continued from cover


And the way information technology research is conducted has also changed to adapt to this revolution. "Emerging within the research space is an accommodation to a new responsibility of IT research that has the user in mind from the beginning and recognizes that some of the research is best done jointly," says Professor Eugene Fiume, chair of the university's Department of Computer Science.

The traditional disciplinary boundaries of computer science (software) and electrical and computer engineering (hardware), though still useful, are no longer absolute. "We are seeing a much more jumbled picture with a set of overlapping players," says Fiume. "The deployment of technologies is happening in a very much more 'bundled' way - it's harder to separate hardware and software."

Aware of this trend, researchers at the university have been forming teams and institutes - the Nortel Institute for Telecommunications, Bell University Laboratories, and the Knowledge Media Design Institute, to name three major U of T initiatives (see sidebar) - capable of looking at the big picture. "Everyone is much more aware that the impact of this technology is through the innovative systems being put together that build on developments in disparate areas," says Zaky. "U of T's advantage is that we have the people to make up the teams."

The breadth of research disciplines at U of T means that fundamental problems in information technology can be approached from many angles. Faculty researching widely differing aspects of information technology - new hardware (Professors Ted Sargent of electrical and computer engineering and Eugenia Kumacheva of chemistry), artificial intelligence (Professors Sven Dickinson and Craig Boutilier of computer science), or human-technology interaction (Professor Kim Vicente of mechanical and industrial engineering and Professors Rob Wright and Ron Baecker and their colleagues at the Knowledge Media Design Institute) - have a full spectrum of expertise informing their work.
This multidisciplinary smorgasbord brings an awareness of the human experience to the physical sciences (and vice versa) and also allows researchers to form just the right team for the problem at hand.

Photonics, a frontier discipline attempting to harness light for communication and networking in the same way electrons have been harnessed for computing, requires just such a fundamentally inter-disciplinary approach, combining physics, chemistry, and engineering. The remarkable breakthroughs recently made at U of T by Professors Sajeev John and Henry van Driel of physics and Geoffrey Ozin of chemistry needed the combined expertise of their teams to create a revolutionary new photonic material (see "Light Heavyweights" in the Fall 2000 U of T Magazine at http://www.magazine.utoronto.ca/00autumn/f04.htm).

In addition to the John-Ozin-van Driel team, Ted Sargent of electrical and computer engineering and Eugenia Kumacheva of chemistry are making some remarkable advances which could lead to a photonic circuit - a miniature system that manipulates photons (quantum particles of light) in the same way a computer chip moves around electrons.

Photonics has the potential to make another giant leap forward in communications and computing, on the revolutionary scale of the computer chip. Light-powered technology is still in its infancy, but it's already blindingly fast. "A strand of optical fibre thinner than a human hair now allows us to send one million books' worth of information across the continent in one second," says Sargent, the Nortel Junior Chair in Emerging Technologies and a Canada Research Chair.

Sargent and Kumacheva are working on taking individual components and combining them like blocks of Lego to make extremely advanced devices. "Eugenia has innovated a very special kind of building block - her assemblies of functionally sophisticated optical materials take on a strikingly beautiful, naturally formed pattern. It's like a crystal, except on the scale of photons instead of electrons."

Over 20 graduate students, postdoctoral fellows, and research associates are working on Sargent and Kumacheva's combined team to realize the power of Kumacheva's crystals. Their new lightwave devices will clean up optical signals so that they travel as pure waves of light across continents and will shuffle photons intelligently to send signals to precisely the right location. Ultimately, the new materials will allow the seamless networking of computers - and, as a result, better human communication.

It's pretty techie stuff, but Sargent also has a philosophical mandate to make his new hardware innovations work in harmony with nature and support the human user of technology. "I see us transforming the current paradigm wherein computers set the agenda for how we interact with them and with one another. We need discoveries of new materials, of new devices and functions, and of new processes to make the materials and devices. This research will shake the Internet at its architectural foundations."

Kim Vicente, professor of mechanical and industrial engineering and head of the Cognitive Engineering Laboratory, is thinking along the same lines as Sargent, though with a different approach: he researches how users interact with existing technology. User interface is a very hot field, but Vicente's research has an added twist. He looks at workplaces such as hospitals or nuclear power plants where a mistake can have life-and-death consequences.

Though hidden, medical error, for example, has an enormous impact on our society. "The statistics show that each year in the U.S. between 44,000 and 98,000 people die from preventable human error. That's the equivalent of having one wide-bodied jet crashing every day or two, killing everyone on board."

With more and more complicated medical devices appearing daily in the health workplace, the potential for human error increases. Vicente's multidisciplinary laboratory, which includes engineers, psychologists, computer scientists, and anthropologists, tries to design technology to fit people, rather than the other way around.

The two most critical parts of the design process, according to Vicente, are user input and performance testing. "User involvement early on is really important, and it also makes it more likely that they will accept what you come up with, so it's a buy-in issue as well. We measure performance as well, instead of just using focus groups, because what people prefer - often what they're used to - isn't what is always best in terms of safety or error."

Ultimately, Vicente says, it doesn't matter how great the device is if people don't understand how to use it. "If technology doesn't work for people, it doesn't work."

It's all part of a move towards seeing information technology in a human context, not as an end in itself. Artificial intelligence and robotics, for example, have moved away from trying to create a fully autonomous human replacement to technology that will enhance human abilities and intelligence. "Ten to 15 years ago, most vision and robotic systems were fully autonomous but could operate only under very restricted, laboratory-like conditions," says Sven Dickinson, an associate professor in computer science, "but recently, by bringing the human in the loop to provide assistance, many systems can now operate under more realistic conditions, such as outdoor environments."

For technology to complement human capabilities, computers must be able to interact with the world in which humans reside. For a robot, that includes the ability to "see" and recognize objects, which is Dickinson's field of expertise. At the moment, it's a manageable (though not simple) proposition to train a robot to recognize a particular object - say, your 10-year-old green desk chair. But it's a problem on an entirely different scale to get a robot to recognize any and all chairs as belonging to that class of things of many different shapes, sizes, and colours we call "chairs."

Without this ability, which Dickinson calls "generic object recognition," computers have to be trained to recognize each and every object in their space - each coffee mug, each book, every different pen or pencil. Both training and recognition, then, become very expensive.

"This is the technology that will take vision and recognition out of the laboratory and into people's homes," Dickinson says. Although it does have definite applications - Dickinson is working on mobile robot systems for the disabled, for example - he cautions that there are no short cuts. "There are a number of important basic research problems that need to be solved before generic recognition is possible."

Rapid advances in IT have meant that we are bombarded by huge quantities of information every day. Craig Boutilier, like Dickinson an artificial intelligence researcher in the Department of Computer Science, concentrates on assisting humans with the mental task of assimilating all this information. Boutilier and other researchers are working on developing intelligent "agents:" software that can sift through masses of information, schedule your day, negotiate purchases, and make choices for you, all the while learning your preferences and how you make decisions.

Boutilier's specialty is research into uncertainty and the types of trade-offs humans spontaneously make when they are confronted with choices. "When someone is faced with several choices of action, there's a lot of uncertainty associated with the consequences. You could buy one make of car, which is very reliable, but the repairs are expensive, or you could buy another make which breaks down more frequently, but perhaps the repairs cost less. So you have to make these trade-offs."

But the problems of creating a helper technology that mimics the human brain are huge. "When you think of a robot roaming around as a kind of administrative assistant, delivering courier packages, et cetera, to different people, suddenly there are more situations to consider than there are atoms in the universe raised to the 10th power."

In contrast, humans are very good at disregarding irrelevant details to make quick choices, and Boutilier is finding computational "tricks" to give agents the same skills. "The simplest things for humans are incredibly hard computationally."

The difficulty of the task makes clear the complexity - and advantages - of the human brain. "The artificial intelligence community challenges us to ask the question, 'What does it mean to be human?'" says Gale Moore, executive director of the Knowledge Media Design Institute (KMDI), a multidisciplinary virtual institute. "While technology may be robust and reliable, it's humans who give resilience to technological systems. Resilience is about flexibility, whereas technology either works or it doesn't work."

People are at the heart of KMDI's mission. "We work at humanizing technology to enhance society's potential for creativity, learning, and increased productivity," says KMDI Director Rob Wright. "That's the broad mission of the field of knowledge media design - to address the needs of the digital revolution that is sweeping contemporary culture and to help shape the way we learn and work in this context."

Researchers involved with KMDI believe that the thorny problem of making technology work for humans must be approached from many angles simultaneously. "Sociologists don't build things; computer scientists will build their view of the world, which is not universal; engineers will build another view of the world; designers will make it beautiful, but it may or may not work," says Moore. "Bring it all together and you begin to develop a community of practice that can realize truly human-centred design."

KMDI founder and Chief Scientist Ron Baecker, a professor of computer science and the Bell University Laboratories Chair in Human-Computer Interaction, exemplifies this people-first philosophy. For example, Baecker and Moore, supported by Bell University Laboratories, are investigating how to turn web-casting, which is currently a one-way medium like television, into an interactive environment. The goal of the "ePresence Lab" is to give both remote participants, who are accessing an event via a webcast, and local participants a sense of each other's presence, and to enable distant participants to actively engage in discussion and questions. As a result of this added interactivity and "e-presence," webcasts will become a tool that supports more typical human interaction instead of a technology that limits communication to a one-way stream.

Professor emeritus of computer science Calvin Gotlieb, the "Father of Computing in Canada" and a member of KMDI, supports this approach. "Because of the spreading of computers in society," he says, "we need an interdisciplinary focus." In addition to bringing the first computer to Canada in the early 1950s, Gotlieb had the vision to create one of the first courses in the world in computers and society in 1973, which he is still teaching today.

With the speed at which research and applications in information technology are moving, this multidisciplinary focus that keeps the human and society at the centre of technology is absolutely essential to create better and more usable technology and to keep human values driving technological innovation, rather than the other way around.

"The world in the future will be more than a world of computers and communications - it will be a world in which our interactions with one another are much more at a human scale enabled by technology," says Ted Sargent. "Right now, we need to gain inspiration from the amazing thing that is the organism - to figure out how we're no longer on this rigid, uni-dimensional, 'faster, faster' path, but to gain an enlightened, balanced view of how we interact with nature, the laws of physics, and each other.