Future computing: Human-centric systems

by Sami Lais

May 7, 2001 —

The Information Revolution won't fulfill its promise until we stop thinking as though we're still in the Industrial Revolution, according to Michael L. Dertouzos, director of MIT's Laboratory for Computer Science. After 40 years of building computers, little has changed, says Dertouzos, the author of seven books, including his latest, The Unfinished Revolution: Human-Centered Computers and What They Can Do for Us (HarperCollins, 2001). The future of computing lies in "making systems serve humans," he says. "That must be our goal." Computerworld writer Sami Lais interviewed Dertouzos about the future of computing.

You talk about human-centric computing, in which the computer isn't a single device but a room where computing is around you and in the air. How is that different from pervasive computing? There is a lot of confusion between pervasive or ubiquitous computing on the one hand and human-centric computing on the other. They are not the same. Pervasive computing implies a lot of equipment, where the focus is on a lot of devices that are themselves computers. Human-centric computing, however, focuses on the human. Today, computers are hard to use. If we make them more pervasive and use more of them, there will be that much more aggravation around us. By focusing on human-centered systems, we declare that our goal is to serve humans. Whether that calls for more or less stuff is secondary.

In the future computing model you describe, interaction will be speech-activated. Why? Much of it will involve speech understanding, not just speech activation. Speech is natural for people, hence easy to use. That's why human-centered systems need to have speech. Remember, the fundamental thing that will set human-centered systems apart is that the computers will serve you. They won't care how you communicate - whatever way is most comfortable for you. Ironically, computing will follow an old model. It's one that is unsavory for humans but perfect for machines, and that is the many-dumb-servants model. The software that serves you will not take human form like a robot, nor will it have a fuzzy face and big ears. It will simply involve programs that sit there doing the things you want them to do.

Pervasive computing is beginning to be fact: With cell phones, laptops and handhelds, we can work pretty much anywhere. How long before the transformation that lets us do this and more, simply by talking to a room? Pervasive computing is easy. It's what we already have, only more of it. Human-centric computing will take a shift in thinking, and it will take time for vendors to offer hardware and software that expresses it. But work in human-centric systems goes on. It has for some time, first at the University of California, Berkeley . . . Carnegie Mellon University, Georgia Institute of Technology [and] the University of Washington. Of course, at the Oxygen Project at MIT. And a lot of work is being done in speech recognition . . . at IBM and Philips and Microsoft. We're starting to see a lot of start-ups in this area, and that's always an indication of a technology that's heating up.

What will it take to cause commercial IT developers to embrace the concept of human-centric computing? It's already happening. Microsoft has announced Hailstorm, a user-centered computer environment, as they call it, part of their .Net system. Ultimately, successful prototypes of human-centered systems will cause start-ups and big companies to go after the new forms.

You say we have to change our attitudes, too, and make them more human-centric. What do you mean? Take e-mail. We have to do something or we'll all drown in e-mail. [One survey says that people] spend an average of an hour and a half a day on e-mail. That's a ridiculous time, and it will get 10 times bigger in the coming decade, as new users join and current users devote more time to e-mail. Surely, people won't spend 15 hours a day on e-mail. And you can only do a little bit by machine. I spend an average of 18 seconds on each e-mail, because I've set up push-button-action responses, but even so, I am only delaying the arrival of my total overload point. We have to start using metadata and XML to put labels on e-mail that describe what's in the e-mail messages so that we and our machines can select or reject the e-mail that comes in. Ultimately, you and I have to change our attitudes and trash a lot of the mail we get. Just because we have become interconnected, we have not acquired the right to bother other people with our writings, nor the obligation to respond to them. E-mail is not going to change thousands of years of socialization.

Will there be a backlash against technology? Not a backlash, but how much better off we'll be with tomorrow's human-centered systems depends on the individual. Look at it this way: You can divide people into their principal components. I like to think of the individual as a four-cylinder car: The four cylinders are our physical, rational, emotional or artistic and spiritual dimensions. Most technologists run on only one or two cylinders, generally the rational and physical. Humanists run on just about the same number of cylinders: the physical and artistic. So if you're a technologist or a business person focused on planning and getting things done, you'll be incredibly better off with computers. If you're an artist, you'll be only marginally better off. You'll have tools to do your creative work, but you won't be able to write or paint better. And if you're a monk, forget it. You won't be better off, because spiritual activity is primarily internal to people. I'd like to see us all learn how to run on all four cylinders. To me, that's what it means to be human.

CEOs want their CIOs to be technology visionaries who also deliver a solid return on investment. In this context, how does an IT manager embrace the human-centric future you describe? Sorry to toot my own horn, but [he] starts by reading my book. Then, [he] browses the Web and reads about these other efforts I've told you about. [Then he] undergoes a total brainwash by focusing on having systems serve people, rather than the other way around. Look at today's speech systems. What are the 200 start-ups doing with speech? Can you use speech in your business? Then look at automation. Can you automate stuff people do that does not require intelligence, so you can relieve them of this work? Can you help your people work across space and time more effectively? Can you customize your systems to your people's individual needs? It's not enough to say you support the ideas. You must show it in every action.

Computerworld