Redmond, Wash., August 17, 1998 — The human brain and computers process information in dramatically different ways. Multiplying large numbers is a difficult task for the brain, for example, yet a relatively simple one for computers. On the other hand, recognizing graphic images is easy for the human brain, yet complex for computers. So does the brain offer clues for making computers "smarter"? And do computers provide insight into the links between chemical and electrical signals within biological organisms, and their resulting behavior?
Computer scientists and biologists will seek answers to these questions this week at the University of Washington/Microsoft Research Summer Institute. This year's topic, "Intelligent Systems: Biological and Computational Perspectives," will bring together 45 top researchers from around the world from August 18-23 to explore links between intelligent systems in the natural and technological realms.
"What can computer scientists and engineers learn from what biology has spent billions of years perfecting?" asks Chris Diorio, an assistant professor at the University of Washington who is helping organize the event. "We as engineers have tended to ignore all those lessons. Likewise, we have lots of mathematical and analytical tools that can be used to study the behavior and communications of biological systems."
Now in its second year, the Summer Institute was created as a partnership between Microsoft Research and the University of Washington Computer Science and Engineering Department to exchange ideas about computer science. This year's topic was the brainchild of Eric Horvitz, senior researcher at Microsoft Research, and Dennis Willows, director of the University of Washington's Friday Harbor Laboratories. "It really hadn't occurred to me that the concept of intelligence in a brain would be directly comparable to intelligence in a machine," Willows said. "But the more I think about it, the more I see that, in fact, it's a no-brainer."
The conference will begin at the University of Washington in Seattle, and will move to the Friday Harbor Laboratories in San Juan Island later in the week to enable researchers to examine firsthand the results of neurological studies on crabs and sea slugs. Marvin Minsky, co-founder of the Artificial Intelligence Lab at the Massachusetts Institute of Technology and one of the world's leading theorists of artificial intelligence, will address the researchers on Wednesday.
There are numerous opportunities for computer scientists and biologists to learn from each other during the Summer Institute, organizers said. For example, the way living organisms process information using different parts of the brain may generate ideas for computer scientists developing distributed operating systems. Such systems spread computations over large networks, rather than relying on a single computer. In the same way, what computer scientists know packet switching, which breaks up messages into small packets for transmission, may be valuable for biologists studying communication between neurons. "We're pushing the boundary between computer science and biology," Willows said.
A division of Microsoft, Microsoft Research has been studying ways to simplify and enhance users' PC experience since 1991. Research on intelligent systems is concentrated within the Decision Theory and Adaptive Systems group, which is investigating principles of intelligent decision making and learning, and exploring ways to make computers more flexible and adaptive.
One topic the group is exploring is how to make computers more responsive to the interests and needs of individual users. By noting a user's patterns of work, software can infer what assistance to provide. And through observations of a user's Internet habits, software can draw conclusions about which subjects interest the user, and offer relevant data as well suggestions for other Web sites to search.
Horvitz, who manages the Decision Theory and Adaptive Systems group, maintains a strong interest in the links between intelligent systems in the computational and biological realms. "Computer scientists have been developing theoretical models for learning and decision making," Horvitz said. "We're interested in opportunities for bolstering these theories and applications by examining clues from neurobiology about the functioning of systems that have been designed and 'time tested' by natural selection. On the other hand, our mathematical models may provide insights for neuroscientists."
Computer scientists do not expect to leave this year's Summer Institute with short-term solutions for making computers smarter. Rather, they are hoping the talks will generate ideas that could influence their approach to innovation several years into the future.
Horvitz mused that the long-term goals of the Summer Institute are captured by the image of the Wright Brothers observing the way birds fly. "The Wright Brothers weren't necessarily interested in building a machine that flaps its wings; they were hunting for a deeper understanding, a pursuit that eventually blossomed into the principles of aerodynamics," he said. "In the same way, we will hunker down and continue our pursuit of principles of 'aerodynamics' for intelligent systems."