Outline of JSAI


Inaugural Address of President


What should JSAI leave behind the AI boom?

Seiji Yamada
(NII/National University/Tokyo Institute of Technology)

I am Seiji Yamada, just appointed as President of the Japanese Society for Artificial Intelligence (JSAI). It is a pleasure to work with you. Upon my inauguration, I would like to address a few words to you.

Artificial intelligence (AI) is said to have been experiencing its third boom for the past several years. To be honest with you, I forecasted that the boom would have peaked out by the time I took office as President. Fortunately, however, my forecast has turned out to be wrong. The AI boom is continuing and the mass media are reporting news about research results in AI and business application of AI every day. As with booms in research, there is a time lag of a few years in any boom in Japan from the boom in the United States. Therefore, the AI boom in Japan will continue for a few more years, i.e., it will continue during my term of office. Then, in such AI boom, what JSAI should do and should leave behind. I think that considering and implementing the answers are the priority duties for me during my term as President. Having said that, however, it would be too late for the realization in my two-year term if I start thinking now from zero. Therefore, I am already pondering some ideas. However, they are just ideas: I cannot guarantee their implementation and I do not like to be bounded by what I have declared. Consequently, I would like to talk about the ideas merely as goals to achieve.

With the current AI boom, the number of members of JSAI is increasing. Consequently, the revenue of JSAI is stabilizing, which leaves a margin to support members’ research and development activities financially. Given these circumstances, I would like to build a framework to promote global pioneering and highly original research originating in Japan and financially shore up as JSAI as a whole, which many thoughtful AI researchers in Japan and I myself feel to be important. This is what I stated in the preface of the Journal of JSAI, Vol. 30, No. 1 (January, 2015). I would like you to read it in terms of the details and methodology of the idea. In any case, AI research in Japan is still insufficient as research that proposes new frameworks and research that can create new fields of research. In fact, research trends in AI research almost all originate from Western countries. I cannot suppress the impression that AI research in Japan is chasing after research trends that have already been prevalent. I am aware that contributing to breaking through such a situation is the primary role of JSAI.

Next, I would like to create a system to promote matching in application development between research at Japanese universities and enterprises. This might be prejudice, but I feel that in the case of Japanese manufacturers, when conducting applied research in AI, they are more likely to point to joint research and development with universities or research institutions abroad than with universities or research institutions at home. Furthermore, when watching demonstrations of AI applications that are performed in current IT ventures, at a research level, I feel a sense of déjà vu of various frameworks that universities and research institutions in Japan conducted more than 10 years ago: I feel that there is a great deal of potential in joint research and development by Japanese enterprises and Japanese universities and research institutions now and beyond. Because there is no system to intermediate them, I think it leaves much more room for vitalization. Probably, among the largest organizations to which Japanese AI researchers and developers belong is JSAI. Consequently, I believe that systematically doing such matchmaking has great meaning.

In addition, I have gradually started thinking about policy changes in the administration of JSAI that I cannot make public here. However, the idea is at a stage that I am thinking to move forward step by step in the help of members of JSAI administration, the Board of Directors, and the Administrative Office.

Finally, although this is my personal goal, I would like to balance the presidency and the research profession: Freeing myself from a common pattern that once becoming a president the person cannot carry out research or suspend research, I would like to promote research that I am engaged in during my term as President as before, or more than ever, if possible. Of course, I think that it would be difficult, as many people point out, to promote research while never neglecting the duties of President. However, considering the remainder of my own life as a researcher, I feel that I should find some way to avoid the two-year stagnation. This is an almost self-centered goal, but I would like to seek a path toward the realization with the aid of research collaborators and members of JSAI.

I have talked about my aspirations for matters of JSAI and private matters. The idea to be the base for all is that while renewing my awareness of the spirit “JSAI is for its members,” I would like JSAI to be an organization gathering needs from Members, Supporting Members, and Student Members and responding to them to a maximum degree. Please feel free to provide your comments to JSAI. I welcome your comments to be sent either via Directors or the Administrative Office.


Address of Editor-in-Chief

Upon my inauguration as Editor-in-Chief

Happy Academic Life 2016: Time to burn our lives

Hiroshi Yamakawa
(DWANGO/Whole Brain Architecture Initiative)

Ten years ago, I created a game called “Happy Academic Life 2006” with fellow JSAI researchers in the winter years of the Japanese Society for Artificial Intelligence as the 20th anniversary memorial project, “AI wakatekenkyusha no tameno kyariadezain noryoku ikusei jigyo: Kofukuna kenkyujinsei ni itaru michi.” The message I wanted to send at the time was that for young researchers, publishing papers and building their networks in academic societies could generate a virtuous circle in their future career. The message used to be applicable to many other occupations.

However, such a traditional view of career design has begun to crumble because of the phenomenon by which various human intellectual faculties can be substituted for AI that many experts, including us, have created. This naturally affects the careers of AI experts themselves, but I believe that because we know AI, we can cope well with it. Now in 2016, I was given the role of Editor-in-Chief. I feel probably I was given an opportunity to think about how AI researchers and engineers establish happy careers with many members of JSAI.

This is my personal opinion, but I would like to think about how research should progress in the AI community from a perspective of career formation below.

It might sound a little rough, but let us assume that AI consists of a comprehensive adult AI handling concepts and symbols that humans can consciously operate and describe, in addition to a fundamental child AI connecting real world information and concepts that humans obtain unconsciously. In the previous situation in which the adult AI took the initiative, human intervention needed, in large part, data preparation and knowledge building to be given to them: experts compensated for a lack of intelligence in children. However, this third AI boom is realizing basic intelligence, such as recognition and motion that children acquire at the developmental stage by combination of generic machine learning, including deep learning and reinforcement learning. In this way, the reduction of a scope that requires the help from experts increases the business value of AI and spurs investment.

To change the subject, experts (not limited to AI researchers) contribute to society through which they understand the subject based on some sort of specialization and produce value using the specialized intellect. However, if amateurs can use professional capabilities by inputting data and computational resources to AI, then the value to demonstrate expert intellect that has been accumulated over time will diminish. For example, ten years ago, the introduction of machine learning into computer shogi invalidated the knowledge of movement rules that had been reinforced strenuously by hand until the time and destroyed the value of specialization to create it. In addition, in the computer vision, the realization of generic object recognition eliminated most of the feature quantity that experts had designed and destroyed the specialization to design it. When an ongoing technology to extract relations between concepts from pair data of image and caption has been established, ontology and even the specialization to design it might become unnecessary.

Here, if we see AI as a car, the engine that propels the car can be characterized as machine learning. Obviously, “car and society” or “AI and society” are interesting themes, but themes such as “engine and society” or “machine learning and society” look faded: I wonder if AI provides valuable mechanisms and frameworks for machine learning.

In the course in which AI irreversibly erodes specialization like this, how should AI experts establish their careers? Under current circumstances, performance of generic machine learning is improving rapidly, in which those who have a good engine can have an advantage in business. Moving forward in such areas is important, but this is an area in which large amounts of computational resources are often required; competition has been fierce. For that reason, it is necessary to consider other possibilities from a perspective of career formation.

How do you feel about applying machine learning as a tool in an individual area of some sort? Even though adaptation to changes is constantly required, it seems promising as engineering. On the other hand, researchers will be required to have thorough knowledge of each area and then to produce original results. Knowledge and know-how that AI experts have polished up over the past 60 years are a deep pool. For example, the capability of redefining problems and knowledge of a particular field from a meta perspective will still require time, even if realized by machine learning. However, considering that one’s own specialization will be replaced by AI sooner or later, rather researchers who have replaced it by machine learning ahead of other countries might leave their names in the world. If that is the case, then there can be a strategy that prepares to seize the moment. This can be characterized as the time to burn our lives that we can enjoy because of having fortunately been involved in AI.

I might receive a reprimand that such is a somewhat biased view. However, considering such a sense of crisis, I would like to make the Journal of JSAI and Transactions of JSAI more valuable with members of the Editorial Committee.


Past presidents


  Name incumbency
The 16th Seiji Yamada 2016.6.24-
The 15th Hitoshi Matsubara 2014.6.13-2016.6.24
The 14th Takahira Yamaguchi 2012.6.14-2014.6.13
The 13th Toyoaki Nishida 2010.6.10-2012.6.14
The 12th Koichi Hori 2008.6.12-2010.6.10
The 11th Riichiro Mizoguchi 2006.6.8-2008.6.12
The 10th Mitsuru Ishizuka 2004.6.3-2006.6.8
The 9th Hozumi Tanaka 2002.5.30-2004.6.3
The 8th Yoshiaki Shirai 2000.5.26-2002.5.30
The 7th Katsuhiko Shirai 1998.6.18-2000.5.26
The 6th Hidehiko Tanaka 1996.6.26-1998.6.18
The 5th Shuji Doshita 1994.6.22-1996.6.26
The 4th Masamichi Shimura 1992.6.25-1994.6.22
The 3rd Saburo Tsuji 1990.6.23-1992.6.25
The 2nd Setsuo Osuga 1988.6.24-1990.6.23
The 1st Akio Fukumura 1986.7.24-1998.6.24