Systems scientists like Steinfeld foresee software that learns from the user, not from the programmer. Basically, RADAR is a kind of computing technology that seems fully aware of its context. Don’t worry—it’s not exactly the evil Skynet system from the Terminator flicks. The developers also refer to the technology as PAL, a personal assistant that learns. Unlike current versions of self-learning software that involve auto-fill text boxes, RADAR will actually get a jump-start on tedious, unwanted work. “All of us get our in-boxes flooded with things we need to do every day,” Steinfeld says. “Whatever the context, there are always actions involved: Don’t forget to buy milk, update this address, and so on. So what if you had a system that could learn about your needs, tasks, and activities just like a good assistant?”

Then Steinfeld pulls up a slide that highlights a startling find. Researchers claimed that subjects experiencing message overload saw IQ test performance decline by more than twice the amount caused by marijuana use. Great. Another year of phone, e-mail, and text messages, and we’ll all be worse off than the stoner characters in Pineapple Express.

While e-mail organizers typically put the burden on users to sort and file incoming messages, RADAR reads and understands the context of your messages and sorts them for you. It then identifies the things it can get a head start on—address and phone number changes, for example—and presents those tasks in a to-do list. Your daily e-mail routine could go like the text-editing feature in Microsoft Word. The program lets you see an editor’s work and click to accept or reject changes. Similarly, RADAR would help you speed through your e-mails.

RADAR also contains a function that creates briefings. The software studies the way the user handles large amounts of data and highlights the parts that seem important. Steinfeld sees the tool making a transition into other industries. “Say you’re a nurse on a hospital ward. You’d like to relay all the relevant information about that ward during your shift change so there’s no problem with the hand-off,” he says. “RADAR’s briefing component could track the status of each patient in the ward and present that information in a briefing for the nurses on the next shift, freeing up more time for one-on-one patient care.” The scenario illustrates the genius of Kind Tech: Instead of replacing nurses with artificial intelligence, expert systems, or white-capped robots, the technology gives nurses more face time with their patients.

One of the most ambitious projects at the Quality of Life Technology Center, a joint project with the neighboring University of Pittsburgh, involves developing a more sophisticated GPS unit that would map out more advanced routes based on very specific preferences, making the current turn-by-turn technology seem primitive by comparison.

“Right now, the preferences in your typical GPS unit are very limited,” Steinfeld says. “You can tell it that you don’t want highways, toll roads, and maybe ferries. But let’s say you were disabled and had a wheelchair van, and you really don’t want to go over any road with a big bump in it. But how do you put that into your system? Trouble is, you can’t.” The proposed device could also take into account preferences that other drivers can’t articulate. Does your car pirouette on icy roads? Find the route most likely to remain clear. Hate dealing with game-day traffic? Take an automatic detour away from the stadium.

“Maybe you hate unprotected left turns,” Steinfeld says. “This device would know that and map out a route that would help you safely avoid them.” One thing the gadget definitely will not do: take control of the wheel. “The important thing is that it gives you a recommendation, but leaves the choice in your hands,” Steinfeld says. “You don’t want a system that nags you, but one that gives you the information to make the right choice.”

Steinfeld powers down his computer and offers a quick tour of the school. Leaving his fourth-floor office, he passes display cases containing ancient computeralia: an old MS-DOS manual, a Datavue 25, and an ST600 tape backup drive from 1988 boasting 60 megabytes of storage space. (That last item took a mere 90 minutes to fill up.) Ads hang on the walls for “Humanoids 08,” a conference in Daejeon, Korea, offering lectures such as “Motion Planning and Physically-based Simulation for Medical Robots.” (If one day R2-D2 shows up to take your prostate exam, you’ll know where to send the thank-you note.) Even for Carnegie Mellon faculty, it turns out, androids are still sexy.

Steinfeld crosses an open skyway with panoramic views of the campus. On one side, the windows overlook a massive construction site—a $100 million computer science center funded in part by the Bill and Melinda Gates Foundation.

Steinfeld leads me into a room called the NavLab where for more than a decade Carnegie Mellon researchers have been working on cars capable of safely driving themselves. It’s a basic garage: concrete floors, high ceilings, fluorescent lighting. Though an Oldsmobile Silhouette now sits nearby, an absent Pontiac Trans Sport is the lab’s most famous former resident. In 1995, two researchers drove the car across the country controlling only the gas and brake. Sensors mounted all over the car worked in tandem with an advanced computer program called RALPH—an acronym that would sound alarming to the easily carsick—to keep an eye on the road. In another corner sit the three-year-old remnants of a competition called RoboCup, where teams of Sony AIBOs—little robotic dogs—were programmed to play a game of soccer against each other. The electronic pups nudged a miniature ball down a small field with their heads, and could even create simple plays and shoot toward a goal. By 2050, the RoboCup project hopes to create a humanoid team that could take on the best soccer players in the world.

PAGE 1 2 3 4 next