Crafting the Next PC Interface
- — 17 March, 2000 17:49
That was the theme of a session titled "@ the Human Interface" at Intel's Computing Continuum here Thursday. Led by Dr. Victor Zue, a senior research scientist at the Massachusetts Institute of Technology, a panel of experts offered their visions of future computer-to-human interfaces.
Wall-size, touch-sensitive displays in interactive workspaces are one way to make computers friendlier, says Patrick Hanrahan, computer science and electrical engineering professor at Stanford University. Interactive conference room tables and workbenches, as well as better projection technology, will make computing easier, he says.
Future computing won't focus on a single desktop or handheld PC, he says. Hanrahan advocates what he calls an "overface," which lets a person communicate with and control multiple devices, but not address each one individually.
Animation is key, according to Ron Cole, director of the Center for Spoken Language Understanding at the University of Colorado at Boulder. Animated conversational agents can improve our interaction with PCs because visual speech augments acoustic speech, and emotions and gestures become more clear and meaningful with animation, he says.
Cole demonstrated animation's power in what he called the McGurk effect. A computer-animated figure uttered three apparently different phrases. Cole pointed out that each phrase was the same, but each "sounded" different as we observed the changing motion of the lips.
"Face-to-face communication is a ballet of auditory and visual interaction," he says. People want to communicate with PCs the way we communicate with each other. Such communication is more gratifying, but today's PCs are totally blind to it. Interaction will improve if computers can understand human nuances.
Ben Shneiderman, computer science professor at the University of Maryland, disagrees. A vocal interaction is primitive, the "bicycle of interfaces," he says. It's time to ditch the keyboard, but visual interaction remains key. The more we can cram on a display, the better, according to Shneiderman.
Instead of trying to create interfaces that react to us, we should make consistent, predictable, and controllable visual interfaces that simply work better. "Microsoft's 'Help' does not," he says.
Search engines shouldn't show hundreds of links on a page--they should show thousands, Shneiderman says. More choices give us more important information.
"A pixel is a terrible thing to waste," he adds.
Finally, Raj Reddy, a Carnegie Mellon University professor of computer science and robotics, took the stage. He asked developers not for a specific type of interface, but simply one that takes advantage of human strengths and helps alleviate problems caused by our limitations.
Humans are good at stuff like working in real time, dealing with language and speech, learning from experience, and tolerating errors. But we also make errors, forget stuff, and get impatient and confused. We're generally a lazy bunch.
"Future systems have to be taught the needs of human beings," Reddy says.