Kazuhiro Fuchi is both the leader and symbol of Japan's attempt to leapfrog America's computer industry and establish itself as the world leader in computer technology.
As director of Japan's Institute for New Generation Computer Technology (ICOT), Fuchi is coordinating Japanese efforts to create a so-called Fifth Generation computer: a technology that would meld ultra-high-speed electronic circuitry with vastly complex operating instructions, producing "intelligent" machines.
Fuchi, a youthful-looking 47, is backed by a 10-year, $450 million commitment from his country's powerful Ministry of International Trade and Industry. He directs a core group of 40 researchers--all under the age of 35. "You do not get a revolution from the old," he said. His researchers come from many of Japan's leading technology companies--Hitachi, Fujitsu, Matsushita and others.
However, "This is not just support for industry," he stressed in an interview. "Our project goal is more than that, I believe: It is to create new technologies for the 1990s."
Although some American computer scientists express their skepticism, others are concerned that America will lose its preeminence in computers to Japan in much the same way as it did its lead in the automobile and consumer electronics industries. Partly in response to the Japanese, the U.S. Defense Department is organizing a "supercomputer" project of its own.
"We did not expect our project would result in such concern," said Fuchi, before a New York Academy of Sciences computers conference. "It's just too early."
The project reflects a profound shift in MITI's philosophy, according to Fuchi. "Japan grew up from the ruins of the war with the government taking care of industry," he said. "Now the role of government is changing to promote more basic research. Government is involved in this basic research because industry won't" do it, he said.
Fuchi said that the project represents Japan's determination to acquire technological self-sufficiency, rather than imitating the West. "We fear that the old style of catching-up research and development will become more and more difficult," says the ICOT outline. Japan's lack of natural resources impels it to seek technology as the means to continue its rapid economic growth.
Fuchi concedes that ICOT's plans are extremely ambitious. "We play the role of a symbol, but we will become a reality," he said. "Basically, I'm saying 'let's hurry up.' "
What Fuchi wants to do is create a new kind of computer architecture to process data in an entirely different way. "The reason we're going to have new hardware is to change the technology of software building," he said. "Current hardware forces programmers to think in restrictive ways."
The goal has been described this way by Edward Feigenbaum, professor of computer science at Stanford University and an authority on artificial intelligence:
"At present, we speak of computer capabilities in terms of millions of arithmetic operations per second. Japanese planners want their machines to handle one million logical inferences per second." One logical inference equals one step in an if-then sequence of reasoning, Feigenbaum has written.
By the end of 1985, Fuchi expects to have a special set of PROLOG computers employing a computer language that expresses knowledge and processes in terms of these logical relationships.
He said that these machines are an interim step in the creation of the supercomputer technologies and will make it easier to create artificially intelligent "expert systems"--computer programs that think like experts.
An expert system in the medical field, for example, could attempt to capture the accumulated knowledge and experience of medical experts in a computer program that would perform medical diagnoses more rapidly and accurately than humans do.
"Our emphasis is on this logical approach," Fuchi said. "Expert systems should be built on logic. We will eventually get to multilayered logic, which will make much richer representations possible."
Fuchi believes that logic has "universal applicability" and that this will lead to what he calls the "inference machine"--the culmination of the Fifth Generation approach.
He said that an inference machine should perform many tasks far beyond the capabilities of today's computers--such as translating Japanese to English by recognizing the subtleties of two complex languages.
Although Fuchi's ambitions are broad, he conceded that several important resources are constrained. "We have far less computer scientists in Japan than in the United States," he said. "One important aspect of this project is that it will grow computer scientists as a byproduct."
Nevertheless, Fuchi is confident that, at least on some levels, the Fifth Generation effort will be a success. Although he would not answer the question directly, he said that he expects Japanese computer science to surpass the United States' before the end of this decade.
"My optimism is not proved yet, of course," Fuchi said.