Interest in AI spread throughout society, and the third AI boom came to an end in the 2010s, and the AI boom is beginning to settle down. At the same time, DX is attracting attention in society, and a larger-scale transformation using data and digital technology is beginning to occur.
Now that the attention in the AI field has settled down, it is necessary to turn our attention to areas other than deep learning among the wide range of AI research fields, and to further invigorate research.
Against this background, the Japanese Society for Artificial Intelligence has compiled and released the “AI Map β,” which maps various research themes in the AI field.
Regarding the fields and outlines of AI research, this is also published in the article.
Explains the fields and outlines of research in AI｜Introducing laboratories and research companies>>
This time, we interviewed Mr. Morikawa and Mr. Tsutsumi of the Japanese Society for Artificial Intelligence who were involved in “AI Map β”.
Mr. Fujio Tsutsumi (left) : In March 1990, completed the master’s course at the Department of Information Engineering, Graduate School of Engineering, Kyushu University. In April of the same year, he joined the Central Research Institute of Electric Power Industry. Currently, he is a research advisor at the Central Research Institute of Electric Power Industry. From July 2020, Deputy Director of the Center for Energy Innovation Emergence (current position). For two years from June 2018, he served as the chairman of the AI Map Task Force as a board member of the Japanese Society for Artificial Intelligence. Doctor of Engineering. His research interests include artificial intelligence, image processing, and user interfaces.Koji Morikawa (right) : In September 1996, completed the doctoral program at the Graduate School of Engineering, Nagoya University. After working as a research fellow of the Japan Society for the Promotion of Science, he joined Matsushita Electric Industrial Co., Ltd. (now Panasonic) in April 1997. He is engaged in research and development of advanced technologies such as artificial intelligence and brain signal processing. Since April 2020, he has been CTO of Connect Co., Ltd. Currently, he is promoting the development of rehabilitation equipment that utilizes the Brain-Machine Interface. He is the vice president of the Japanese Society for Artificial Intelligence. Doctor of Engineering.
- An AI map that provides a bird’s-eye view of the spread of AI research
- The first step in using AI is to learn about AI technology
An AI map that provides a bird’s-eye view of the spread of AI research
ーーWhen did you start working on AI Map β?
Mr. Morikawa: We started about three years ago. In June 2019, we released “AI Map β” and in June 2020, “AI Map β 2.0”.
At first, it was difficult to gather members. We created a task force of less than 10 people and met regularly to proceed with production.
ーーPlease tell us how you decided to create “AI Map β”.
Mr. Tsutsumi: We started production for the purpose of “expanding AI research.” In order to continuously upgrade the version, Mr. Morikawa said “You should add β”, so I named it β.
Originally, former vice president Tsumoto (Shimane Univ.) and former president Yamada (NII) of the Japanese Society for Artificial Intelligence said, “AI is difficult to understand because there is no overview of AI.” The Japanese Society for Artificial Intelligence also tried many times to solve this problem, but it did not go very well. When creating the map, I had the cooperation of many researchers, but as the discussion progressed, the content unavoidably became complicated and bizarre.
It was a struggle because the content was not organized, but with the tide of deep learning, I came to feel the need for AI maps even more. When we started producing “AI Map β”, deep learning was attracting a lot of attention, so we didn’t pay attention to other technologies.
Of course, the AI field is not all about deep learning, so I thought that the Japanese Society for Artificial Intelligence should create a diagram that would give us a bird’s-eye view of the spread of AI research and the content of research.
ーーWhen you actually created the AI map, how did you feel about the number of fields and complexity?
Tsutsumi: Looking at the research in the AI field, the amount of documents for research papers is enormous, and it has become unmanageable. Also, since the granularity of words varies, that point was also a challenge.
“AI Map β” is divided into four diagrams instead of just one diagram. First, I made a map and shared it with the members, who then created their own map. As a result, we were able to express the whole picture of AI because all four of us created it from completely different perspectives.
Vice Chairman Morikawa, who saw it, said, “If there are various ways to approach it, let’s publish all four maps without narrowing down to one.” I made it
I think everyone in the Japanese Society for Artificial Intelligence recognizes that deep learning technology is wonderful. However, if we focus entirely on deep learning, our understanding of AI will not progress.
The first step in using AI is to learn about AI technology
ーーDo you think that we should also engage in research other than deep learning with a view to social applications in the future?
Mr. Tsutsumi: That’s right. It is up to the user to decide how to use AI, so I think the first step is to learn about research content and how it can be used in society.
I think that there are some researchers who do not have enough budget. However, even in such a case, it would be wonderful if you could believe that “this field is a technology that expands human possibilities, and there are ways to use it,” and open up a new research field.
Mr. Morikawa: When AI experts emphasize, there are things that amateurs cannot fully understand, so there are times when opportunities are lost. Even though the content of the research is wonderful, there are times when the budget cannot be allocated, so AI research will not progress as it is.
Therefore, it was decided to create “AI Map β” with a target set for AI beginners. I became interested in the AI industry by telling people, not people working on AI research, but people who thought, “AI is amazing and interesting,” and said, “It’s useful and effective in this way.” People can deepen their understanding and think about how to use it in conjunction with themselves.
In 2019, the Japanese Society for Artificial Intelligence held an event called “Activities of the AI Map Task Force -Overview of AI Research for AI Beginners and Researchers in Different Fields-“.
At this event, the AI map β version and AI research content are released to those who are interested in AI, and by creating opportunities for opinion exchange, it is a mechanism that allows you to get a bird’s eye view of AI research.
▼ Reference article
ーーI feel that it is quite difficult to select the AI technology to use. What do you think about it?
Mr. Tsutsumi: That’s right.
Therefore, in “AI Map β 2.0”, we released “Task Cards”. In the “issue card”, many issues related to AI are described.
In some cases, researchers and engineers who are researching specific areas do not understand the issues and knowledge related to other areas. I understand that researchers and engineers want to use the technology in their own field to solve problems, but if there is another technology that is optimal, I think that recommending it to users will further advance the use of AI. increase.
I hope it will be an opportunity to change your perspective and reconsider the issues from a bird’s-eye view.
Morikawa: As the use of AI advances, I think it will become so obvious that we will no longer be able to tell the difference between “AI or not”. However, since new research is progressing at academic societies, it would be great if we could find areas of blue ocean and discover new ways of using them by changing the perspective while looking at the map without missing the timing.
As the understanding of AI and the application of technology progresses among users, I think that there will be talk of “Isn’t it necessary to use AI here?” In most cases, when AI is applied to the field, various improvements are made. Each time, the people in the field put out their wisdom and make improvements.
From now on, we will take advantage of it and enter the next step of researching AI in the next layer. Since AI evolves by incorporating human intellectual activities, I believe that if such a cycle can be achieved, AI research will develop significantly during the fifth and sixth booms.
ーーWhat kind of response did you receive when you actually released the AI map?
Mr. Tsutsumi: We allowed the reprinting of the map, so it was quoted by many media and books. At the time it was released, I think it had an impact in various fields, but one year after the release, I heard that some AI beginners and researchers in different fields were wondering, “What should I do next?” I’m here.
These people have a clear problem and want to do something, but they don’t know how to connect with the AI map. Unfortunately, I ended up looking at the AI map.
ーーWhat do you think about the advent of the AI winter era?
Morikawa: Next time, I hope that AI technology will take root in society. I don’t want AI technology to lose its focus, but rather to become the norm.
On top of that, I hope that the Japanese Society for Artificial Intelligence will play a central role in enlivening the blue ocean field.
Currently, the integration of inductive approaches such as deep learning and rule-based deductive approaches is beginning to progress, and the application of the technology is going well. We will continue to improve it by adapting it to the site. I would like everyone to research the next layer of AI.
In particular, various technical issues are discovered in the field. I would be happy if it would be returned to AI researchers and the next AI research would develop.
ーーHow do you plan to improve in the future?
Tsutsumi: From now on, I hope that AI maps will become a tool that connects researchers, AI venture managers, and students.
If so, it’s worth creating an AI map. It’s still just a map, but if it plays a role like a platform, I think it can make a big impact.
Because AI technology has the potential to fundamentally transform various fields, there are many different research areas mixed up behind it.
The AI boom sparked by deep learning will continue to spread to a wider range of research fields, leading to further research development.