I had the rare opportunity last week to discuss the progress of research on Artificial Intelligence (A.I.) with two Japanese geniuses in this field—Joichi (Joi) Ito, Director of the MIT Media Lab, and Hiroaki Kitano, President and CEO of Sony Computer Science Laboratories, Inc. Ito also serves as an outside director for a number of major companies, including The New York Times Company.

　　　　　It was my first meeting with Kitano. A rapid succession of statements, conveying a rich assortment of innovative ideas about the latest information technology and trends in information-related industries poured from Kitano right from the start of our session.

　　　　　When it comes to A.I., Japan is usually regarded as being “a lap behind” among the world’s advanced nations. Is that truly the situation? Here’s what Kitano had to say:

　　　　　“In the A.I. circles in the US, one is at quite a disadvantage without the ability to comprehend English spoken with a Chinese accent. China has been doing practically everything the US has in terms of A.I. research. Students from China’s major universities—such as Beijing, Qinghua, and Shanghai—have rushed to US shores in large numbers, happy to be able to conduct A.I. research in major universities there.”

　　　　　The recent increase in the number of those researching A.I. in China is astonishing, according to Kitano. In point of fact, Chinese researchers were hardly visible at the 23rd International Joint Conference on A.I., held in Beijing two years ago. But their number has since grown explosively, with talented A.I. researchers concentrated in leading universities like Qinghua, Beijing, Shenzhen, and Hangzhou. Meanwhile, the Japanese presence in the field is barely discernible. There is a fundamental difference in the spirit between Chinese and Japanese A.I. specialists, noted Kitano, explaining:

　　　　　“In India, as well as in China, the fruit of success is truly enormous. In Japan, however, one’s success is at the most rewarded by a promotion to professorship perhaps two years ahead of others. Around March last year, I drew up a basic A.I. research program at the request of the Ministry of Education. I was then put in charge of the program, with the RIKEN Institute of Physical and Chemical Research designated as the venue of our research. But the government allocated only ￥1.3 billion (approximately US$11 million) of the ￥10 billion (US$83 million) requested for budgetary appropriations.

　　　　　“Leaving that aside, however, I wish to point out that the government’s action created a fundamental mismatch, because RIKEN operated on the principle that they must first comprehend the basic factors pertaining to human intelligence before conducting research on artificial intelligence. A.I. today must be an indispensable part of the full-fledged industrial policy of the government. It’s a worldwide battle, and must be treated as such. Unfortunately, the government’s decision indicates little understanding of this stark reality.”

　　　　　What will happen when A.I. is highly developed? Even at this stage, A.I. provides us services via our smart phones, giving us such information as dinner menus and driving directions. Eventually, A.I. will replace humans in certain functions–such as driving a taxi.

When Driving Becomes Unnecessary

　　　　　Explained Kitano:

　　　　　“Baidu, Inc., a leading Chinese search engine, introduced a self-driving car out of the blue last December. Because Beijing is not a party to the 1949 Geneva Convention on Road Traffic, China is entitled to test self-driving cars on its highways at any time. Although the Chinese are ahead of us, a firm we’ve formed, called ZMP, is scheduled to conduct self-driving tests using a taxi in the Shonan area near Yokohama this month.”

　　　　　Automobiles are entering an era in which driving no longer will be required. What then will happen to Honda’s popular “Fun to Drive” advertising slogan? With A.I. technologies advancing rapidly, it will be impossible for the Japanese automobile industry to avoid fundamental structural changes.

　　　　　“Last week, I attended a conference in New York titled ‘Future of Artificial Intelligence,’” Kitano continued. “During the conference, it was reported that Ford had just formed an alliance with Google. However, the attendees at the conference were in no mood to congratulate Google. Instead, the general view was that, by forming the alliance, Ford managed to find a place for its survival. In the same vein, GM has taken steps to align with Lyft, which resembles Uber—a service using a mobile-phone application to facilitate peer-to-peer ridesharing. Lyft is bullish about its position, ready to control GM in due time. A.I. is in the process of prompting a full game change in the US automobile industry.”

　　　　　What gives Lyft the advantage over GM, Kitano says, is its lead in the development of self-driving knowhow. Hardware is of course absolutely necessary, but going forward it won’t generate enough profit for the auto industry. Already, hardware manufacturers are being obliged to follow the directives of A.I. corporations. Kitano is convinced the balance of power will continue to shift in favor of A.I. industries in the future, continuing:

　　　　　“Already some major electrical appliance manufacturers are in the process of functioning as subcontractors for companies like Apple and Amazon. The US auto industry will follow suit without fail. Over the next five to ten years, one must regard as unavoidable the infiltration of key domestic industries in Japan by firms like Apple, Google, and China’s Baidu.”

　　　　　How then is Japanese A.I. research coming along? Ito noted that Japanese firms concentrate almost solely on physically making things, as is demonstrated in the case of Toyota, observing:

　　　　　“Toyota has been making great efforts to make its products intelligent. But they don’t seem to realize that A.I. is all about networking. Therefore, I feel it will be quite difficult for Toyota to develop its own self-driving technology. In point of fact, the better record a corporation has chalked up in the past in manufacturing excellence, the more difficult it is for it to break out of that shell of concentrating on physically making things. ”

　　　　　The greater the success story of a corporation, the more backlash will there be to its efforts to forge ahead with a completely new attempt founded on the denial of the objectives that sustained it in the past.

　　　　　“It may perhaps surprise you,” continued Kitano, “but the US and Japan share the same problem today. The Islamic State has gathered strength thanks largely to the Internet. The US military, on the other hand, has also networked its organization, but there are limits to how much stronger it can become due to its compartmentalized structure.”

　　　　　Ito had this to say about China, which is in hot pursuit of the US:

　　　　　“The Chinese pay absolutely no attention to the rules we in the West abide by. For instance, the US never lets the private sector access intelligence obtained through espionage, stringently implementing a thick wall of law. However, it is a different story in China, where intelligence is shared across the board, as the state, the military, and the private sector are snugly interconnected. When you tell Chinese that the US limits espionage operations to areas of national interest, vowing to never allow its use for commercial purposes, they ridicule the US practice as silly. Why not use intelligence for commercial purposes, they ask, asserting that a nation’s economy is synonymous with its competitive power. They are just not on the same wavelength with us when it comes to democracy and the rule of law.”

“Open Artificial Intelligence”

　　　　　“Naked capitalism”—an irrepressible passion for making money—is the driving force behind China’s rapid A.I. development. Meanwhile, an entirely different wind is blowing in the US. Observed Ito:

　　　　　“In 2014, Google purchased DeepMind Technologies, a British artificial intelligence company comprising only 12 personnel, for a whopping US$500 million. This transaction reflected a fierce competition for highly skilled human resources in A.I. But in this deal you could also see an entrepreneurial spirit to build a better society. In fact, my colleagues—founders of A.I. businesses including Tesla, Y Combinator, and LinkedIn—have announced they will invest US$1 billion in “open source A.I.,” data that will be made available to anyone.

　　　　　“A.I. is fundamentally based on data and algorithms. Google, Facebook, and Amazon now hold a treasure chest of this valuable commodity, and they will continue to accumulate further data, ultimately succeeding in creating the world’s most sophisticated A.I. As a hedge against A.I. being manipulated and controlled globally by a single entity, these companies are nurturing open source A.I. This represents America’s wholesome private entrepreneurship at its best.”

　　　　　What about Japan, then? To state the conclusion first, Japan is not to be underestimated in this respect. Unmanned taxis will be easier to introduce in Japan than in the US. Japanese taxi drivers generally boast a high level of education, allowing for a smooth introduction of new technologies. In Makuhari, Chiba Prefecture not far from Tokyo, there are plans for a test of self-driving delivery trucks. We can expect sophisticated scientific experiments to be backed by our well-educated average citizens.

　　　　　Japan is prepared to keep these experiments open, accumulate data, and utilize that data properly when the right time comes. If at the same time we manage to speedily establish a venue for independent research by able specialists, including the two geniuses introduced in this column, we can be optimistic about the future of A.I. in Japan.

(Translated from “Renaissance Japan” column no. 691 in the February 11, 2016 issue of The Weekly Shincho)