A New York Times article entitled “Making Friends with a Robot Named Bina48” has lit up the blogosphere, even to the point where I received emails from friends and associates who are not in the robotics business. Bina48 is an armless, upper torso humanoid robot, modeled on Bina Rothblatt, vice president and co-founder of the Terasem Movement, a 501c3, not-for-profit charity chartered with researching and educating the public on human lifetime extension. According to the Times article, hours of conversations with Rothblatt have been uploaded into computers that serve as Bina48’s mind (and personality).
There are two reasons for the article’s popularity. First, it was the Times… the Sunday New York Times. Second, the article was accompanied by a video that could be referenced, reposted, and go viral. For those of you who have not seen the piece, or other Bina48 videos (they are out there), well, let’s just say that it was not a good day for those involved with the Bina48 project or for the robotics industry.
Bina48 was designed by David Hanson, CEO of Hanson Robotics and developer of the well known Albert-Hubo robot, the Philip K. Dick robot, and the Zeno robot. Hanson’s goal is to eventually develop robots that can interact with humans socially in a deep, complex, and meaningful way. I have always been taken with Hanson’s systems, as well as with his dedication, drive, and intelligence. That’s why the Bina48 piece was such a letdown (or as one of my associates phrased it, “an epic fail”).
But why the negative reactions? Demonstrations of the Albert-Hubo, Philip K. Dick, and Zeno robots captured on video did not elicit the same response from me. Yet these systems and Bina48 are basically the same. Why did I, and many others, find the Bina48 videos so off-putting? I believe it comes down to expectations, experience, and video editing.
The Times reporter actually expected to have a context-sensitive “conversation” with Bina48. However, to her dismay, and unlike the human-robot verbal interaction on display in videos of the Albert-Hubo and the other Hanson systems, many of Bina48’s responses were random, with others bordering on incoherent. In some scenes, the reporter seemed uneasy (and made the viewer uneasy). I could imagine her thinking to herself, “I drove all the way up to Bristol, Vermont, for this?”
The story (and video) also included samples of a number of less-than-successful verbal exchanges. For the vast majority of videos of robotics systems available on the Web or shown elsewhere, these types of scenes are cut out of the final video. Clearly, the New York Times reporter was unfamiliar with how to posit questions to robotics systems like Bina48, so as to lead them to answer with a correct response, or at least one that has some relationship to the question asked. This is not the case for those who demonstrate these types of systems on a continual basis.
I believe that those familiar with the primitive state of human-robot verbal interaction subconsciously give the systems a pass. And why not? It is difficult stuff and the achievements thus far are extraordinary. For the robotically uninitiated, such as the Times reporter, however, the assumptions and expectations are vastly different.
As an armless upper torso with a lifelike, talking head complete with rolling eyes and sporting wires out the back, Bina48 is a perfect exemplar of the “uncanny valley,” the description applied to the level of unease or revulsion that comes over humans when viewing lifelike robots. Bina48’s stilted, random responses increased the level of creepiness, pushing the robot into the realm of the uncanny Grand Canyon, the walls of which are made up of the general public’s expectations for human-robot verbal interaction.