechanically and electronically upgrade them with open source software the team develops. They even expect to add an all-terrain capability.
“We’re going to de-domesticate the dogs, if you will,” she says with a gleam in her eye. “We’ll equip them with new noses to detect environmental toxins and new brains to track those toxins.” Then they’re going to pick a site of community interest, such as the Mission Bay landfill in San Diego , in which to run experiments with the dogs.
The team, which includes students from as far afield as Ireland, England, Norway, Australia, and Yale and Cornell universities, plus UCSD, will be faced with the typical design challenges any engineering students face -- but in the context of a socially useful experiment. They will publish their initial designs to establish the foundation for an online distributed community that can inexpensively adapt the dogs and add to their trove of capabilities. “Our approach is like what you see happening in the Linux community,” says Jeremijenko, who holds a Ph.D. in Computer Science.
This approach is different from other models, she says. In robotics competitions, for example, if you need a motor, you buy a motor made by the competition’s sponsor. “That’s what I call the ‘sports model'," she says. "Everyone plays on the same field and agrees on the rules beforehand.”
But that teaches students to be technology consumers, she says. It doesn’t teach them technical competence to specify a real motor, or understand why you need torque or the tradeoffs involved. It’s these issues that she’s interested in: “I try to encourage students’ resourcefulness and entrepreneurial spirits by challenging them with open systems.”
Furthermore, to design a good system, the students need to understand the social and political context in which their design will operate. Adds Jeremijenko: “That’s one of the problems I see with technical education today: Students are not encouraged to think through the real-world context of their engineering projects.”
By using dogs “with a gazillion dollars of marketing promoting them,” Jeremijenko is also rethinking the use of toys. Why not use them as suppliers of scientific data about topics of interest or concern for the public?
“Unlike peer-reviewed papers, which are our societal standard for conveying scientific data, the dogs will display information through their movements that children and 95-year-olds alike can understand," she says. "Displaying data for a range of people provides an opportunity for evidence-driven discussion and public discourse.”
The sensors used on the dogs will also include cameras. It’s traditional in robotics, says Jeremijenko, to put cameras where the eyes would be, underscoring the fiction that these beings are alive. But this team plans to put cameras on the “non-barking end” so that the cameras point back at the people following them who are trying to track and interpret what the dogs are doing.
“That will be the dogs’ real value,” says Jeremijenko. “It’s not the data per se that matters, but rather what sense people make of that information. We live in a participatory democracy after all: This is one way to script that participation.”