• 7 minute Read

This Artist Is Teaching Tech Execs How To Make Devices More Lovable

Apple might make a Siri-like assistant its mediator for the Internet of things. One of the uses imagined for IBM’s Watson is as a supercomputer assistant for doctors and researchers. MIT Media Lab grad Alexander Reben is ready for this humanizing revolution.

Robots don’t need to have artificial intelligence and a voice like Scarlett Johansson for people to form emotional bonds with them. (Those things, it should be noted, do generally help however.)

In fact, it takes surprisingly little, as Alexander Reben discovered while working on his masters thesis at the MIT Media Lab. His robot BlabDroid looks like a cardboard Wall-E and totters around asking people personal questions. Watching recordings of BlabDroid’s interviews, Reben was struck by the way people opened up to the robot in ways they wouldn’t to a human stranger, and he decided to leave his engineering job at Cisco to explore the ways people interact with technology. Though his work has far-reaching implications for the tech industry, Reben initially found that the best place to pursue his inquiry wasn’t a lab or a startup, but an art gallery.

“I think the art world provides lot more latitude to look at these things,” Reben says. “You can learn a lot by having things in an artistic space where people are open to having experiences.”

But Reben didn’t leave the tech world entirely when he began his art career. Rather, he’s using art as a way of exploring problems tech companies are just beginning to confront–and those companies have started to notice the work.

“Everyone wants to create technology that people just love, that they form an immediate bond with,” Reben says. “That’s what I’ve been able to do, even in the absence of functionality.” Reben has given talks on how to create better emotional bonds with technology, design less frustrating interfaces, and make devices that are more intuitive to use. “It’s always interesting for me to show art to a CTO,” Reben says. “A lot of these ideas are very philosophical and emotionally based, and it helps to show them through demonstration rather than through a PowerPoint.” Next month, he’ll give a talk to C-level executives at the TTI/Vanguard conference, titled “The needy robot and our relationship with emotional machines.”

In The Beginning, There Were Bots

One of Reben’s early findings was that a machine doesn’t have to look as minimally humanoid as BlabDroid for people to connect with it. His installation Pulse Machine, a collaboration with the sculptor Alicia Eggert, was simply a kick drum beating 60 times a minute, counting down to its death in 78 years, yet gallery visitors felt sympathy for it. One of Reben’s more recent works, shown at the Volta art fair in New York last month, consists of two metallic Mylar balloons, one of which is tied to a static electricity generator and the other to the ground. As the charge builds the two balloons are drawn to each other; when they touch, a blue bolt shoots between them, balancing the charge and sending them flying temporarily apart. Viewers ascribed emotions like love and aggression to the balloons’ oscillation.

“My theory is that behavior creates a stronger connection than looks even can create,” Reben says.

Now Reben is working on a series of robots with personality disorders. Robots are programmed to be precise, which is what makes even the ones with human features seem so uncannily inhuman. “If we made robot like a human it wouldn’t do things quite so exactly,” Reben says. “It would do a task but never quite well enough. It would always be futzing. Would people anthropomorphize that more? Would it seem more human if it were less perfect?”

The great paradox about much of Reben’s work, is that though we we may know that it’s absurd to empathize or engage in an emotional way with machines, we do it anyway. Whenever you curse at your broken-down car or plead with your unresponsive printer you are, in a limited subconscious way, treating them like living things. According to iRobot CEO Celeste Biever, 80% of Roomba owners name their vacuum cleaners. As social animals we’re hardwired to treat things socially, even when objectively they’re nothing like us at all.

Much of the corporate interest in social robots comes from companies trying to establish themselves within the Internet of things, the term for a network of digitally connected appliances and objects, from fitness monitors to refrigerators. As more of our devices go online, Reben says, we’ll need a central nexus, an ambassador for our appliances, lest we be overwhelmed with streams of data and separate interfaces for every smart toaster, speaker, and thermostat. Because we’re social animals, it makes sense for that interface to be a social machine, something that exploits our ability to process and convey complicated information using social cues. The more we know about the ways we anthropomorphize technology, the easier it will be to design such machines.

More Man Than Machine, Now?

Companies designing robots to work alongside humans have already begun moving in this direction. The industrial robot Baxter, designed by Rodney Brooks’ company Rethink Robotics, has a digital cartoon face. When it looks to its right, the people working alongside it know it is about to do something in that area; when it looks puzzled, they know something is wrong, and they know it much more intuitively than they would if, say, a red light started blinking or it presented them with a blue error screen. Presumably a puzzled cartoon robot face would also be less enraging than the impassive blue screen of death.

Apple recently won a suite of patents that indicates it may make a Siri-like assistant its mediator for the Internet of things. One of the uses envisioned for IBM’s Watson is as a supercomputer assistant for doctors and researchers. It would be easy to imagine Google Now, the company’s want-anticipating personal assistant, combing with Google Glass, Nest, and other projects into something similar for the layperson.

Creating a social robot won’t just be an engineering problem. As complex as it is to design machines that can parse spoken language and interact coherently with humans, there is also the psychological question of how to design a robot that people like, one that’s not creepy (HAL 9000) or super annoying (Clippy).

“There’s a fine line between something that’s fun to interact with and something that’s annoying,” Reben says. “Not many people talk to their Furbies for years.” The trick, he says, will be making something that is a little like a dog, which Reben refers to as a “genetic technology” developed over thousands of years for optimal cuteness and companionship.

BlabDroid, the cute robot that emerged from his masters thesis, is touring film festivals around the world, collecting candid interviews for a documentary Reben hopes to make. Now in its third iteration, the robot has become a precision engineered machine of cuteness. It’s smaller with big, wide-set eyes; it has the voice of a 7-year-old child. “It’s pretty universal what people think is cute,” Reben says. “It’s interesting how quickly our brains shift to do that, the child part of my brain says, ‘I better take care of this thing.’”

Reben thinks we’re so hardwired to treat tiny vaguely anthropomorphic things as human that the ambassador of the Internet of things will probably look something like BlabDroid. It would be a powerful and intuitive way to communicate with your devices, much better than, say, addressing your refrigerator. But what he’s seen with BlabDroid also points to some of the pitfalls we’ll encounter as our devices become our companions.

Emotional Rescue

In Amsterdam, one of the BlabDroids asked people whom they love most, then followed up by asking them if they could tell that person now. The idea was that it might make for a touching scene of people professing their love for each other, calling their loved ones, and so on, except that one woman’s answer was her mother, who was deceased, and when the robot asked her to call her, she started crying.

“That was an error case we hadn’t anticipated,” says Reben. Machines can play on our social conventions but they don’t understand them, even if we often act as if they do. “Everyone had good intentions but because of ignorance of robot, that can happen. They don’t understand what you’re saying–they’re just asking questions.”

“As technology becomes more attached to us I think these questions will be very important,” Reben says. “I believe we’re going to need to guard people’s psychology as well. Engineers who work on things like elevators have to be certified so they don’t kill people, but engineers who work on social robots aren’t psychologists. There will be unforeseen risks. A lot of my work deals with this sort of thing. I like to call my installations experiments and refer to viewers as subjects.”


More Stories