67th MUMEI-JUKU (01/12/2023)
Theme: Generative AI and Japan’s Strategy
Guest Speaker：Dr. Daisuke Okanohara, Preferred Networks / Co-Founder, Chief Executive Researcher, Preferred Computational Chemistry / Chief Executive Officer, Preferred Elements / Chief Executive Officer
Obtained Ph.D (Information Science and Engineering) from the University of Tokyo in 2010. In 2006 while in graduate school, he founded Preferred Infrastructure Corporation, the predecessor of Preferred Networks (PFN), with Toru Nishikawa and others in 2006.
In March 2014, he founded PFN to accelerate the practical application of deep learning. Currently, as the Chief Research Officer of PFN, he is working on research on deep learning and its practical application. He also serves as President of Preferred Computational Chemistry, Inc. which markets Matlantis, a general-purpose atomic-level simulator jointly developed by PFN and ENEOS, and Preferred Elements, Inc. which develops multimodal infrastructure models. Ltd. which develops multimodal infrastructure models. He is currently writing a series of articles in Nikkei Robotics titled “AI Frontline. His publications include “Is Large-Scale Language Model a New Intelligence?” (Iwanami Shoten), “Diffusion Model” (Iwanami Shoten), “The Forefront of AI Technology: 73 Advanced Technologies for Reading Future AI” (Nikkei BP), and “Technologies Supporting Deep Learning” (Gijutsu Hyoronsha)
Commentator: Dr. Junichi Tsujii
Information Scientist; Fellow, National Institute of Advanced Industrial Science and Technology (AIST) (Former Director, Artificial Intelligence Research Center)
Completed graduate studies at Kyoto University. He is a PhD in Engineering. He has been an associate professor at Kyoto University, a professor at the University of Manchester since 1988, a professor at the Graduate School of the University of Tokyo since 1995, and a chief researcher at the Microsoft Research Institute Asia (Beijing) since 2011, current position after these. He also holds a professorship at the University of Manchester.
He has also served as president of the Association for Computational Linguistics (ACL), the International Association for Machine Translation (IAMT), the Asian Federation of Language Processing Societies (AFNLP), and the Association for Language Processing, President of the International Committee for Computational
Linguistics since 2015. He has received numerous awards, including the Medal with Purple Ribbon, IPSJ Achievement Award, Funai Achievement Award, Okawa Award, AMT (International Association for Machine Translation) Honor Award, ACL Lifetime Achievement Award, and the Order of the Sacred Treasure, Gold Rays with Rosette.
<Opening Remarks by Dr. Takeda>
Today’s topic, “Generative AI and Japan’s Strategy,” is a very fascinating topic. there have been many events surrounding OpenAI. In November, OpenAI’s board of directors suddenly fired Sam Altman. As a result, the majority of employees indicated their intention to resign along with Altman. While Microsoft was attemot to drow him into their company , Altman was reinstated as CEO of Open AI and the board of directors was revamped as well.
Generative AI is the English term GAI, but there is also the concept of AGI (general-purpose AI), AI with human-like intelligence. Altman has stated that he will start working on AGI, the original purpose of OpenAI, immediately after he returns to work. I myself think that it would be impossible for AI development to progress so easily all at once, but on the other hand, Altman is a superman, so I wonder if he has a hand in it.
However, I believe that a very big leap has also been made in current Generative AI (GAI). It is in this context that Mr. Okanohara, co-founder of Preferred Networks, will speak to us. Last month, I had a chance to exchange views with a top executive of a telecommunications company about the next generation of AI and other important cutting-edge technologies. He told me that Preferred Networks is the company that could challenge the world in AI, especially in the area of GAI. After Mr. Okanohara’s lecture, we will hear comments from Dr. Junichi Tsujii, Fellow of AIST (former Director of Research Center for Artificial Intelligence), who is actually Mr. Okanohara’s mentor. In Japan, there are many “AI scholars” who talk about various aspects of AI, but I believe that Dr. Tsujii is the one who truly understands the whole picture of AI in Japan.
<Professor Takeda’s comments after the lecture>
Mr. Okanohara mentioned the importance of making rules and fighting for the world, and Dr. Tsujii mentioned the importance of making common friends, and I totally agree with him. After the emergence of OpenAI, I feel that the world has moved into a different next era. In this trend, the Biden administration of the U.S., European Commissioner Breton of the EU, and the Xi Jinping administration of China have started their own moves. In this context, Japan must consider how it can cooperate with like minded countries and how it can establish a domestic structure for this purpose.
In addition, it is very important to establish AI standards and to cooperate with other countries. Japan alone cannot set the standards in AI. The U.S. also acknowledges that it cannot develop generative AI on its own. The U.S. is trying to create a cooperative framework for the expansion of generative AI beyond the existing government system. In this context, Japan must rethink its industrial policy.
The Xi Jinping administration in China is aware of the crisis that the trade deficit in semiconductors is many times larger than that in crude oil, and is desperate to develop semiconductors and to create rules for this purpose. As Mr. Okanohara mentioned, the digital deficit has started to exceed the oil deficit, and we will need to think again about our national development and create a digital system to make Japan stronger. An industrial policy is needed to achieve this. OpenAI has about 770 employees, but it runs the world. I believe that the time will surely come in Japan when companies started by young people like Mr. Okanohara and others like him will move the world. If this does not happen, I do not see a future for Japan. To this end, the role of the Japanese government will be crucial.
The Japanese government must talk about how to cooperate with like minded countries, such as the U.S. and Europe, after Generative AI is released. As I have little basic knowledge about the developments in China, I would like experts on China to conduct a survey on the kind of competition and cooperation that can be expected. I would also like the government to establish a system for this purpose.
The Hiroshima AI Process, launched in response to the results of the G7 Hiroshima Summit last May, established guidelines for the direction of AI. This is a great achievement. And on October 30, President Biden declared his readiness by issuing a presidential decree on a wide range of areas where AI will have an impact, including AI, security, and risk. In this sense, we hope that the Japanese government will take care of industry and also universities. I am very grateful that METI has made supercomputers available to many researchers. Otherwise, research would not be possible at universities. I also believe that the way to promote collaboration among countries or between Japan and the U.S. should be done on a grass-roots basis. I think the biggest thing is to enlist the help of university professors in CET and other programs.
Also, we tend to forget the importance of standard making and rule making, but the US, Europe, and China are putting huge numbers of people into international positions. China is also strategic in this regard. I would like to ask you to once again try to maintain Japan’s strength in the world. Even for a single semiconductor, companies are not only Micron, but also Intel, AMD, etc. We should not be satisfied with just one. We should not be satisfied with just one. As Dr. Okanohara and others have said, we must move up from one to the next. I would like to ask you all for your best wishes and good guidance for the next generation.
