Digital Extraordinary Administrative Advisory Committee Working Group (3rd)
Overview
- Date: Tuesday, February 22, 2022 (2022), from 4:00 pm to 6:00 pm
- Location: Online
- Agenda:
- Opening
- Proceedings
- Hearing from companies, etc. that provide digital technologies that can be used for regulation reviews
- Exchange of opinions
- Adjournment
Materials
- Agenda (PDF/47KB)
- Document 1: Documents submitted by Professor Hiroshi Esaki (PDF / 957 kb)
- Exhibit 2 Submission from the World Economic Forum Japan Center on the Fourth Industrial Revolution (PDF / 1,397 kb)
- Exhibit 3: Data submitted by Vantiq Co., Ltd. (PDF / 7,409 kb)
- Exhibit 4: Documents submitted by SoftBank Corp. (5x5 Technologies) (PDF / 3,862 kb)
- Exhibit 5: Documents submitted by Professor Yutaka Matsuo (PDF / 216 kb)
- Exhibit 6: Documents submitted by GRID Co., Ltd. (PDF / 28,407 kb)
- Material 7: Materials submitted by DeepX Co., Ltd. (PDF / 2,687 kb)
- Appendix 8: Image of Correspondence between Digital Technology and regulation Review (PDF / 297 kb)
- Minutes (PDF/660KB)
Minutes, etc.
Date
Tuesday, February 22, 2022 (2022), from 4:00 p.m. to 6:00 p.m.
Location
Online Meetings
Attendees
Chairman
- Su Beng Seiichi Kobayashi
Members
- Junji Annen (Attorney, Professor of the Graduate School of Law, Chuo University)
- Tatsuhiko Inadani (Professor, Graduate School of Law, Kyoto University)
- Katsuya Uenoyama (President of PKSHA Technology, Inc.)
- Takafumi Ochiai (Attorney at law, Atsumi & Sakai, Foreign Law Joint Enterprise)
- Akiko Sugawara (Managing Director and Head of Policy Planning, Keizai Doyukai)
- Katsunori Nemoto (Senior Managing Director, Japan Business Federation)
- Masakazu Masushima (Attorney-at-Law, Mori Hamada & Matsumoto)
Minutes
Secretariat (Suga): Now that it is on schedule, I would like to open the third session of the "Digital Extraordinary Administrative Advisory Committee Working Group."
Nice to meet you again this morning.
Members are invited to participate online this time as well.
In addition, members of Ochiai, Masushima, and Uenoyama are scheduled to participate from the middle of the course.
First of all, following on from the morning, I would like to ask for an address from Senior Vice-Minister for Digital Affairs Kobayashi, who is the chair of this working group, prior to the opening of the meeting. Thank you.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, , the third meeting will be held consecutively, so I would like to ask all of you to continue.
Welcome to the Digital Ad Hoc Working Group, everyone from private sector who will be participating.
I think there are people who have not seen the second edition, so when I shared it, I heard about existing technologies that can be used immediately for implementation in the second edition. I thought it was very helpful that the Ministry of Land, Infrastructure, Transport and Tourism arranged the technologies in a catalog in advance and made a visualization that they can be used immediately.
In addition, I believe that the explanation from private sector companies was based on the idea of substituting machines and technology for the parts that had been done by people, premised on existing methods, and that they were taking on the challenge of doing so.
I believe that all of you who attended the third session today are thinking from the perspective of what it would be like to do with technology or machines, rather than doing it with people or even methods that were originally used. I hope that we will be able to switch our minds from the previous discussion and discuss new ways of regulation from here. I have great expectations for you. Thank you very much for your time today.
Secretariat (Suga): . Now, I would like to begin today's agenda. I would like to ask Vice Chairman Annen for your continued support.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister . Thank you for your cooperation.
In the first half of the third session, we will hear from four parties: Professor Ezaki of the Graduate School of the University of Tokyo, Project Director Hirayama of the Japan Center of the Fourth Industrial Revolution of the World Economic Forum, Vantiq Co., Ltd., and SoftBank Corp.
In the second half, we plan to hear from The University of Tokyo Graduate School, Professor Matsuo, GRID Co., Ltd., and DeepX Co., Ltd.
As the Senior Vice-Minister said, I am looking forward to hearing from people who are a kind of game changer.
First of all, I would like to have a presentation from Professor Hiroshi Esaki of the University of Tokyo. Mr. Ezaki, nice to meet you.
Ezaki Member: . I would like to briefly talk about how the external environment has changed in Rule digitalization because it is design and online.
First of all, what is digitalization? Basically, digitalization is that the function is separated from the hardware. Therefore, it is essentially digitalization that the existing combination will change or may increase, and that it is necessary to understand that unnecessary things will be removed or eliminated.
From the perspective of genes and human-like survival machines, if we think that programs and rules are genes and hardware is a survival machine, I think it is good to think that this COVID-19 is a digitalization in which great things begin to happen everywhere because RNA can move freely.
The procedures for setting rules are written at the bottom, but all procedures must be considered from the rules based on physical assumptions to the settings based on cyber assumptions.
As for the profit structure, when making the rules, it is necessary to embed the fact that the rules and incentive design from the exclusive structure to the shared and altruistic structure are necessary, and the fact that the single KPI must be changed to multiple KGIs.
What is meant by further progress in softwareization is that it will change from the Tech-Driven water-flow type to the Issue-Driven user-driven type. To put it another way, even if the cutting-edge research to date is Ready/Advance, it is the same as why Japan lost in the business structure. I think it was not possible because Japan did not realize that it was changing from Tech-Driven to Issue-Driven.
The second thing that has changed is that we can now take data, which is a big change. It is very important to embed the fact that we can continue to take data in the rules.
As a business incentive, the SDGs have a very strong bias, so we need to change the design from a short-term premise rule to a medium - to long-term premise rule, and from a replacement premise that is assumed to be disposed of until now, to a recycling premise.
In terms of external requirements, technology can be changed from a human-based system to a robot-based system as mentioned by the Vice Minister. This will match declining birthrate and aging population very well, and in many aspects, cutting-edge technology, mainly digital technology, exceeds human abilities. This can be regarded as supernatural powers, and we must make institutional design aware that humans will acquire new abilities.
In other words, we must consider a system design based on the premise that TO BE will come out, not AS IS. By changing the environment setting based on the premise of replacement to the premise of update, it means that we should set the rules based on the premise of system asset management.
On the other hand, what has not changed is that hardware is fast and software is slow. It has been common sense until now, but the fact that hardware is evolving at a tremendous pace has not changed at all. Then, what is important is that the ease of moving has increased by about 1,000 times in 10 years. It has not changed at all in the past 40 years. This means that we can think that people are heavy but things have become light. Thanks to technology, there are so many rules that presuppose that things will not move unless we make institutional design based on moving.
As I write on the bottom, IoT, from the premise that things are connected, we think that things are connected. It can be seen that what was premised on incomes will change to digital signatures, and what we saw with the eyes that we have a story about today has changed to be seen with technology. The obstacles to TO BE, that is, innovation in the future, are AS IS, current regulation, business practices, and momentum.
Next, I would like to talk about Dig Once. The common sense up until now has been that if we reduce the construction work, we will be scolded and we cannot raise the unit price, so we cannot do it, but I think this assumption will change fundamentally.
Therefore, in order to realize this, it is important to consider how to "By Default" the design and procurement Rules. As I am trying to share them, I would like to make my first introduction on how to design systems based on the premise of globalization and the unbundle of hardware and software, in which design must be based on the premise that hardware and software can be moved.
That's all from me.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , thank you very much.
Ladies and gentlemen, please let me know what anyone has to say.
Member Inadani: Kyoto University.
Thank you for your very exciting talk. I would like to ask about what you said in your talk, from things to things and issue based, in relation to what regulation should be. For example, in the morning, in order to replace visual inspection with digital technology, I think there was a talk about image recognition, which is a method of image recognition that replaces eyes. As Dr. Masushima said, in the end, I think it is necessary to change the approach in accordance with the essence of regulation's purpose, such as finding defects or finding dangerous spots.
At that time, I thought a little during my talk this morning that there is a gap between the ways in which humans are good at and the ways in which machines are good at. As I was asking you, I had the impression that it would be good to redefine the essence of regulation objectives in a way that draws out more and more the ways in which machines are good at, and to create rules in line with that. Am I correct in understanding that way?
Ezaki Member: There are many options for how to realize what you want to do, and the rules must be set so that you can do it. Actually, that is the purpose of the Internet, but design is done from the beginning so that the hardware can be changed into anything. Then, applications can freely select it.
That is where I wrote the relationship between genes and survival machines that I talked about. Genes can be placed in various individuals. Or, individuals themselves can be selected from genes. If individuals are excellent, we will consider how to bring in genes. In that case, it is very necessary to have a rule design that programs and executions can always be interchanged. If this is done, new innovative hardware and software will come out, and resources can be used very effectively. I think three people will tell us examples of this.
Member Inadani: .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , may I speak with you?
I think it was so shocking that we were told that the jobs of you lawyers would disappear. It is a wonderful future that we have been implementing based on the exclusive right of ownership, and that we have been premised on the self-evident distinction between ourselves and others, and it seems that it will disappear.
Ezaki Member: , but the barriers will be greatly reduced.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister That's right. I thought so. I think I will be shocked all night today, so Mr. Ezaki, I would like you to participate in the discussion again. Thank you very much.
First of all, I would like to hear Mr. Ezaki's speech. Then, I would like to have a presentation from Mr. Yuta Hirayama, Project Director of the World Economic Forum Japan Center for the Fourth Industrial Revolution. Thank you very much.
Hirayama Member: Nice to meet you. I would like to share some materials with you. Are you all here? Let's get started. Thank you for your introduction. My name is Hirayama from the World Economic Forum Japan Center of the Fourth Industrial Revolution.
Today, I would like to talk about the Dig Once Policy that Mr. Ezaki mentioned earlier and the current situation of road management in Japan.
First of all, the G20 Global smart cities Alliance was launched in 2019 under the Japanese presidency of G20. Now, we are working together with 36 local government around the world to create a model policy. smart cities is a problem that cannot be solved by one person alone, whether it is security or privacy, so it is an initiative project to work together on such a matter.
In 2020, we announced a model policy called the Dig Once Policy. This is a concept that has spread widely in Australia, the United States, and other countries. In countries with vast land, for example, it is difficult to install optical fiber cables for one house. At that time, water and electricity lines are usually installed underground on roads. When maintenance work is carried out on these lines, the government obliges the operator to obtain information on the construction work and install optical fibers. In that case, it is not necessary for everyone to dig the road many times for public works. Instead, it is necessary to dig the road only once. This is the Dig Once Policy.
If we look at the current situation, we can say that the situation of roads is extremely difficult. It is still the case that they are managed by paper, such as the Road Register and the Appended Figure of the Road Register. Whether or not it is a problem to be managed by paper is not a problem if the accuracy is high. However, when I actually heard the voices of the site, when they dug a hole as written in the Appended Figure of the Ledger, they thought it was actually 1 meter below, but it was 50 centimeters below and damaged the pipe. If the pipe is damaged, for example, if it is an optical fiber network, the Internet around it will be stopped. It is a very serious problem.
As of 2022, there were about 126 accidents nationwide. You may think this is a small number, but when you actually look at it, it says that there were no instructions to dig up to 50 centimeters around the artifact. In other words, the reason why there were no accidents was that at the site, there might be something 1 meter below, so they dug up to 50 centimeters before with heavy machinery, but they dug up the remaining 50 centimeters by hand. This is what is done in road construction throughout Japan. Considering this inefficient situation, they think that it would be better to improve the road a little more.
I just happened to have a global reference case, so I would like to introduce it. It is French PCRS, and it is plans de corps de rue simplifiés. It is difficult to pronounce, so I cannot say it, but even if I search for French PCRS, only information on PCR tests can be found, so when I search, I need you to type in all the words. However, what I am saying is that it is a law or rule that requires business entities that handle basic local government and utilities such as electricity, gas, sewage, and underground facilities to deliver digital data to the ledger.
I have heard that implementation will be completed in French cities by 2019, and that efficiency will be expanded to the whole of the country by 2026. According to Ms. Laika's presentation, efficiency will increase the number of French airports by about 60%. Actually, I have not been able to confirm the progress of Okinawa, but I believe that this is a very useful example for Japanese people.
Based on this, what I would like to suggest or talk about is that the age of digital Taiko-kenchi may be just around the corner. Now, such information on underground installations is being exchanged on paper, but when I actually hear from the site, survey technology has improved, so we are measuring with digital and three dimensional information. However, when we deliver it, we take the trouble to transcribe it on paper and deliver it to the road administrator of each local government. There are national roads, prefectural roads, municipal roads, and in some cases, private roads, and the management methods are completely different, so they are all separated.
If possible, it is naturally more efficient to unify these things into one on a digital platform, and then there will be no unreasonable things. When I write digital data, it may seem as if I deliver it in PDF, but I think it is extremely important to deliver it in a state where it can be understood by a machine, even if a person can see it, in a list of 0 and 1, which is called binary data.
As mentioned in (ii) and (iii) of the materials, I believe that various discussions on real estate ID, space ID, and building OS are taking place in parallel in the government. Mr. Digital Agency will firmly lead the creation of a base registry that links these with digital information in the area of integrated ledger development. Mr. Ezaki talked about the unbundle of hardware and software earlier. I think that road administrators can continue to be as they have been until now, but if ledger management can be realized in Digital Agency, I think the world will become very efficient.
Finally, I think it is important to advance such efforts through public-private cooperation, but in our organization, we are working on the whole organization while advocating agile governance, so I would be grateful if I could be of any help. That is all.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Hirayama, Project Manager. This is also a very shocking story, and I think it is amazing. If you have any comments, please. Mr. Inadani, please.
Member Inadani:
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister . Mr. Nemoto, please.
Nemoto Constituent: Thank you, As Dr. Annen said, I think what you just said is a project that regulatory reform is also taking up. I would like you to do it in the form of digital twins not only in the ground space but also in the underground space.
In addition, if I were to add one point in my explanation, road construction, which is a nuisance to residents, is extremely wasteful in terms of people's lives because the work of excavating the same part over and over is repeated. I would like to emphasize that the improvement of digital data will be considerably eased and will be extremely beneficial for people's lives. Your proposal is for Digital Agency, but it is being carried out in various ways in various places, so I would like to ask for a summary within the government departments.
In addition, I would like to say that you can apply for road occupancy online, but if you ask us to come and get the permit or if you send it by e-mail, we will be told that we cannot do construction unless we print it out and post a paper copy on the site. I would appreciate it if you could do something about it.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister .
I'm a lawyer, so I briefly thought of the registry. It might be better to re-create it from scratch.
Thank you very much. I would appreciate if you could participate in the discussion. Thank you very much, Mr. Hirayama, Project Manager. I will be in a hurry, but next, I would like to have a presentation from Mr. Jun Kawakita, Representative Director of Vantiq Co., Ltd.. Thank you very much.
Kawakita Member: Thank you very much, . I'm Kawakita from Vantiq.
Vantiq provides a platform in the category of event-driven architecture. This is not to run applications after data is stored in a database. It is to monitor in advance how stream data and data flow, and to determine in advance what action to take when the data becomes what form. Therefore, it is a way of creating a system called event-driven architecture.
It is an American company, and I will be the representative of our subsidiary in Japan.
These two are the founders and key people, and it should be noted that they are the two who were the vice president and chief architect of the company called Ingress. When the Ingress became open source, a database called Post-Ingress was created. In other words, they are the incarnation of the database. The incarnation of the database has come to the age of IoT, brought the database to the center, pointed out that the system does not work well, and proposed Vantiq.
If I briefly explain why we need an event-driven architecture, including Vantiq, first of all, with the advent of IoT, there is too much data to put away.
In the case of the Internet of Things (IoT), there are various types of data generated by sensors, and a database cannot be created until the design of a schema, which is a form of data storage, has been completed. This is one reason why the design is becoming impossible.
As represented by smart cities, there are too many stakeholders, and your security policies are different, so it is difficult to share the database.
There is another phrase, "smart something." For example, it is not smart just to analyze traffic volume with big data and make a timetable. If a grandpa in a wheelchair appears there, you can turn him green before you finish crossing. It is smart only when you can do this, but the grandpa information is not in the database. If you cannot solve these problems centered on the database, you can solve them by creating an event-driven system.
However, there is a weak point in event-driven architecture, and although EDA has been mentioned in the past, there are no decent applications. The reason is that databases cannot be shared. On the database side, multiple applications can work together in a single database, but the mechanism of EDA is that there is a broker in the middle, and there are sensors, update information of legacy databases, and various events. If you want to use data, you write the type of data you want to use as a topic in the event broker. Then, from the event data that rains and falls, if it matches the topic, it will be pushed to you.
For example, the above subscriber could easily create an application by using temperature as a topic and turning on the air conditioner when the temperature is 30 degrees Celsius or higher. The middle subscriber could create an application by using humidity as a topic and turning on the air conditioner when the temperature is 50% Celsius or higher. However, the middle subscriber could not do this. The temperature is 30 degrees Celsius or higher, the humidity is 50% Celsius or higher, and the air conditioner is not turned on when the temperature is 50% Celsius or higher for 60 seconds. Therefore, they cannot check each other.
So what happened to all the other EDA vendors was that they said they didn't want to use the database, but they brought a database called Microbatch here, and they kept writing temperature data and humidity data, and they tried to check that they had been in the same place for 60 seconds, and then the nightmare would come back.
What Vantiq has created is an event bus, a memory space, in which data generation timing, frequency, data payload, and shape can be left as they are. If you do so, you will have a rule engine that can write rules on the Vantiq event bus. There are various tools there, and you can see them on a skewer. For example, there is a clock in the rule engine. Now, the temperature is 30 degrees Celsius or higher, and the humidity is 50% or higher on the same clock. Of course, it is the same clock, so it is very easy to make the operation of turning on the air conditioner after confirming the same step for 60 seconds. This is the mechanism of Vantiq.
I'm sorry for the busy page. The mechanism of EDA of other companies is here, and it is slow because it goes through I/O once it is put into micro-batch. It is slow because the batch is kept around. In addition, design of schema is required to put in data, so agility is lacking. There are more and more sensors, but how to do it?
In Vantiq's case, again, you can write rules in the event bus. As for how to write these rules, we have a popular low-code development called Activity.
For example, one of the activities is statistics, so in the previous example, if you just turn on the air conditioner once per minute, you can average all the timing data for one minute, even though various timing data is coming in at random, and you can make the meaning of the data on the stream. For example, the machine number of the sensor can be the room of A, and the machine number can be the room of B.
After the completion of the entry, it is time to join. I set the data of the thermometer in room A and the hygrometer in room A to 10. When joining, the data becomes dirty, but I transform them on the stream. I clean up and lose or delete unnecessary data.
In the previous example, a conditional expression "if 30 degrees, 50% or more, continues for 60 seconds" is entered, and screening is performed. All data processing from the time data is entered to the time action is performed is processed on the stream on the event bus. Therefore, there is no need to perform the design of a schema or the design of something that calls it, unlike a database.
The data can be in the form of each frequency data and payload, so if you can stream it, you can see it in a skewer, which is called complex event processing, and it can be connected to various actions.
This is not to say that Amazon is bad, but it is a very famous EDA engine, so I compared them.
The purple Kinessis is the microbatch I mentioned earlier. Various algorithms are connected by lambda, but every time there is a Kinessis, there is a microbatch process, so if the processes are connected one after another, there will be a delay and the design will become very complex.
In the case of Vantiq, one of the orange activities is the previous activity. In the previous example, if you want to include the temperature difference of 3 degrees, for example, in this case, you need to change the database again. In the case of Vantiq, you only need to add one more orange activity and draw one line, and the program will automatically change. In the case of Vantiq, you only need to add one more orange activity and draw one line, and the program will automatically change. design
If you connect all of them, you can use low-code in the backend to create this, and a JAVA program is automatically generated. It runs in one environment. So, there is a huge performance difference between the string pattern and the speed at which a complex program runs in one environment.
This is not just saying that it is fast at the front, but in the United States, Verizon's 5G' Mech' platform for development applications is available. The environment called SingSpace is an OEM of Vantiq. At first, Amazon was doing it, but it was 5G fast and the applications were slow, and it was impossible to do anything about it, so Vantiq was adopted.
Another feature is that the database is not in the middle, so in the previous example, if 30 degrees and 50% continue for 60 seconds in the cloud, turn on the air conditioner. If the air conditioner does not switch on properly, call for maintenance or write a program here.
After that, you can install Vantiq on Edge as well. The thermometer and hygrometer are here, so you can run it here until it's 30 degrees and 50% for 60 seconds. The server is next to the air conditioner, so if it doesn't switch on properly, you can deploy it like a single stroke, like calling maintenance here. This is also a feature of Vantiq that doesn't have a database at the center.
I think there are roughly four ways to connect. For example, in a system like Uber, someone places an order. Location and time are dynamic information. At that moment, which empty car is within 3 kilometers is also dynamic. By writing rules, linking becomes the best field.
The driver of the empty car is registered almost every morning, so this is a legacy database. We will also link static information and dynamic information.
Regarding the information you mentioned earlier, in cooperation with the web service, at the end, you are the right person for the driver, but if you can get on the car, the workflow will proceed. In other words, if you write the workflow first and connect everything you want to connect, Vantiq will be able to do many things without creating an application on top of the database.
In summary, the EDA came out, but the microbatch database had some drawbacks, but Vantiq solved them, and it became a platform that can connect various data sources from IoT to legacy.
Mr. Daikin has also introduced it. On the platform of Mr. Softbank, which is scheduled to be announced later, Mr. Softbank has made various investments, and various devices will be released. It is very difficult to make a design for the database and bring it to the application every time, but with Vantiq, you can make an application if you put all the data in. So, Mr. Daikin is advertising the use of it for two consecutive terms.
It was a rush, but you said that there are already various cases in the world of EDA. Thank you very much.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Kawakita.
There will be one shocking story after another, but please don't hesitate to speak up. Go ahead.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, How long will be left for the Vantiq bus you just mentioned?
Kawakita Member: Thank you very much, That's a wonderful question. In the previous example, if you don't clear the fixed time on the thermometer and hygrometer, you don't need to turn on the air conditioner, so it's unnecessary data. In the previous example, it's 1 minute. As a Vantiq platform, I think 2 minutes from there is basically enough. If you have to wait for more than 2 minutes, you can just go to an external database to ask. So, I think it's unlikely that you have to wait for more than 2 minutes by creating a real-time workflow. Is that the answer?
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister If so, do they dispose of them one after another quickly?
Kawakita Member: Thank you very much, That's right. It's slow because we have a platform to store and a platform to react in real time together. I'm not saying that we don't need a database. We analyzed various things and found the trend value. If we found the trend value, we should take action. If we can get a weather map like this, it will rain. If so, we should put an umbrella, and we can properly instruct the GPS terminal there. These two are related.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, By the way, is the cache that made the judgment, or the history that made the judgment because these three were gathered here in 60 seconds, left in another place?
Kawakita Member: Thank you very much, You're right. It's not so much about keeping a history as it is about preparing activities and writing them out.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, That's right. Thank you very much.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Would you like anything else? Here you are.
Member Inadani: : It is almost the same as the question you just asked. When something like a trouble occurs, I was a little concerned about whether I misunderstood the date or whether I could validation that it did not go well. I should be able to confirm only that. The point is that I should be able to put those things together well.
Kawakita Member: Thank you very much, You're right.
Member Inadani: I see. Thank you.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister privacy information, one after another. Is it very compatible with that?
Kawakita Member: Thank you very much, That's right. Each database is fine. So, you access only what the workflow needs. If you access the database, doesn't it answer something? Each answer can be in a different form. Vantec will shape it. For example, in the conventional way of thinking, external applications that want to be linked with My Number will be discussed from the perspective of how to link with the database, so the budget will increase, but there is no need to make any design.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister I see.
Kawakita Member: Thank you very much, This is the point that everyone can say that it is a scale from the eyes.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , please.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, : We are trying to establish a base for the dependents data connections, and we would like to discover the risk of overlapping incidents. I don't think it is only for children, but on the other hand, we have to consider the personal data, and the location of the database is different for each local government. This is Issue. What kind of work will be done to connect the databases? I am sorry that we are talking about the implementation of the system.
Kawakita Member: Thank you very much, Keep multiple databases as they are. Write a workflow like Vantiq. For example, suppose you write an application that notifies a nearby parent or teacher that it is dangerous for a child to talk to someone they don't know for more than 3 minutes when they leave school. Then, for example, the place where they are meeting someone is real-time information, but who they are is a legacy database.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, In that sense, I would like to hear about a more systematic layer. Is it possible to connect the existing database and Vantiq by setting it up?
Kawakita Member: Thank you very much, Vantiq has a general interface for IoT data, so it has a standard one. In addition, it can be connected to web applications by rest. An SDK is also prepared, so it is okay to build it in.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , if you have any comments, please do so.
Ezaki Member: In addition, the difference of troublesome data is that it is a structure that has a compiler with intelligence that can be absorbed, so it is correct that things that people said they could never do worked.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, I understand very well. Thank you very much. It's wonderful.
Kawakita Member: Thank you very much, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister administration, or for the regulations that the administration will carry out, in other words, our immediate Issue, we must interpret and digest what this mechanism means. I felt that this itself is a big Issue.
Kawakita Member: Thank you very much, My way of thinking will change completely.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister That's right.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, You can handle multiple data and make decisions at a very low cost and without maintaining data.
Kawakita Member: Thank you very much, That's right. In addition, there are IoT data that is updated very frequently, and there are various databases that are updated gently. All of them are connected to the workflow base.
Ezaki Member: In addition, the database to be written out is also made so that it can be selected from multiple databases. In the front-end system, there is basically no recorded data, and it is made so that it can be read or written from an external database.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, , we conducted the survey at the plant in advance. I believe that the continuous monitoring system and the post-check type, which is the specialty of the members of Inadani, can be introduced very easily if such technologies are available. Thank you very much.
Kawakita Member: Thank you very much, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , I have received a number of comments. Is there anything else? If there is at least one more thing, please let me know. Please go ahead.
Member Inadani: : I would like to repeat what you just said, but in light of the relationship with Dr. Ezaki's talk today, the very nature of regulation will fundamentally change. If we grasp the essence of the problem of the matter subject to regulation and decide to deal with such an event in this way, we will be able to do it completely automatically. Moreover, we will be able to reduce privacy infringement and errors at that time, and if errors occur, we will be able to do validation later. In addition, we will be able to do EBPM driven, so if we devise how to write rules and connect IoT well, we will be able to realize a completely different law. I had a sense that this is becoming a reality, but I believe that this is the correct understanding of what you just said.
Kawakita Member: Thank you very much, That's right. It has already begun in various places, including the field of communication, the field of security, and on-site talks at manufacturing companies.
Member Inadani: Thank you very much. It was really interesting.
Kawakita Member: Thank you very much, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , are you all right?
Mr. Kawakita, thank you very much for your very exciting talk.
Kawakita Member: Thank you very much, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Then, I would really like to make a quick start, but next, I would like to have a presentation from Mr. Hideto Funayoshi, General Manager of the Next Generation Network Department, Advanced Technologies development Division, SoftBank Corp. Thank you very much.
Funakichi Member: Nice to meet you. I'm Funayoshi of SoftBank.
Thank you for giving me this opportunity.
Today, I would like to introduce you to the technology of infrastructure inspection, including the history of our company's efforts. I would like you to understand what we have achieved so far, and think about how to utilize it in the future.
I have just projected the materials. Could you take a look at them?
Let's get started. Of course, we have a limited time, so I think we'll have to run a lot. Thank you for your understanding.
Needless to say, these are data from three to four years ago. I think one point in time for the deterioration of infrastructure is 50 years after construction, but in the next 10 to 15 years, the number of things that are out of reach will increase rapidly due to the decrease in the current working population. I think this is the latest Issue.
There are 720,000 bridges in Japan alone, which is quite a surprising number, but there are also various other things called infrastructure, and I think that this is the environment around us. Under these circumstances, we wanted to solve the problem of what to do somehow.
We are a telecommunications carrier, but we are working hard as a digital trading company, so we are wondering if we can solve this problem by communication and digital. For now, we also have a mobile tower, so we are conducting infrastructure inspections every day. This is also a situation where we are in the same situation, so we have been conducting activities since 2019 to improve our technology in use cases.
When I actually reviewed what we are dealing with as a Issue in infrastructure inspections, it is the middle of the left side. This is not about cost. The height of the steel tower is 30 meters, and the height is 50 meters. People attach safety ropes and wires to it and go up. Even if we are safe, there are sometimes unexpected accidents. It is written that we can avoid danger in these places and reduce the number of inspections. People actually go up and check at the visual inspection. We are in a situation where we depend on people not only for going to the site, but also for preparation for going there, travel time, and the total number of inspections after returning home. We are wondering if we can solve this problem digitally.
Third, we have been using AI and other technologies to look ahead, not to look at the past or the present, but to look ahead. We can make good efficiency of the plan for the work itself, so we have been turning around the parts that we cannot handle in such a case, and we have been advancing the project with the aim of improving the economy.
At that time, we didn't have any hands, so we looked for various partners and ventures at that time. However, we are still continuing to do so, but as a development partner, we would like to work with an American venture company. It is 5x5 Technologies, a company in Florida, and we have been thinking about cooperating well with them since 2019.
They aimed to launch the digital twin early for use cases, where infrastructure inspections are being conducted, and to provide customers with not only analysis of the current situation but also future forecasts for the digital twin on the web. At this point, they are doing so, so they were a company with such strengths.
As for the outline of the service, first of all, the infrastructure that needs to be monitored and maintained is photographed. Then, it is modeled on the cyber, analyzed, predicted, and provided on the web.
This is the form in which everything was modeled in actual cyberspace. This is a visual evaluation. With the current ability, I will repeat once again, but even with this much visual inspection, the form is no different than the actual one.
This is an objective evaluation. I have actually taken a close look at our assets. The dimensional error is a deviation of less than 5 millimeters with respect to tens of meters. The blue one is an antenna. It is covered at an angle to the ground surface. The angle is very important, but if it changes, there will be problems. The error is less than 1 degree, so we know that current technology alone can sufficiently sample.
If you can do this, it will be on the left side. Although I could not bring it out this time, when it is completed, the inspector checks it in advance by looking at the dimensions written on the paper and the still images on the left. After it is completed, the inspector checks it on site by looking at the 2D photos and the drawings at the time of completion. On the other hand, in cyberspace, the range is expanded, so it is very efficient and can measure the dimensions. Therefore, not only steel towers but also historical cultural properties may not have detailed design maps, so they may cause various problems in the event of disasters. I think it can be used for digitalization technology.
Up until now, I have tended to copy what is actually in cyberspace. I almost saw what is actually on the web, but I will make better use of AI to predict the future. I am thinking of a digital twin.
To do such a thing, this is rust. It is made of iron, so it rusts quite a bit. Especially the seaside rusts very much. Depending on the degree of rust, there are places where you need to go for inspection early and places where you can take your time. But we don't know right now, so we have to go there and decide on the spot whether repair is necessary. If we collect this data and predict the future, we can go for inspection and maintenance in a planned manner according to the ratio.
In addition, it is a bolt or a crack in a bridge. Invisible things themselves are simulated in AI cyberspace using a sensor called a camera, so that the level of risk can be firmly determined and maintenance can be performed in a planned manner. I think that by assembling data from these places well, it can be applied to various places.
In the remaining time, there were a few orders, so I would like to introduce modeling technology, although it is not AI. What we are doing now is modeling using an ordinary visible-light camera and coordinate information. When it comes to whether it can be done well if we actually do that much, there is surprisingly know-how, and there is only one method of shooting, such as flying drones.
One is how to fly. The way to fly is written here, but by combining the feature points of the asset you want to model with the feature points of the camera you use, the numbers will come out mechanically. If you actually input the flight data as a profile, the number will fly automatically, so it can be done, but there are things that need to be done well in advance.
The performance of the current camera itself is 42 million pixels, but your smartphone has about 10 million pixels. It has about four times the capacity, but this is a commercially available camera, and it has this granularity.
In addition, regarding the method of shooting, if I put the camera that I just mentioned on the market as it is, for example, it will be a monument of a rocket, and it will be transparent, but this area can be photographed close or far away, and by layering images, it can be modeled well.
We already have the know-how, and we have already accumulated such know-how. By applying such know-how well, all of these digital twin models were actually created in cyberspace, and various types of digital twin models have been created, so they can be put into practical use immediately.
Last but not least, since we now have this much technology, it has become a matter of course rather than a new thing. As for the future prediction using the digital twin mentioned earlier, it is not possible to predict tomorrow for the first time, so how to accumulate the past to the present, and by doing this simulation hard, the correct future can be predicted, so by applying it quickly, it will lead to safe and secure maintenance, and I believe that we will go to an economic place by managing the number.
I'm sorry to rush you, but that's all.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , I would like to thank you very much. I also heard a very exciting story. May I ask you a question? You said that it can be put into practical use with existing technology. Looking at the reality, steel towers are of course important, but in the case of electric power, telephone poles and telephone poles are two or three orders of magnitude more than steel towers. All of these things must be inspected, but it does not seem that the method of predicting the future with AI by looking at it with a machine is generalized. What is your recognition, Mr. Funayoshi? If not, what do you think is the problem?
Funakichi Member: , I think there are two main points. The first is that there are lines, such as the telephone poles and electric wires, that make it difficult to fly drones. Depending on the situation, I think it is okay not to actually fly drones where you can reach and where you can see, but if you fly drones, electric wires will be in the way. It is true that there is one environmental factor.
The second is, I think it is common including DX products, but there is no place for users to use it. It is like a traditional craftsman's technique, and people don't transform it because they think it is a straightforward method. In fact, we also offer a similar drones service other than this, which is a bit more legacy, but we take only completely 2D photos. We don't do modeling. The reason is that it is troublesome to see the model.
In fact, 10% of the people who have carefully looked at the modeling say that this one has better performance, but the majority think that the problem is that the voices of the craftsmen are loud and it is difficult to use it.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister I see. It's a deep insight. Thank you very much. Mr. Inadani, please.
Member Inadani: You mentioned earlier that the merit of scale is necessary in relation to economic efficiency. You mentioned that the users have not yet grown up. In that case, is it necessary to stimulate demand in order to successfully obtain the merit of scale and to rapidly expand such initiatives?
Funakichi Member: , I think you are right. If you do not raise awareness about this, it will probably not be used. If it is used, the database mentioned earlier, drones, and the development of people will be necessary. I think initial investment is necessary, but if there is an opportunity like an initiator, if only initial investment is made, people will grow and the number of people will be controlled, which I think is an advantage, so I think economic efficiency will come out. I personally think that it is very important to create an opportunity as a national initiative.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister 's government. Thank you very much. I would like to take a break now. At the end of the first half, if you have any comments from the Senior Vice-Minister, please let me know.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, .
I think you mentioned many new technologies that are suitable for the second half of the war, and I think we need to apply our wisdom to how we can change the system of rules.
If we can do that, as Mr. SoftBank said earlier, we will need to create a market, or rather, create demand, like demonstration experiments. That is my image.
It is very selfish, but I think that Vantiq is good at the place like data connections, dependents, and I think that SoftBank is talking about the possibility of sharing local antenna towers with carrier antennas in Conference on the Realization of the Digital Garden City Nation Vision. I felt that it would be easy for the country to stimulate demand if we could do 3D modeling of the things built jointly by carriers and make them into a database. In any case, we need to think more about how to use them, and I thought that it would be good if everyone who made a proposal today would be interested in the slightly complex system of rules and propose that we shine here. Thank you very much.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister . We will take a five minute break.
(Breaks)
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Then, may I have your attention, please? It is two past five in the afternoon, so we would like to reopen. Thank you very much.
I'm sorry to bother you again, but I would like to have a word from the Senior Vice-Minister here again.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, I see. Thank you.
I think we were able to have a very good talk, and in terms of management, we were able to communicate quite frankly like this, and I think the third meeting was also very meaningful. Attorney Masashima also participated, so I would like to study again and do my best from here, so I would like to ask for your cooperation.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister . Next, I would like to ask Professor Yutaka Matsuo of the University of Tokyo Graduate School to make a presentation. May I speak to Professor Matsuo?
Member of Matsuo: Thank you, Thank you very much for your time. I would like to talk a little.
What I would like to talk about, although it will be short, is that AI technology will advance in the future, and through various trials and errors using AI, there are various things that have been understood, including visual inspection and inspection of various human activities. I would like to tell you about them.
First of all, as you know, AI, especially deep learning, has made great progress in technology. I think it is becoming more and more involved in daily life. Recently, DeepMind released Alpha Code, which is programmed automatically. It seems to be as good as the average programmer in competitive programming, but it says Problem Description on the left side, and it seems to be corrected into the program on the right side.
That's why technology is advancing rapidly, but AI technology is still only a part of it. The reason is that, for example, in automated driving, many people are doing their best, but it has not been put into practical use. To put it bluntly, it calculates the route from the current location to the destination, and it calculates the route by comparing it with detailed map information. In addition, it measures it to the unit of several centimeters using GPS and other devices. In that case, it is basically just a matter of accurately following the route, but it is troublesome if there is an obstacle there, so it recognizes oncoming cars and pedestrians. It is like using deep learning of image recognition here.
In fact, only a small part of AI is using deep learning. The most fundamental problem of AI today is that, like automated driving today, humans build the entire system and use deep learning for only part of it. On the other hand, when we think about what humans do when they learn to drive, it is not that they learn only recognition. They learn recognition, steering wheel and brake operation, and what to check and when to check it. By combining these things, they learn the entire act of driving. The current AI mechanism cannot learn large acts by combining learned acts. Therefore, this is a bottleneck. Tasks that include partial tasks, which AI is not good at, have a distorted structure in which the accuracy of a single task is extremely high.
In terms of facial recognition, translation, and other single tasks that require a large amount of data to be collected, it works very well, but in terms of dialogue systems, automated driving, architecture, agriculture, and other things that involve tasks, it has not worked well yet. However, I believe that there will be technological breakthroughs here someday, and I myself would like to produce them from Japan, but I believe that great things will happen when they are made.
In that sense, the current situation is that, for example, visual inspection inspection and visual inspection are simple tasks, so they are easy to do, so they are progressing very much. So, what we have learned from the automation of skilled techniques in visual inspection is that humans are lazy. Humans are bossy, but in the end, there are very clear OK zones and very clear NG zones. This means that both AI and humans are correct, but the area in between is the boundary area between subtle judgments. In the case of humans, it is easy to be an object of authority or mounting. I have cultivated it through many years of experience, and while I am confident, it is quite wrong. For example, in visual inspection, Shogi, and Go, young people are very active, and it is very interesting to know that what a bossy person said was actually wrong. In short, there is such an area.
If we can improve this area properly, there is actually a lot of potential. The field has changed to that of medical care, but it is now difficult to produce drugs for dementia. The reason is that in the case of cancer, it is known that it is effective if the cells are reduced in size, but in the case of dementia, it is not known whether the drug is effective or not because there is no evaluation axis. However, if AI can properly quantify it, drug discovery will be greatly facilitated, and I believe that there are many industries and technologies that will grow by properly measuring and scoring with AI.
You said that you stopped using AI because of discrimination against women who used AI in Amazon's hiring. In the end, that is also a sign that human beings are irresponsible. I think that it is not a matter of pretending to be someone else and pretending to be me, but it is a matter of pretending to be AI and pretending to be me. However, I think that it is necessary to have a process in which AI reflects it as it is as a mirror, and the human side corrects it by looking at it. In the end, I think that it is necessary to re-discuss and raise the issue of whether or not the law should be realized by AI, while re-examining the sloppy parts of human beings that have not been clarified until now.
That's all.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , thank you very much. Please make your statement. Please.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, is familiar with the Government's efforts and the real world, if I were to ask him, if it is as he said that he would like to make it possible for Japan to solve new problems, do you have any specific fields or tasks that you think would be challenging and high in this field?
Member of Matsuo: Thank you, .
You pointed out a very important point. In terms of current Japanese AI research development strategy, I don't feel like I can win even if I do it properly, and I think I have to think about how to do it. As you said, I think this is a place where there is a considerable chance of reversal.
In terms of where to start, there is a lot of discussion about this in the laboratory, but the easiest thing to do is GAFA itself, but Internet services are fast and scale, so I think it is faster to start up by doing things that include tasks.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, What are the tasks contained in the Internet, for example?
Member of Matsuo: Thank you, It is a common white-collar job. There are various ways to start this, and it may be legal tech, sales, or sales, but I think it is such a system.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Legal Tech is not a big industry, so it is not interesting, but it is an accumulation of sloppy work and is all authority, so in the end, it was sloppy, but I think it will be very obvious. I always tell students that law is not a study but an entertainment. I don't care about that.
Member of Matsuo: Thank you, Just in case, OK and NG are correct.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister That's certainly true.
Member of Matsuo: Thank you, is sloppy.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister That's right. In the middle, we have been eating. I am aware of that. If there is a chance to discuss it later, I will let you do it. Time is pressing, so Mr. Matsuo, thank you very much for your time.
Member of Matsuo: Thank you, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Next, I would like to have a presentation from Mr. Kan Sogabe, President and Representative Director of GRID Corporation. May I speak to Mr. Sogabe, President?
Sogabe Member: Thank you, Best regards
Mr. Matsuo, thank you for your introduction. It's been a long time.
Today, I would like to briefly introduce our efforts. Our company is GRID Co., Ltd., which is engaged in development of AI. Recently, we have been engaged in development of quantum computer algorithms.
Our mentality is to apply AI technology to social infrastructure, so we are only doing projects that focus on infrastructure, and the provision of solutions is also quite specialized in social infrastructure.
In terms of the content of our businesses, we are engaged in such areas as electricity and energy, distribution and supply chains, urban transportation and smart cities. In addition, in order to improve infrastructure planning, we are working to solve a large number of optimization problems using AI technology.
As a technical trend, starting with machine learning, deep learning came out in the early 2010s, reinforcement learning came out, and quantum algorithms came out, so I understand that various algorithms are evolving rapidly.
I understand that the use of AI is widely spread in image recognition and speech recognition, but I believe that AI will expand to areas such as automation and optimization in the future. I believe that AI is working as an extension of discussions on image recognition and technology in 2015 and around that time. As Professor Matsuo presented earlier, I believe that AI technology will be used in the creation of programs automatically, mechanisms for automation, and optimization.
In terms of big data, we are currently collecting a large amount of data across Japan, but there are also places where we are collecting a large amount of data that is quite meaningless. We are working to create things through collaboration, like simulator technology, and are working to create data ourselves.
We have compiled the technical architecture, and we are putting business rules and physical models in a simulator, learning them with algorithms, and putting them in a business system.
Conceptually, the concept is to reproduce such social infrastructure with digital twins and optimize it with AI.
This trend is also worldwide. Last year, Mr. Tesla held AI Day in the summer. Among them, we are using simulators to complement automated driving. We are using simulators in virtual spaces to train automated driving digitally. This is a global trend. We are using simulator technology in digital spaces to complement data and make AI smarter.
We have prepared several use cases. We are doing something like optimizing the operation of power plants, and we are using AI to automate the operation planning of power companies. We are reproducing the entire power system network in a virtual space digitally, and learning how to operate power plants with AI. We are making AI remember which turbines should be installed in which order and how they should be installed economically, and the AI automatically thinks about how to operate the power plant according to the daily changing demand pattern.
When operating a power plant, there are many things that should not be done, but we understand all of them, understand that this timing is not enough, and think about how to operate the power plant. With such a system, we adopt a plan that is made by the power company based on how to operate the power plant, and actually operate the power plant.
Next, in regard to the optimization of AI ship allocation, in a project with a Japanese oil company, we are using digital twins to digitally reproduce all of the locations where oil is produced by the oil company, ships, and places where oil is stored. There are about 100 things that should not be done in various ways, and while observing these things, we are automatically planning the operation of nearly 100 ships in detail by the hour for the next few weeks. Taking into account weather information, we are creating a system to automatically plan the operation of nearly 100 ships while observing the rules such as avoiding Japan when waves are high and the working hours of sailors.
In the context of smart cities, we will recreate the entire city with digital twins, and optimize the whole city. We will recreate the city in virtual space, simulate all the movements of people, and optimize the entire city's movements using digital twins.
In addition, we are now in an age where AI is used to create various compounds through the fusion of simulation technology. We are searching for chemical mixing patterns, creating our own parameter design of semiconductors, creating digital twin simulators for the design of semiconductors, automating the design of semiconductors, and creating AI that improves efficiencies. Algorithms may perform better than the parameters that researchers think.
In conclusion, I would like to conclude by saying that until now, I believe that big data has been the main source of information. However, including the case of Tesla, I believe that there is a movement to complement virtual space simulator technology, including digital twins, while making good use of it. In fact, the project that we introduced is solving planning problems for social infrastructure by combining both left and right technologies.
Finally, from the perspective of mitigating regulation, speech recognition and image recognition, which you mentioned earlier, are actually smarter than humans, as you said, so I think it is necessary to think about how regulation should be in accordance with the times and technologies. In the future, regardless of speech recognition and image recognition, if it evolves a little more, I think it will happen in the next few years that AI will be used for automation and optimization, which I mentioned earlier. So, if you consider how regulation should be in the future, I think you can take the lead, and I think it will be interesting to think about it from such a perspective, so I have made this recommendation.
That's all for my introduction. Thank you.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , President. Now, please make your statement.
Ochiai Member: I'm Ochiai . I thought it was a very important point of view while listening to your story.
I believe that image and speech recognition can basically be solved by writing that they are technologies that substitute parts that are supposed to be done by human hands for various works such as visual inspection, which we discussed this morning. If you have heard that there is regulation in such parts from other perspectives, I would like to ask you one thing.
Another point is that during the actual analysis, for example, when we change things to agile, I recognize that there may be cases where it is difficult to update programs, medical care equipment, and storage batteries. For example, if we want to use this in the real world, I think it is necessary to change regulation based on city planning and various plans so that it is easy to change what is approved in advance. Is this correct understanding?
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister ?
Sogabe Member: Thank you, .
In the context of image and speech recognition, I believe that a large number of stories such as regulation's have been discussed, so I believe that a sufficient number of examples have been provided. However, I would like to point out one thing. I believe that it would be desirable to have such a discussion around 2015 or 2016. Image and speech recognition are becoming a considerable mature technology for us, so I would like to ask you to consider a discussion that looks one step ahead. This is my first answer.
The second point is that when it comes to planning work, most of the various companies are run by people who plan something and run their business. It is common to talk about how many things to make, and how to move logistics. Society as a whole plans something and operates it. However, when we look at it, there is a limit to the plans that people think about, and it is quite wasteful. It is good to move something once, but it is transported two or three times. It is wasteful everywhere, and it is emitted as CO2 emissions.
It is a premise that we should use new technologies to handle such things more smoothly and smartly. However, when we try to achieve an optimal state in various places, some can be controlled, some can be changed immediately, and some cannot. I think we need to think about things in 10-year units, including investments. Some can be changed and some cannot. However, if there are many elements that can be changed relatively flexibly, they can be changed flexibly and it is easier to re-economy. Therefore, if there is a meaningless regulation, we plan on the premise that we must keep it. But there are also things that are not.
In the case of the ship I mentioned earlier, I think that the working hours of sailors are a regulation that should be protected, so in the sense of protecting workers, I think it is better to have such a regulation. On the other hand, if there is a regulation that does not have any meaning at all, we have to protect it on the law, and we may end up doing something wasteful. I think that there are naturally good and bad regulation, and I think that the people of the administration are always aware of this. If such an influence is created, it will become a mentality that all business operators must protect. I feel that it is very good to have a flexible policy.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister . Mr. Nemoto, your hand was up. Here you are.
Nemoto Constituent: Thank you, .
Thank you very much for your introduction. I got the impression that practical applications and utilization are limited to the private sector field, such as power plants and ship operations.
However, as far as I know, it is extremely difficult to obtain administrative or public sector-related data. Do you have any examples in this regard? In addition, it is extremely difficult to obtain data on traffic volumes, but if you have any ideas on how to solve problems in areas other than regulation, I would like to ask you.
Sogabe Member: Thank you, .
In the field of transportation, we are working as a smart cities initiative. It is very difficult to measure traffic volume, so we are working with railway companies. For example, we are working with railway companies as partners at terminals in front of stations. In addition, we are working with Nextco on traffic congestion prediction. Nextco has road data, so I think there is no other way than to cooperate with people who have such data. Therefore, when it becomes open, I think people from completely different ventures can use it to do things on their own. However, since the data folder is limited, if you want to do something, I think it is impossible to do what you want to do in the first place unless you work with such people.
Nemoto Constituent: Thank you, .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , don't you? Thank you very much. Mr. Inadani, I'm sorry, I would appreciate it if you could be brief.
Member Inadani: To be very brief, I think you can solve optimization problems using simulations. Then, I think risk management and such problems can be considered to be fairly typical optimization problems. Now, if we change regulation from pre - regulation to post - regulation, I think there will be many problems among companies about how to do risk management. I wanted to ask if you could think that your technology can be used for such problems.
Sogabe Member: Thank you, .
We often talk about VR and ER in financial engineering. We are working on technology to create a plan by simulating such a scenario and considering how much profit will deteriorate and how much assets will decrease. However, if we create a plan that is resistant to demand fluctuation, market fluctuation, and so on, we do not think that re-economy is the best, because it is risk management and optimized. Now that we have technology to create a plan after controlling risk, it can be used in such a way, and I understand that it will spread more and more in the future.
Member Inadani: .
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister . I feel that the regulation of the nuclear power plant can be replaced in real time, but I don't know in my mind. Thank you very much, President Sogabe.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister Next, I would like to ask Mr. Kaoru Nasuno, President and Representative Director of DeepX Co., Ltd., to make a presentation. May I speak to President Nasuno?
Nasuno Member: Thank you for Thank you very much. I'm Nasuno from DeepX. Nice to meet you.
I would like to introduce myself. My name is Kaoru Nasuno, and I am the CEO of DeepX Co., Ltd.. As a company, we have a mission to innovate production sites around the world by automating all kinds of machines, and I was a member of the New Governance Model Review Committee in Society5.0.
The company was founded in 2016, and we are currently operating with a system of about 40 people. As I mentioned earlier, we have set an ambitious mission to innovate production sites around the world by automating all kinds of machines. We are working to solve Issue at production sites, such as labor shortages, shortages of skilled workers, hard work, and dangerous work, by automating machines with AI and putting them into practical use.
I would like to introduce an example. It is an example of heavy equipment automation that we are promoting. At first, we are promoting heavy equipment automation technology development by utilizing AI technology and cyber-physical fusion technology. It is a demonstration video in which a hydraulic excavator is automatically operated to perform excavation, dump loading, and other such operations.
What is shown in the upper right is the space recognized by AI. An angle sensor or the like is attached to the hydraulic excavator, and the status of the soil around it and the position of the dump are automatically grasped in real time, and it is realized that it moves according to circumstances.
I am writing what I have just introduced here, so I will skip it.
In addition to these hydraulic excavators, in cooperation with manufacturers and others, we are also promoting technology development for the automation of mobile cranes used at construction sites, factories, plants, and other places.
Here, I would like to introduce the technical features. We are making it possible to perform AI learning and validation in cyberspace, so-called simulation. In that case, we are greatly reducing the time required for real data and collection in physical space and physical space.
This is about control. In a general AI learning method, I think it is common to collect data from the field, let the AI learn it, and use it. In such a case, the cost is high compared to the fact that the operator actually operates the machine to collect the data. In the case of a hydraulic excavator, it may take a certain amount of time to operate the machine, and it may be impossible to send it again unless it is dug and returned, or it may be fuel-expensive. In order to overcome such various Issue, various conditions are prepared by simulation, and the AI is learned and realized in the Issue.
Below is a video showing what specific simulation elements are configured. For example, by simulating hydraulic pressure, earth reaction force, earth volume, and various machines, we are able to respond to various conditions at the site.
In addition, we are making efforts to recognize and grasp objects, which is also generally said to be FA, and in such programs, we are making efforts to generate data in a simulation and train the AI to create the actual data required. By doing so, it was originally necessary to go to a factory, install cameras, lights, and other equipment, collect data, and make complicated preparations before automation, but we are aiming to significantly reduce such work.
We are currently conducting technology development for a wide range of machines, such as work robots and hydraulic excavators, and we plan to conduct social development for a wide range of industries. We are conducting technology implementation for a wide range of machines.
I would like to introduce why we are doing this. Although it is somewhat personal, I am working on automation of all kinds of machines because I think it is very interesting and a big opportunity in the big picture. I have listed the elements of (I), (ii), (iii), and (iv) here, and I would like to talk about each of them.
The first is about the Japanese working-age population. It has been announced that the working-age population will decrease to about 30 million from 2020 to 2060. It is estimated that it will be difficult to automate production sites that need human resources as they are. I believe that medical care needs will drive technological development.
I have listed the operations that are performed by humans using machines and manual operations by humans. I believe that these operations will become more difficult, and I believe that automation will be the key to how they are supported.
When we talk about automating such operations, we are talking about moving machines automatically. However, there are hardware manufacturers that are targeting various machines and are able to improve or optimize the hardware necessary for automation, which has been difficult to do so far.
In addition to that, as I mentioned in the introduction of the case study earlier, it is becoming possible for various machines to recognize and control AI from the human eye, grasp the situation, and move according to the situation, and a wide range of potential operations can be automated.
I have shown the past on the left and the present and future on the right, and the parts that have already been automated and the parts that have not yet been automated on the top. Basically, what has been automated until now is limited to the same environment, the same shape, and the same movement. For things such as those that cannot be controlled in the outdoor environment, those whose shape is unknown, those that are soft, and those whose movement is irregular, it is necessary to move them according to the environment, the object, and the content of the work. Such things are difficult to automate.
With the advent of AI, it has become possible to recognize the surrounding environment and to move flexibly depending on the situation. We believe that by combining such technologies, it will be widely automated, and we are making such efforts.
In addition, as the Government of Japan has stated, we call ourselves an advanced country in Issue, and in the future, Japanese Issue will be globally actualized. Japanese production sites have Issue caused by population dynamics, and prediction, optimal technology, and AI technology are growing, and by combining the two, domestic Issue will be solved. At the same time, we will be able to take on a Issue that will be globally actualized in the future. We will be able to solve domestic Issue and take on a big challenge that can be deployed globally. That is why we are working on this.
The above is why we are working on it. In the end, I would like to share the Issue and episodes that we and our partners felt in relation to laws and other matters in the process of machine automation by AI.
Here are five points. The first is for demonstration experiments at construction sites, the second, third, and fourth are for considering the operation of automation, and the last is for technical development.
First of all, when we try to automate operations at construction sites, we are advised to put safety first from the site. If accidents occur and we are suspended by designation, the loss is immeasurable. We are told that we cannot use equipment and operations that do not cause accidents at the site. On the other hand, we are told that it will be difficult to prevent accidents if we try to maintain the current operation of the demonstration experiments.
As for the suspension of designation, it seems that there is a suspension of designation in the construction industry, and if a serious accident occurs, you will not be able to receive guidance from the government for a certain period of time. In that case, each project costs billions of dollars or more, so if we are unable to operate for a certain period of time, there is a possibility that we will suffer an economic loss of that amount. When we consider the balance between the possibility of suffering such a loss and the fact that heavy machinery is operating automatically, we feel that it is difficult for force to work in the promotion of automation. This is one of the reasons.
The second point is about the so-called license. The current license is based on the premise that a person will basically operate the aircraft in the cockpit. However, it is unclear whether it is necessary for a person to be on board when automation is performed, whether a remote cockpit is acceptable, or whether there is no physical cockpit.
In light of such uncertainty, when conducting demonstration research on automation, we conduct operations by sounding out the competent authorities in Ministry of Health, Labor and Welfare about the content of the initiative, including its security and appropriateness. However, I have heard that there are cases in which the decision differs depending on the person in charge, such as area.
In addition, if we consider a crane as a work robot, it will be impossible to move it unless it is in an environment where people do not stop by. However, in most cases, people are involved in the unloading and unloading of cranes at construction sites. In the first place, it is unclear whether the safe regulation applied to industrial robots should be applied and whether it can be automated when applied.
Lastly, with regard to the difference in efforts between Japan and overseas, AI, which is being discussed in Europe, is classified as high-risk AI, but while guidelines for procedures and systems for AI vendors have been shown, policies have not been seen in Japan, and I feel that there are concerns about the setting of technology development targets.
So, that's all for the presentation from DeepX. Thank you very much.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , thank you very much. We are almost running out of time, but at least one of you will be fine. Mr. Ochiai, please.
Ochiai Member: I'm Ochiai . Thank you for your explanation. It was a forward-looking talk and was helpful. I received a discussion on automation, which I think will be in the second half of Principle 2 or after 3, and I thought it was very helpful.
The first is about the technology development that is currently being projected, and I believe you have pointed out that the future shape of regulation is unclear. In addition, if it is a region that receives public funds for medical care and nursing care, it is difficult to make plans unless we can see how investment will be made, that is, how the Government will develop industry. I thought it would be easier to understand if the Government made plans for how to deploy the technology not only in regulation, but also in other regions. Is it okay to proceed with both sides?
Second, regarding automatization, it is also included in the Digital Principles that it is better to clarify the responsibility relationship. Regardless of whether or not the civil liability relationship is finally determined, it may be easier for demonstration experiments and others to proceed even if we just leave aside the matter that we can respond in this way under the business law. Therefore, we considered that it is necessary to consider this as soon as possible.
Finally, with regard to demonstration experiments, I believe that the principles of infallibility are sometimes too strong. I recognize that there is a relationship of safety between areas where such principles can be relaxed and areas where they cannot be relaxed, but I felt that it is necessary to rationalize them as much as possible. I would like to make a comment if there is any place where you can give your opinion.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , President of the Company, what do you think? If you have any opinions, please tell us.
Nasuno Member: Thank you for 2 and 3.
Regarding 1, as you pointed out, we would appreciate it if you could step into both sides.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister .
That's all for today's scheduled presentation. Thank you very much to all the presenters. Well, let's wrap up the proceedings for now, and I would like to have a word from Parliamentary Senior Vice-Minister for Foreign Affairs Kobayashi at the end.
Senior Vice-Minister for Digital Affairs Kobayashi: Thank you very much, .
Thank you very much for the introduction of Mr. Matsuo and the fact that people in the field of AI came out.
So far, I think you have put into practice that we must think about things based on the premise that the accuracy can be considerably improved in the world of simulation. In that sense, even if there is a regulation that you want in advance, I thought that the time would come when we could find a more optimal checkpoint in the world of simulation or make a judgment that it would be okay if you could overcome this. As you mentioned earlier about optimizing ships and tankers, I thought again that the time would soon come when we could use digital twins to simulate and optimize based on various conditions, and how the revision of the law would affect society.
In that case, if we can simulate legal reform, we will be able to conduct more trial experiments, so I think there is a possibility that we can transform into a Japanese society in which legal reform can be realized more flexibly and more speedily, so I would like to follow up on it a little more without giving up.
Last but not least, you mentioned that it is difficult to take steps because the rules are ambiguous. This may be the case with Mr. Inadani. With regard to the suspension of designation mentioned earlier, it may be a suspension of designation after one major accident has occurred. Although it is not a car log, there is a video of the construction site, and if it is disclosed including the causes of the accident, I wondered if it would be possible to consider the ideal way of execution by regulation to avoid a sudden suspension of designation.
In any case, in the first half of today, there was a discussion about replacing the clearly analog regulation with digital and technology, and in the second half, as we enter an age in which we can simulate in the cyber world, I think it was an opportunity to think very much about how to think about the ideal way of rules, regulation, and execution.
I think the Secretariat has organized its thoughts. Based on today's discussion, we would like to start discussions on coordination with each ministry and agency, so I would like to ask for your continued support. In addition, I would like to give my closing address by telling everyone in the Secretariat to do their best. Thank you very much for your time today.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister , I would like to thank you very much for your words many times. Thank you very much to all the presenters. I have learned too much, and this is my frank impression. Finally, I would like to ask the Secretariat to explain about the next Working Group.
Secretariat (Suga): The details of the next Working Group meeting have not yet been decided, and the Secretariat will contact you later. Thank you for your cooperation.
As I mentioned in the second session, I would like to proceed with the handling of today's proceedings and materials by making them public unless there is something that is particularly difficult to make public.
Thank you very much for joining us today.
Mr. Annen (Vice-Chair): I would like to express my sincere gratitude to all the members, including Senior Vice-Minister . Thank you very much.