Samantha — Just Another Pretty Voice?

May 2014
By Bryan Bergeron

The AI named Samantha in the sci-fi movie, HER is the latest attempt by the film industry to depict the possibilities of human-computer interaction and bonding. The concept of human bonding to a computer, robot, or other machine isn’t new — sci-fi writers have been exploring the issue since at least the early ’50s. Then, there’s the classic Eliza psychotherapist program from the ’60s, and the bonding between user and the ubiquitous cell phone. Still, the film has value in reminding us of what makes a good computer/robot/phone-human interface, and what doesn’t.

For example, take voice recognition. Seri — perhaps the most popular voice recognition system in general use — hasn’t proven to be the solution to fulfilling the narcissist’s dream of the perfect, ever-available, and ever-attendant companion. For starters, talking is often inconvenient. Think public transportation system, nice restaurant, working in a cubicle farm, or a long boring meeting at work. A few quiet keystrokes would be tolerated by others where speaking would be practically impossible.

Then, there’s the distraction factor of voice interaction. Most workplaces wouldn’t put up with employees in constant phone contact with their significant others (organic or otherwise) because they’re supposed to be focused on work. Unless you happen to have a job that doesn’t rely on real social interaction — such as the writer in HER — a real Samantha AI would never do.

So, what is the best user interface to computers, robots, cars, and other intelligent machines? The ultimate interface is — I believe — a direct conduit to our conscious thought. After all, haven’t you ever just wished your mousecontrolled pointer would move to a certain word or radio button by simply reading your mind? I have. What about the microwave oven? Why repeatedly punch in the same one minute and 20 seconds to heat a cup of coffee? The voice recognition interface depicted on StarTrek is inherently flawed. Why ask for “Earl Grey, Hot?” In the future, the computer interface should determine what you want without you having to ask.

Of course, the cognitive interfaces currently available to consumers are maxed out by tasks such as controlling a simple drone, much less knowing what kind of tea you’d like at the moment. With time, however, such technology is inevitable. We just need to be sure that what sounds good today will be worthwhile when we achieve it in the future.

For example, I wouldn’t trade a microwave or 3D printer that could read my mind for a future that provides the government and corporations with access to my thoughts as in the movie, Minority Report. What about the situation where there’s one TV and two watchers? Does channel selection go to the viewer with the most focused mind or simply the most stubborn one? And what about the auto food synthesizer? Does the dinner menu respond to the four year old who is fully focused on ice cream, or to the adult who is juggling a dozen other decisions? It’s fun to take in a film like HER, but as roboticists it’s also important to think about the implications of the underlying technology on practical future robot designs. SV