acm - an acm publication


The human-centric approach

Ubiquity, Volume 2001 Issue January, January 1 - January 31, 2001 | BY Michael L. Dertouzos 


Full citation in the ACM Digital Library

Michael L. Dertouzos is director of the MIT Laboratory for Computer Science, and author of the best-selling book What Will Be: How the New World of Information Will Change Our Lives. His newest book is The Unfinished Revolution: Human-Centered Computers and What They Can Do for Us. In the following interview, Dertouzos talks about bridging the technology gap and demonstrates a speech-recognition system.

UBIQUITY: Bill Gates has recently been saying that technology doesn't offer much to third-world countries, such as those in Africa. What do you think of that idea?

MICHAEL DERTOUZOS: I'm very much in tune with him. Incidentally, he's given me a wonderful quote that's on the cover page The Unfinished Revolution in which he praises the human-centric approach. I was writing about the digital divide long before it became popular and famous. One of the first things I found out was that even if you had all the computers and all the training in the world, if you magically provided all this to the Nepalese, you wouldn't achieve anything. The reason I know is because I tried.

UBIQUITY: You tried, how?

DERTOUZOS: Well, we hired a Nepalese scholar and conducted some studies, both in Nepal and in Africa. We focused on Nepal because it is so poor. We figured if anything could be done there that would demonstrate a quick jump in the country's GDP, it would be a great heroic act. Those were the na�ve days when we felt we could just get computers, connect them to the Net, and let them be used for health care, education and the usual things. We found out -- and this is where I got my conclusion -- that only 27 percent of the Nepalese are literate. The remaining 73 are not. So that is the first obstacle. The second one is that of the 27 percent who are literate, only five to eight percent speak, read and write in English. Of those, only a few hundred had any skill to sell. The obstacles of poverty became extremely visible and they drove the lesson home.

UBIQUITY: Does that lesson mean that it's essentially futile to try?

DERTOUZOS: It is not futile but very difficult. There are both technological and human approaches we can take.

UBIQUITY: What are they?

DERTOUZOS: It's a very, very difficult issue. Before we get to them, keep in mind that the first thing I wrote about the information revolution was that, left to its own devices, it will increase the gap between the rich and poor people. So my position is even more aggressive than saying, as some do, that the information revolution will help the poor. I'm saying that, left to its own devices, the information revolution will increase the gap.

UBIQUITY: Why do you believe that?

DERTOUZOS: The reason is very simple. Computers increase productivity. They make us able to do more things with less effort. If we're rich we can buy them and can get more productive, hence richer using them. The poor can't afford to buy them and they stand still. That's the formula for an exponential growth of the rich-poor gap.

UBIQUITY: So how can the problem be solved?

DERTOUZOS: "Solved" is too strong a word. Here are a few helpful things we can do. One is low earth orbit satellites that whip around the earth. They work like a cellular system passing information to each other. When they're over Boston and New York, they're very busy. When they're over Africa, they are totally silent. We could leave them on for a very low marginal cost (where "we" is anything from the countries of the West under foreign aid to the owners of the low earth orbit satellite systems). I've talked to Bill Gates about this and he is willing to help. So that's one technical approach -- letting them use our low earth orbit systems for almost zero marginal cost.

UBIQUITY: And another solution?

DERTOUZOS: The second technical approach is to provide better speech-understanding systems. Because the systems do not require the ability to read and write, we can reach a couple of billion people through medicine, training, agriculture and things of that sort. We can open tomorrow's world of information to another billion people, the Chinese and the Koreans and other people who use ideograms. Because they cannot use keyboards they are largely excluded from the predominately English Internet. So two technologies that I can think of are low earth orbit systems for communications, and speech understanding for bridging the linguistic and illiteracy barriers.

UBIQUITY: What is a realistic timeframe for speech-understanding systems actually helping?

DERTOUZOS: When is the last you've used a top-of-the-line speech system? Have you been exposed to anything recently?

UBIQUITY: It's been a while.

DERTOUZOS: OK. Stand by. I'd like to show you one. [He calls a number on his speaker phone.]

COMPUTER VOICE: Hi. Welcome to Jupiter, the MIT lab for computer science weather information system. This call is being recorded for system development. You may hang up or ask for help at any time. How can I help you?

DERTOUZOS: What's the weather forecast for Boston please?

COMPUTER: In Boston this afternoon, mostly cloudy, high in mid-50s. Tonight, cloudy. Lows in mid-40s. What else would you like to know?

DERTOUZOS: Do you want to ask about a place yourself? Go ahead and ask.

UBIQUITY: What's it like in Atlanta now?

COMPUTER: In Atlanta Thursday, cloudy in the morning. Then showers and thunderstorms in the afternoon. High in lower 70s. 80 percent chance of rain. Thursday night. Cloudy. Chance of showers and thunderstorms until midnight. Lows in lower 50s. Is there anything else?

DERTOUZOS: Please connect me to Voyager.

COMPUTER: Welcome to Voyager. How can I help you?

DERTOUZOS: What are the traffic conditions on Storrow Drive please?

COMPUTER: Here are the full traffic reports for Storrow Drive. On Storrow Drive and Soldiers Field roads heading in both directions. The traffic is moving freely with an average speed of 32 miles per hour.

DERTOUZOS: Back to Jupiter.

COMPUTER: Welcome back to Jupiter. What else would you like to know?

DERTOUZOS: Good-bye.

COMPUTER: Thank you for calling. Have a nice day.

DERTOUZOS: OK. I just wanted you to have a live performance.

UBIQUITY: That's very impressive.

DERTOUZOS: This is the state of our systems, developed by Dr. Victor Zue and his group at LCS, which are considered among the top in the world. Any speaker, with no training, can ask questions as long as the domain of discussion is restricted. For example, when talking about the weather or about the traffic these systems understand about 95 percent of what you're saying.

UBIQUITY: When will that technology be ready for prime time?

DERTOUZOS: I think that kind of technology should be available on the street within a five-year span. I know there are some 20 spin-offs in this area pursuing various threads of it. There are big companies like IBM and Phillips who are putting a lot of money behind speech now. It's a technology that's finally bursting. It's been submerged for a long time, but now it's really reaching the bursting stage. In ten years, I think it will be widely used. In one year, you will see TV sets out of Phillips that you can command with your voice.

UBIQUITY: What other kinds of useful applications are you doing?

DERTOUZOS: One of our doctors, an M.D., Hamish Frasier, is doing a wonderful experiment in which people in the bush, who can neither read nor write, are trained to take X-rays of the villagers. They just put the X-rays against a window and photograph them with a digital camera and then stream the picture digitally to Boston where he diagnoses and helps them out. So there are possibilities for helping. But when you are dying for food, getting bits of information is not the best thing.

UBIQUITY: How do you orchestrate the two approaches?

DERTOUZOS: Well, I'm not crazy enough to try to orchestrate anything in the developing world. You just explore various opportunities, and you ask people. You try things rather than talk about them. I would like to see the World Bank do the same. I'd like to see them dedicate maybe five or ten percent of what they're doling with their 30-plus billion for structural loans every year be applied to worthy uses of Information Technology in the developing world. The best efforts so far are the ones that involve the local people and entrepreneurship. And they all start with telephony. Before you can use sophisticated Internet exchanges and interactions, you need telephony.

UBIQUITY: Do you think it's possible -- or probable -- that some of the developing countries could leapfrog over the more developed countries?

DERTOUZOS: No. That kind of stuff has been tried before many times and it doesn't work. France tried it in the '70s with Jean Jacques Servan-Schreiber. They tried to help people in Senegal to leap frog the Industrial Era and go to the Information Era directly. They failed.

UBIQUITY: Can you think of any examples of success?

DERTOUZOS: I'll give you some positive signs. If you go to India you do see a little leapfrogging in Bangalore. It's very small. You're familiar with the programming that's being sold by InfoSys and others? These people sell software services to the United States and England from a distance. I say it's small because it involves only a few thousand of the best-educated Indians who can program and speak English. Meanwhile, there are 50 million Indians who read and write English and who can deliver office work. Now, that's getting into some real numbers. This means that there is a workforce comparable to that of the United States office workforce sitting out there in India able to deliver clerical services at a distance, and able to do it at a quarter of the cost. Nobody is talking about that in today's Internet. Yet it is much bigger than the content that everyone is hyperventilating about. The traditional broadcast content, whether it's print or video or songs, that you chop up and sell for a dollar a piece, is worth, at most, one trillion dollars. I'm talking here about five-plus trillion of office work that's going to flow over the Internet within the next 15 years or so.

UBIQUITY: What would you expect the socio-economic and political impact of that to be on the Western countries? I mean, suppose that started to happen within the next 10 years in large numbers?

DERTOUZOS: I forecast a major redistribution of labor. Fifty percent of the industrial economy is what we used to call in the old days, white-collar work. Now we call it office work. So 50 percent of the industrial world workers are office workers. Of that office work, I estimate a third, maybe a little more, can flow over the information marketplace. That's where I get five-six trillion dollars a year.

UBIQUITY: Are you not worried that the American workforce would take that amiss?

DERTOUZOS: No more and no less than I would have been worried in the industrial era, with the corresponding shift. The history of the world is full of these socio-economic shifts. Some jobs will be lost. Other jobs will be created. If you look at the industrial era, you can list them. Before the Industrial Revolution, we didn't have psychiatrists. We didn't have jet pilots. We didn't have engineers. Now we have all these kinds of jobs and we don't have crafts people. And we don't have people who dig the earth with axes. So there are going to be some corresponding changes here, and we must help the people who will suffer from these shifts. Nobody is talking about these things. Everybody is focused on the rather short range.

UBIQUITY: Are there any projects going on now that look at long-range solutions?

DERTOUZOS: We're building this thing at MIT called Oxygen, which is a prototype of a pervasive human-centric system. There are another four or five other places around the world that are beginning to build human-centric pervasive computer systems. UC Berkeley is doing something, not very similar but along these broad lines. Georgia Tech is another one. Washington University is another one. Carnegie-Mellon is perhaps the closest. Most of them put the emphasis on pervasiveness and ubiquity. Here at MIT, we put the emphasis on both. I personally put all my emphasis in human-centric.

UBIQUITY: Why doesn't everybody? And that's a serious question.

DERTOUZOS: It's easy for people to envision a great abundance of what they already have. It's much harder to go after something radically different, because it calls for a big mind shift. Inertia is attracted to pervasiveness.

UBIQUITY: Is your philosophy embodied in MIT undergraduate courses?

DERTOUZOS: Not yet but I'm hopeful it will be. I don't worry about it that way. I worry about building human-centric machines and systems that really serve people. What I'm trying to do with my colleagues at MIT is make sure that we convey this philosophy to the world. We'd like to see a radical shift in the way we build and use computers. One that really puts the person in command, that brings the technologies close to people, and that focuses everything around what's good for people rather than the other way around. If we do this job well, all of tomorrow's students will be studying it.


Leave this field empty