This page has been translated using TexTra by NICT. Please note that the translation may not be completely accurate.If you find any mistranslations, we appreciate your feedback on the "Request form for improving the automatic translation ".

Working Group of the Digital Ad Hoc Administrative Study Group Technology-based Council for Promotion of Regulatory Reform (2nd)

Overview

  • Date and Time: Wednesday, November 16, 2022, from 10:00 a.m. to 12:00 p.m.
  • Location: Online
  • Agenda:
    1. Opening
    2. Agenda
      1. Technology-based Council for Promotion of Regulatory Reform Direction and issues for future discussion (explained by the Secretariat)
      2. Explanation from Ezaki Staff
        • Proposal for a technology map "How to proceed with risk assessment of security, etc. in the use of technology"
      3. Explanation from Mr. Shimada
        • Toshiba Corporation's Approach to AI Quality Assurance
      4. Exchange of views
    3. Adjournment

Material

Minutes, etc.

Date and time of the

Wednesday, November 16, 2022, from 10:00 a.m. to 12:00 p.m.

US>

Held online

Members present

Chairman

Hiroshi Esaki (Digital Agency Senior Expert (Architecture))

Member

  • Yusaku Okada (Professor, Department of Industrial Engineering and Management, Faculty of Science and Technology, Keio University)
  • Keiko Ogawa (Banking Capital Market Leader, Regtech Leader Partner Certified Public Accountant, EY Strategy & Consulting Co., Ltd.)
  • KATO Shinpei (Associate Professor, Graduate School of Information Science and Technology, The University
  • KAWAHARA Yoshihiro (Professor, Graduate School of Engineering, The University of Tokyo)
  • Yumi Kawabata
  • Yutaka Saito (General Manager of Information-Technology Promotion Agency, Japan Digital Architecture Design Center)
  • Taro Shimada (Representative Executive Officer, President and CEO, Toshiba Corporation)
  • Takao Someya (Professor, Graduate School of Engineering, University of Tokyo)
  • Keisuke Toyoda (Project Professor, Institute of Industrial Science)
  • Takao Nakagaki (Professor, Faculty of Creative Science and Engineering, Faculty of Science and Engineering, Waseda University)
  • Osamu Nakamura (Professor, Faculty of Environment and Information Studies)
  • Ayumu Nagai (President and Representative Executive Officer of Astamuse Co., Ltd.
  • Masanori NEMOTO (Japan Business Federation Consultant)
  • Yu Noboru (Director, Information-Technology Promotion Agency, Japan Cyber Technology Laboratory)
  • Yutaka Matsuo (Professor, Graduate School of Engineering, The University of Tokyo)

Overview

The Secretariat (Suga): The next Committee meeting in Good morning. As time is up, I would like to open the second session of "Technology-based Council for Promotion of Regulatory Reform".

Once again, members are invited to participate online.

Mr. Endo, Mr. Ogino, Mr. Kyuma and Mr. Suzuki are not attending this time due to business reasons. Mr. Kawahara and Mr. Someya are scheduled to leave in the middle of the meeting.

In today's proceedings, we will give an explanation in the first half and exchange opinions in the second half. During that time, we would like to receive opinions and questions from you through chat. We would like you to use the Slack of the committee as much as possible as a chat medium. If it is difficult, we would like you to post it on the WebEx chat column. We would like the secretariat to check the posted comments and pick them up in a timely manner.

I would like to begin today's agenda.

I would like to ask Chairman Ezaki to lead the proceedings from now on. Thank you very much.

Chairman EZAKI: Are there any other members who did not speak, including Thank you very much.
Now, let's get down to business.

The second meeting will be held on three topics: (1) the direction of future discussion and issues in technology-based Council for Promotion of Regulatory Reform, (2) how to proceed with the assessment of risks such as security in the use of technology for the development of technology maps, and (3) AI quality assurance as considered by TOSHIBA.

At the end of the meeting, we will have time to hear opinions from the members. However, we would appreciate if you could join the chat. Professor Someya and Professor Kawahara will come first, so if you have any comments, please feel free to do so.

First, Mr. Director Suga of the Secretariat would like to explain the direction and points for future discussions on technology-based Council for Promotion of Regulatory Reform.

The Secretariat (Suga): The next Committee meeting in I have turned over one page of the material, and although this is a review, I have explained the outline of the initiatives of the digital ad hoc committee. I have summarized the goals by looking back on page 3 as well. There is no particular change.

The next page is a table of contents, and I would like to explain in line with the three pillars.

First, we conducted an advance public call, and based on the lessons learned from it, we have compiled suggestions from the secretariat on what kind of information should be included in the technology map and under what kind of thinking.

Second, based on the lessons learned from the advance public call, I have prepared a draft of what areas and subjects should be included in the map and what process should be followed to compile the map.

As for the third remaining issue, although it is only partial, I would like to explain the editorial policy, how to design incentives that can be used, and how to secure trust in the content of publications.

First, the first pillar. The next pages are about the ideal way of listing information, and the next page summarizes the overview of public call, which was implemented in advance. Thankfully, although it was a short period of time, we received 24 applications from various people, including large companies, small and medium-sized enterprises, and venture companies. Currently, 20 applications have been published on the Digital TV Agency's website after the secretariat closely examined them according to several criteria.

On the next page, as items to be included in the catalog, we held in-depth conversations with the regulatory authorities in advance, and after hearing concerns, we included them as mandatory response items. The items highlighted in yellow are the summaries of what the authorities wanted us to hear, and all the items up to the next page are the items to be included in the catalog. Mainly, we are concerned about whether they can take measures against unfair entrance examinations, as they are national qualifications.

The next page summarizes the lessons.

First, as a matter of course, this catalog must be coordinated with the regulatory authorities, and through the dialogue with the regulatory ministries and agencies in the face-to-face training, we have learned that they think it is essential to detect fraud such as spoofing and cheating, and we have asked them to respond to it.

When a part of the essential function is not equipped, even if the other function is well-equipped, it will look inferior. This has been pointed out by the business operators, but at this point in time, based on the needs of the procurement side, I think it is unavoidable.

We took the trouble to show the public call results we received to the people from the ministries and agencies in charge of the regulations and the organizations that conduct the tests and training, and asked them if the results could be used. We received a lot of very positive comments. We also received opinions such as, as points to keep in mind, making the outline of the technology easy to understand at a glance, not writing too much about it in the free text, clarifying the evaluation criteria and technical level, designing a layout that makes it easy to compare and search because it is published in a simple format and search is not possible, and making efforts to some extent so that business operators do not have to access each individual because they are concerned about the introduction cost.

In any event, we reaffirmed that it is necessary to cooperate with the Japanese authorities before the implementation of public call.

As the second major issue, regarding the public call this time, the secretariat considered the course and the test as one process, and made a public call for products that could cover the entire process from the application for the course to the payment of the fee, implementation of the course, implementation of the test, and issuance of the certificate of completion. We received comments from many companies that they wanted to apply but declined because it was impossible to cover all the cases in full specifications.

On the other hand, there are some elemental technologies, ranging from large ones to small ones. In an extreme case, a company with only a personal identification technology or a company with a technology that can detect human movements with a camera applied for the public call by filling in the required fields even though they did not meet the initial airport reservation requirements set by the secretariat. At present, these elemental technologies are posted on the website separately because they were applied for, but I thought again that it is necessary to decide how to deal with them in advance.

There are various possibilities, such as making use of ingenuity, including start-up companies, discovering new technologies that we are not aware of, and collaboration between companies on elemental technologies. Considering that the unit of technology search is a convenient decision made by the secretariat at its discretion, I think at present that we should allow applications for only some elemental technologies. I would like to hear your opinion.

Third, if elemental technologies are widely accepted for the second point, another problem will arise. In the case of personal identification technology, I think we will have to have people apply for elemental technologies over and over again for every catalog development public call in the future. For such common technologies that are assumed to be used across multiple technology fields, rather than accepting them over and over again as elemental technologies for individual fields, we may need to extract them separately and develop catalogs. I think this is also an issue.

I have also written three other points.

Going forward, through dialogue with the ministries and agencies responsible for regulations, I would like them to proceed to procurement if they think it would be good to adopt the technology they applied for as it is. Before that, if there are regulations that require face-to-face meetings, they will be reviewed. On the other hand, if it is determined that this application information alone is insufficient and verification is required, we will conduct a technical verification, but I believe that the necessity of a technical verification will ultimately be determined through dialogue with the ministries and agencies responsible for regulations.

The second is, as I said, there were some cases that the secretariat did not publish after careful examination, but one of them is the pre-release service. It will be an advertisement, so it is very grateful that there is a need to apply in terms of marketing, but in fact, the secretariat does not have the ability to keep track of whether it will be released for sure, and it is difficult to conduct fact checking and verification at this point, so this time, it was automatically excluded from the subject of publication.

In fact, there were several applications after the public call period ended, and while I think it is not very fair, we have made an additional announcement for the sake of convenience this time.

On the next page, I will summarize the advantages and disadvantages of listing the main listed information, including what I mentioned above.

First, I explained earlier whether the requirements I mentioned earlier would be made mandatory or not. I also mentioned that elemental technologies would be listed. However, cross-cutting technologies could be listed in a separate catalog.

What follows from here is new, but many people have pointed out that best practices are actually useful. By publishing best practices that correspond to adoption examples, in addition to the technology itself, the context in which it is applied and operated, and examples can be shared firmly. We want to use it in a similar way, so it is very useful that people will decide that it is likely to work.

On the other hand, as a disadvantage, it has been pointed out that start-ups without a track record of adoption may be excluded, and that there may be cases where market opening through regulatory review should come first because there are regulations and there is no market in the first place to accumulate track records. Although we are not yet aware of any specific examples of difficulties, I believe the current hypothesis is that best practices will be included in the technology map, but that it will not be an essential requirement that they cannot be included without track records.

I think it is natural to point out that I would like the disadvantages and restrictions to be listed in the next column, and that it is more useful from the user's point of view to provide information on things that cannot actually be done even if all that can be done with technology is explained in a splendid manner.

On the other hand, as a disadvantage, there was a comment that even if we communicate with the companies that applied this time, to be honest, the companies that own the technology can provide information on the restrictions that should be noted, but it is difficult to write about the disadvantages or what cannot be done because the companies that own the technology have the right to decide whether to apply in the first place, and they would not apply if it would be a negative marketing. Therefore, regarding the necessity of listing, I think it would be good if there was a mechanism to have companies express the points to note, and on the other hand, to have users, including the ministries and agencies in charge of regulations, supplement the information by using it.

Then, as I mentioned earlier, it's the price issue. Those who want to adopt it say that price information is essential for making an adoption decision in the field, and if it's too expensive, it's impossible.

On the other hand, when I had a conversation with companies, they rightly pointed out that the selection focusing only on price is too much ahead of others around the government, and they would like us to evaluate the performance and order the performance. In addition, they also said that it is natural to flexibly set prices after understanding the requirements and scale of users, and it is quite difficult to set a uniform price. Therefore, while we will also leave it optional for the time being, as pointed out during the JFTC meeting, we may consider cooperating with the digital marketplace when it is established.

In addition, as for other matters, I will not forget to point out that information on the relevant guidelines should be included, such as what kind of cybersecurity certification has been obtained, what kind of standards are being followed, and when the information was available.

Regarding the next two pages, I would like to ask whether it would be good if we could continue a little in the future, rather than just making a public call once. This is my current idea, and I would like to receive your frank opinions on this as well.

The first is that there are many regulations that require inspections, inspections, and diagnostics to be conducted on site or on a regular basis. This time, we have made it a target of the reform of the seven preceding items, and there is a need to know the technology necessary for it.

The objects of inspections, checks, and diagnoses are extremely broad, and there are needs to diagnose the risks of accidents, thefts, and fires within the land, sea, and air domains and areas, as well as to monitor the conditions of specific structures and natural objects, diagnose and predict the risks of degradation and disasters, estimate the remaining life, and determine the timing of repairs. Regarding the second point, in particular, the existing catalog created by the Ministry of Land, Infrastructure, Transport and Tourism is very well made, and it will be important to coordinate them. If we suddenly search in such a large area, for example, if we receive 2,000 or 3,000 applications, it will be difficult to handle, so we would like to first consider the scope of the technology search, including such areas, which is the point of contention.

As for the target technology and search technology, I am writing sensor monitoring technology as a typical example, and I think image analysis will also be included. Since there are various sensor technologies, I am writing the catalog items as images below, but it will be a very complicated selection, so I would appreciate it if you could give me advice on how to do it.

The second point, which is a little different, is that there is actually a privacy protection technology called secure computing from privacy tech companies. By using this technology, personal information can be transferred and processed securely without being touched by the person to whom it is directly transferred. Although such a technology has been released, according to the current interpretation of the Personal Information Protection Act, no matter how confidential it is, it is still personal information. However, as an operational device, it is interpreted that if it is highly encrypted, it is not a problem. However, it is unclear whether the use of secure computing corresponds to the fact that it is highly encrypted, so there seems to be a problem that the implementation of the technology is not progressing. I introduced this as a possible target.

In addition to this, we have received consultations on several interesting themes, so I would like to continue it.

The second pillar is from the next page. This is the state shown in the diagram. Vertically, there are cases where there are regulations and cases where there are not. Horizontally, there are cases where there is technology and cases where there is not. Roughly speaking, which areas should be covered by the technology map or catalogue? I would like to sort this out a little.

The first point is on the vertical axis. When there are no restrictions, it is the blank column below, but in light of the purpose of the digital ad hoc committee's initiatives, it will be excluded from the target for the time being. When there are analog restrictions, there are cases where it is allowed to use the technology and cases where it is not allowed to use the technology. I believe that if it is not allowed, the regulations will be reviewed as much as possible and the restrictions on the adoption of technology will be lifted.

As for the horizontal axis, of course, I would like to apply it to cases where there is technology, but there are many zones where there is technology and there is no technology, such as when there is technology and there is no technology, it is ambiguous, and it is not certain unless the capability is confirmed, or when the capability is not achieved. I think it is important to set the scope of technology exploration by taking a Hirome for those that are said to have no technology as much as possible and not to overlook changes in the middle area, and the entire area is indicated by a dotted line with a white outline.

Within that range, technologies that are certainly mature and have no regulatory restrictions on the use of technologies may be listed in the brown area of the Technology Map. If all of them are listed, it may go bankrupt, but if there is a need, they can be listed on the map even if there are no regulatory restrictions. If there are restrictions, they will be listed as unrestricted if the regulations are revised.

In terms of technology, I think it is difficult to make a completely objective judgment between areas where there is a certain mature technology and areas where performance needs to be confirmed or where it has not been achieved, but we will make a judgment in some way, and for those that are judged to require technical verification, we will first review the regulations after that and implement the process of listing them on the map. In addition, if there are regulations and we would like to innovate compliance using technology, but it seems that there is no technology yet, we will write development goals on the far right, and we organized them conceptually so that the authorities can set such things.

On the next page, based on this, I have come up with a four step process for preparing a technology map.

The first is, as we are currently doing in the digital ad hoc consultations, we will identify analog regulations and categorize them based on the purpose of the regulations being implemented across ministries and agencies. What are the authorities concerned about a certain category, what are the essential requirements for complying with the regulations, and what is the acceptable range of risk? This will be a very difficult conversation, but I think we have no choice but to clearly state them to the extent possible. The clearly stated contents will correspond to the items to be placed in the catalog that I mentioned earlier, and such requirements will be reflected in the public call and the items to be placed in the catalog. This process will be necessary first.

Next, we will enter the technology public call phase by searching for technologies that will contribute to the review of the categorized regulations. While looking at the technologies that were identified at that time, we must have a dialogue with the authorities again and decide whether or not technical verification is necessary. At this point, I think that whether the authorities are concerned or not, and whether the authorities think that they cannot say that it is good without verification are the criteria in a single sense. If there is any other wisdom, I would like to know it.

On the other hand, for mature technologies that seem to require no verification, they will go through a minimum of fact-checking before they are suddenly posted on maps and catalogs.

The items that the authorities said need to be verified will be subjected to the third process of technical verification, and those that were found to have insufficient performance or no technology will be set as development targets.

Once the technical verification is completed, the final process will be to review the regulations. I believe that the purpose of this initiative is not to review the regulations and end with them, but to connect to the point where the technologies listed in the map and catalog are procured by the public and private sectors.

On the next page, in order to classify regulations, when the purpose of various regulations is broadly summarized as an abstract concept, what kind of technology is there, what is currently being widely used, what is under technical verification, and what is possible to be used but has not yet been unlocked, as far as the secretariat can grasp, this is a provisional table. This actually has a lot of information and I think it will be interesting if you read it carefully, but unless we do this work first, I think we will not be able to complete the regulation classification and technology search phase.

Next page. Since cooperation with the ministries and agencies responsible for regulations is a key point, we are currently making a budget request to establish a mechanism in which the Digi Agency will cover the cost of technology verification for cross-ministerial verification in the development of the technology map.

On the next page, the ministries and agencies in charge of regulations have pointed out that we should cooperate, and I believe that it will be necessary to clarify the essential requirements and draw up a process diagram.

The next page is the third pillar. The editorial policy is an itemized summary of the discussions held to date by Chairman Ezaki.

First of all, I think it was about how to think about the main body of editing, publishing, maintenance, and operation of the catalog. Next, it is about the responsible body and the scope of responsibility when using the technology described in the catalog. I think this may be the most important. At present, the catalog is provided as just reference information, so the editing body of this catalog is not responsible at all. I think we will consider including whether it is really good.

As for the target of catalog provision, the decision maker of the ministry or agency in charge of regulations is the first target, but we would like to go on to the actual procurement, so we need to increase articles that appeal to those who make procurement decisions in government organizations or organizations of companies subject to regulations, as well as those who have substantial technical decision-makers, which you pointed out last time. That is my understanding of the issue.

As for the function to be included in the Technology Map, I believe it would be desirable to have a function that matches needs and seeds, a function written with technical evidence that shows how technology will change the future, and a region where catalogs are automatically created on maps according to the axes and range required by users.

In addition, I think that the text editing method should be a technology-neutral format that is easy for automatic processing and secondary use.

It will be continued on the next page, but it is an incentive design for continuous update operation. Of course, the Digi Agency will be established initially under the leadership of the government, but I believe you pointed out that it can develop as an ecosystem under the leadership of the private sector, rather than standing as a gatekeeper. In that case, what kind of organization and capabilities are required for maintenance and operation is a matter for consideration.

In addition, we will thoroughly consider a system that makes it easy for technology holders to update and a system that motivates them to write, which was pointed out by Mr. Noboru.

In addition, we would like people to not only create the map but also use it. In terms of incentives for use, it is necessary for the ministries and agencies in charge of regulations to establish a system in which regulations are reviewed with reference to the Technology Map. This is currently under consideration.

In addition, for technology holders, we will consider cooperation with the DMP that I mentioned earlier so that technologies directly procured by the government will be smoothly connected to government procurement as much as possible. In addition, as incentives for business operators, I believe that the value of this catalog will increase by firmly setting mandatory items for regulatory compliance, properly cooperating with the ministries and agencies with regulatory jurisdiction when deciding whether or not technical verification is necessary, and firmly drawing out their thoughts. That is what I have written at this point.

It took a long time, but I wrote on the last page that this is how it is now because of the trust collateral system.

The first point is that in the preceding public call, the technology catalog is only reference information, and the government does not certify or approve it, but in addition to that, I think we are making efforts to give meaning to it by cooperating with the ministries and agencies with regulatory jurisdiction. Furthermore, I wrote that it may be possible to make it possible to appeal the status of various certifications, including security certification, and determine whether it will be reflected in procurement requirements in the future.

If there is any other good system, I would like to hear your opinion.
That's all from the office.

Chairman EZAKI: Are there any other members who did not speak, including Thank you very much.

Although there are two more questions, I would like to ask you first, or if you have any questions or opinions on the explanation from the Secretariat, as I believe many of the questions were quite well organized as I mentioned earlier, as there are some members who are attending under very difficult circumstances.

, please.

Member SHIMADA: I'm glad that the materials have been compiled in such a comprehensive way.

My only question is about incentives. What is actually happening? For example, from the perspective of gas companies and others, after all, even if they introduce IoT and reduce costs, their profits are suppressed and it is not profitable for them. If they do such a thing, there is a need for companies to take risks and make efforts in the first place. There are such issues, so I think it would be good to consider including the design of such a system.

That's all.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

That's exactly right. People won't invest if they don't make a profit, so I hope we can share our practices on incentive design in that area.

Nakamura-san, please.

Nakamura: Mr. I'm glad that so many things have come together.

There is one point. Although it is about feedback on the need to change regulations and laws, I feel that it is the other way around. There is no breakdown of the purpose of regulations, what kind of regulations are there for what purpose, and what kind of quality is needed to achieve that purpose. Instead, methods such as simply conducting tests three times a year or once every five years, and even writing tests or how many hours of training are required, are probably written in the law. I feel that this is the essence of the need for reform.

At the time of making the map, I think you are talking with the ministries and agencies that regulate the area and discussing what the purpose of the map is. I think you can do the same for all regulations. In other words, you need to clarify what you have to do and for what purpose. I think the method has been developed first.

How many times should the test be conducted? Should this level of test be done properly? For example, should these things be inspected once a year with sensors? Visual inspection? Each purpose should be broken down a little more. What is being inspected and what is being done? Breaking down the purpose is very important in mapping. The inspection is done to satisfy the purpose. Then, what methods are available? Unless this is the story, I think that the Digital TV Agency will not be able to proceed. That is probably the part where the flag will be waved.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.
Mr. Suga also nodded.

The Secretariat (Suga): The next Committee meeting in Assuming that the regulations we have had so far were reasonable, now that the wave of the digital era has come, we have had a dialogue with each regulatory authority to review whether the regulations are really reasonable even if they are premised on digital. I think that this is a process that is incredibly costly, but at the same time, I also think that this is the heart of it. For example, speed limits are set for the purpose of safe driving without contact with people, but I think that it is to reach a conversation that speed limits were the best means even assuming autonomous driving technology, so I would like to keep it in mind.

Nakamura: Mr. Perhaps creating a catalog is one method, and in the communication of creating a catalog with that method, I think it would be good if we could support the work of so-called breaking it down by having the regulatory ministries and agencies rethink such a part.

The Secretariat (Suga): The next Committee meeting in Thank you.

Chairman EZAKI: Are there any other members who did not speak, including Perhaps it would be good if examples of such a review were published as practice. I will guide everyone to do it using this as a reference.

Are there any other opinions?

I will make some time at the end, so if you have any opinions, I would like you to organize your thoughts and comments at the end.

Next, I would like to briefly explain how to proceed with risk assessment, including security, in the use of technology for the development of technology maps in line with Attachment 2.

What is important is current practice, and it will continue to change, as Mr. Suga mentioned earlier. We must not forget to consider how the private sector and the government can support each other, and this is not an issue that is limited to the government.

Also, if it is not introduced and operated, it is meaningless, so I think that how it was introduced and operated is not only an issue of the technology map, the so-called technology, but also important. I think it is related to Professor Nakamura's talk earlier about sharing this practice, but rather than having such a method, how to have a method of introducing operation that is to be reviewed, and migration will be a very important practice to be shared.

As for incentives, as President Shimada said, not only is it something that must be done, but in cases where it does not lead to cost increases or cost reductions in digitalization, it will probably be necessary to share practices on how to respond, and if possible, incorporate them into policies. Then, how to share knowledge, expertise, and experience across industries should also be included in this map.

The third point, as Mr. Suga mentioned, is that we need to think about how to avoid excessive demands, regulations, and shackles.

In the next page, we are talking about a matter of course. Trying to avoid such a thing is an essential technology, so we have to proceed. In that case, we need to properly anticipate essential requirements such as cyber security, which Mr. Suga mentioned.

In doing so, I believe that not only knowledge but also operating experience will be very important practices to be shared in terms of what kind of current practices will be shared, as I wrote earlier, including rephrasing. As knowledge, it will of course be easier to write if you study it, but what should be done in the area of operation? We will of course hear about the introduction and operation of the method, and in the earlier section on incentives, I believe the key will be how to design incentives, including money.

There is also the issue of verification. It is very costly for the government to do this, so it is probably necessary to provide information on how to industrialize this, and I think it would be very effective to introduce successful examples. Then, I think that the government will support and the private sector will take the lead in creating such a verification system in an agile manner.

Also, although the ASIS is basically easy to understand as a map, I believe that we must also consider using this map to make it To Be, that is, to change to a new function or a new structure.

In that case, the best way to create it is to go from KPI to KGI. In today's document from the secretariat, it is decided only by price. This needs to be changed. This is what is written as a problem awareness. In that case, I think it is necessary to make KGI a requirement for procurement as a practice.

In doing so, the government was able to successfully drive the idea that these four things can be done in one, and this is the Cloud by Default.

As I have participated in the discussion on Cloud by Default, it is not only to simply promote data sharing. The second point is to leave it to experts to reduce labor costs and costs related to the development and securing of human resources. In addition, CAPEX and OPEX will decrease, and the fourth and fifth points will include CSR and BCP. If we are not satisfied with this, there will be no effect as Cloud by Default, so I think we are moving in the direction of making this a requirement. I think it is necessary to share KGI, not KPI as a practice like this.

Also, if we think about digitalization taking this means of manual work, which is the first person, and bringing it to the third person, which is digital, basically, with digitalization and going online, conjecture and falsification of data will not be possible. Auditing work will also become very easy. Whether this will really happen or not, as Mr. Shimada said, if there are no real cost advantages, the private sector will have no incentive to enter, so what do you think about this?

In addition, considering that the locus of responsibility shifts from people to computers, I think it would be good to include the KPI as one of the benefits of digitalization, that is, psychological stability will be shown to employees when Web3 is introduced further.

In the last page, as one of the important requirements of the technology map, I think that information on security requirements should be provided and understood in one-stop. In terms of systems, it is important to evaluate not only so-called cyber security but also operational stability design, so that something equivalent to what has been done by people will not come out.

Another important point is operation. In addition to simple element technology and system requirements, it is necessary to share methods on how to operate well and incentives for doing it well.

As for the technology verification system, I think it is also an important task to think about how to industrialize the verification business in the private sector, and to present examples of this properly so that industrialization can proceed.

Also, in relation to the issue of price alone, while shifting from KPI to KGI, I thought that the technology map should cover not only new methods and incentives, such as Minister in charge of Administrative Reform Okada or carbon neutrality, which are currently being considered as one of the axes, but also how to increase incentives and how to bring them beyond price, which was my proposal.

That's all. Thank you very much.

Next, Mr. Shimada, will you explain TOSHIBA's concept of AI quality management?

Member SHIMADA: Then, I would like to say that before I say about TOSHIBA's idea of AI quality assurance, I also want to say that we need to carefully consider the operation that you have just pointed out. I am trying to say what is the process to make it possible to operate properly. I think you will understand if you look at the content.

After all, the automation of inspection and the like means quality assurance. We need to be able to confirm whether the required functions are met or not. However, it is not easy to assure the quality of AI. Rather than regarding AI as a robot, I think it is better to treat it in the same way as a human being. In other words, it is useful, but unless it is properly trained, it cannot produce stable quality assurance. That is the problem.

To be specific, as everyone here is well aware, it is not possible to achieve 100% prediction accuracy. It is extremely difficult. In addition, the reason for such a result is also unclear. In addition, it is difficult to guarantee the operation after the start of operation. As will be explained later, it is often said that it will decay over time, but there is a problem that the accuracy decreases or the results are unpredictable if it is used in an unexpected way.

That's where AI quality assurance comes in. It's a good tool, but you have to make it transparent like a black box. You also have to think about the qualities of the training data and fairness to remove bias. You also have to design it with attention to the fact that it must be protected by misidentification, fail-safe, etc. That is very important for the qualities of the training data and the AI model, and you have to look at the transparency, fairness, accuracy, and robustness of the training data itself. Even so, the answer may not be correct. It is very important to ensure the quality of the system. I used to design aircraft, but the people who design aircraft do not trust computers. Therefore, if you want to make it fly-by-wire, you have to quadruple it. We have to think about the same thing that happens in AI.

For example, when we think about automatic driving of cars, we are currently trying to address this with AI using various methods. For example, if the AI produces an abnormal output, a human can override it, or it can be corrected by automatic control such as this.

Various measures are taken in AI. For example, when AI is also duplicated, as an aircraft engineer, I am not sure if it is all right to duplicate it, because I don't know which is wrong. I think there must be more than three. In addition, we will solve this problem by combining it with a transparent AI, which is a rule-based AI that always gives the same answer. In addition, we are trying to enhance the robustness of the AI system for automatic driving with high image recognition by making it a monitoring system that protects abnormal output when it occurs, so that it is very stable and the driver can output manually in the end. AI

Using this, for example, in the maintenance of plants, for example, in the area of abnormality prediction and detection AI, first of all, in the case of such plants, indents are often specialized items, so they are specialized AI. However, there are many cases where AI performance has not been improved. Therefore, while judging the necessity of maintenance, etc., maintenance personnel feed back AI data, and by adding more such data that must be input, AI and humans will gradually work together to improve function. In fact, we are providing such things.

I would like to briefly mention this because this is another example. At the Toshiba Carrier as well, using the system I have just described, we are actually carrying out early confirmation of refrigerant leakage. Kurita Water Industries' water treatment and sedimentation tank automatic monitoring services are also adopting this method.

The quality assurance of manuals is determined by the capability of the AI system to meet customer expectations. Learning data, AI models, control systems, and operations are important elements in the system capability of AI. The quality assurance of AI is to maintain the capability of the AI system to meet customer expectations. In other words, it is necessary to guarantee the quality of learning data, AI data, control systems, and operations as a whole. Even so, the quality of AI is not sufficient, and the quality assurance of the system must be performed in combination with other elements.

That's why we have established quality-assurance guidelines. First of all, we have established quality-assurance guidelines and a quality-assurance process for implementing them. Based on these guidelines, we will guarantee the AI of the model. Furthermore, when our clients invest in the model, we will provide them with something like a quality-assurance card so that they don't have any problems. By doing so, we will be able to utilize the AI as a whole. This is what I am proposing.

The guidelines are to gain basic understanding of what everyone needs to pay attention to as a whole. The list of perspectives is a basic understanding of what there are and what we need to pay attention to. As I said earlier, a AI is like a human being, so unfortunately, you should think that there is almost no such thing that will work as it is once it is introduced. There may be some, but when used for such surveillance, humans tend to complain about why we do not move when we are robots, but it is natural that people who leave school cannot work properly, and we must consider including the educational system.

We have created a process for that. This is a general process, so if we follow that process, we will be able to bring it to a stable operation. This is just one example, but there are things like what are the business issues, whether the obtained data is safe, whether uncertainties are properly considered, whether relearning is done, and so on. When actually operating it, we will operate it while always thinking about the quality of the data, and that the model quality card must be in front of the customer and must be seen properly. This is an example, so I would like you to see it later.

Using this, we will introduce and operate a system that analyzes AI from drone images during power line inspections, and image diagnoses of products in the manufacturing process. You can check the details later.

Finally, MLOps is a continuous update of machine learning. Basically, AI is often said to rot, but it sometimes degrades. We have to understand that AI often causes a decrease in accuracy or a completely different answer due to a change in the characteristics of the device. Therefore, from the development phase to operation and maintenance, we need to properly feed back that feedback, and we have to think about the maintenance of operation that can make us smarter and smarter like humans while we are using it. Since the maintenance of a AI model involves re-learning of the AI model, we have to think about a mechanism that can do this continuously.

What I wanted to say was that we have to think about the operation of the system as well. As Mr. Nakamura mentioned earlier, you are absolutely right. If we do not define the process for doing this and many people do not understand it, we will not be able to carry it out, even though it is true.

Taking this opportunity, I would like to explain the Appendix a little. In fact, I think that the Japanese are leading the world in terms of AI quality assurance. I am carefully watching trends in Japan and the creation of rules in the international community, but I feel that the rules coming out of Europe and the United States are more focused on the perspective of how to protect individuals from risks and how to operate them appropriately by imposing regulations.

Among the institutes of Japanese academic societies, there is a group called QA4AI. This is a consortium for assuring the quality of AI products, and I feel that it is a group of people who really like it and do it on their own. On the other hand, what is important is that more than half of the people involved are AI people, not quality assurance people. In fact, Japan is an advanced country in terms of quality assurance, and I think it is an important point that there were many people with knowledge of various quality assurance systems and mechanisms who have released Japanese products that have won the world's honor. At one time, the world had studied it.

By incorporating this in the development of AI, we believe that we can promote the application of digital technology to the world, not only in AI.

If I were to say one more thing, I would say that it would probably be better for the Government of Japan to come out with a guideline or something like this to study the process thoroughly before the regulations, and I think it would be ideal to shift to regulations only when the guideline is completed and specific things become clear.

That concludes my brief presentation. Thank you very much.

Chairman EZAKI: Are there any other members who did not speak, including , thank you very much.

In operation, you focused on AI and assurance. Process is not a simple technology, and it is extremely important, so how to share it is necessary. As an example of this, I believe you mentioned AI, which is deeply related to digitalization.

Lastly, the fourth point is that we will have approximately one hour to exchange views. I would like to hear the views of those of you who will be absent due to time constraints. If you would raise your hand in the function, I would like you to speak.

Doctor Someya, please.

Member Someya: I am Someya of the University of Tokyo. I would like to ask just two questions about Professor Ezaki's explanation.

The fourth point on the summary slide at the end refers to carbon neutrality and Minister in charge of Administrative Reform Okada. The explanation on carbon neutrality was very well understood. I would like to ask you to explain the perspective from which Minister in charge of Administrative Reform Okada needs to be considered within KGI, which I believe is a very important point.

The other point is, I agree with your point that incentives are very important because it may increase costs in digitalization. Under such circumstances, for example, if digitalization and analog coexist for a transitional period, which will only increase costs, I would like to know your thoughts on how to overcome this and introduce incentives. Thank you.

Chairman EZAKI: Are there any other members who did not speak, including Then, in terms of my Minister in charge of Administrative Reform Okada, I think that checking the supply chain will be one of the important points that you are already aware of. As for risk management, including what kind of structure the business is doing, how it is related not only to technology but also to operation, will be very important from the perspective of service sustainability. However, this will change considerably depending on the external situation, so we should continuously evaluate where the risks are and what we need to think about.

Also, from the perspective of Minister in charge of Administrative Reform Okada, we will probably come up with a broad direction of where to focus on. In a sense, I believe we must be extremely careful about how to formulate our approach to that, while being aware that the Government should not interfere too much.

With regard to incentives, I do not believe there is a single method, so the question will be how many successful examples can be shared. Then, for example, support grants as policies or support for research and development will share examples of successful and unfortunately unsuccessful cases. This will contribute to accurate and effective measures, and in the same way as business, there will be many companies that have successfully overcome difficult problems with such usage methods. The point is how to share this.

Member Someya: I understand very well. Thank you very much.

Chairman EZAKI: Are there any other members who did not speak, including Do you have any other comments?
In the chat, Mr. Kato told me that naming is very important.

Member KATO: Technology Map, a completely different thing called the Technology Roadmap comes out, so I feel like it's a waste. I think you are talking about regulations and certification, so if I say technology map, I don't look at it. I think I know more about it, but when I hear regulations and certification, I think it's a national issue. I think I know more about it, but at least I will look at it. The name represents the body, so I think it is worth considering at this time.

When you say technology, you probably talk about regulation this time. If you say technology in general, especially in Japan, there are many real technical people and real hacker people, so in a sense, it will cause misunderstanding. Therefore, I think certification and regulation are very important this time, and if you don't make it clear, your target users won't see it. At least I think I am one of the target users of this, but even if the government issues a press release saying that a technology map has been released, I would think quite objectively and probably not see it. I think it would be better to associate it with regulations or certification, which is a little easier to understand.

And as Noboru mentioned, I think the opposite is also true.

From a corporate perspective, whether it can be used or not is a matter of policy, I think. If it's just about technology, Accenture and McKinsey have great reports that you can look up on your own, but regulations and the like don't come from these digital agencies. There are METI, MLIT, and National Police Agency, and if they are organized properly, I think it will be a very valuable map and catalog. That was my first question about the naming.

Chairman EZAKI: Are there any other members who did not speak, including After all, there is a slight nuance of the kanji for regulation and the nuance of regulation.
Suga, you react to this.

The Secretariat (Suga): The next Committee meeting in Yes. When the name of this committee was decided, I think there was a substantial discussion on what to do here in the first place. This technology map is the same. Although the name is plain and has no edge, I think it is an important and substantial point of discussion, such as whether to include the keyword of regulation or to have the function of certification, and I have written a little in the secretariat's material. Even if we cannot go to certification, I think we must first discuss how to guarantee trust and how to take responsibility.

As you pointed out, we would like to make it accessible to those who are not interested in the regulations. That is what we also think. On the other hand, what I wrote a little bit in the secretariat's document this time is that after making a preliminary public call, we reconfirmed that there is a lot of value added in visualizing and verbalizing the ideas of the regulatory authorities. I think that there is an aspect that is inseparable from the work of visualizing, rationalizing, abstracting, and typifying the ideas of the regulatory authorities, who are at the top of the target users of the map. Therefore, I think it is also a matter of consideration whether to intentionally distance ourselves from regulations.

Chairman EZAKI: Are there any other members who did not speak, including . It is not a simple explanation of the rules of regulations, but what to do.

Doctor Nakamura, please.

Nakamura: Mr. , this is a question of where you stand. Mr. Esaki tends to use the expression "led by the private sector." For example, do you want us to conduct some kind of tests? Projects like the one in public call are being conducted by the private sector as well. If there are such success stories, would you put them on this map, and if there is something that can be used, would you like the government to use it? Or, on the other hand, does the name change depending on whether it is positioned as a map for the government to take the lead in using such technologies to change the world?

Simply saying that this is a technology map, it sounds like it means that there are various technologies in the world, please tell me about them, and the government will also refer to them for digitalization. Personally, I am not reluctant, and if the so-called government takes the initiative in digitalization and makes a catalog that shares successful examples with everyone, I think it will be seen differently from the outside. When we talked about whether or not to include the elemental technology in the catalog, if it is written that the elemental technology could be used in this way, it is very interesting, and I think many people will be eager to include it in the catalog and work with the government.

However, if the regulatory authorities simply think that this kind of thing can be done in the so-called world, that there is such a track record, and if we think that it is a document to be used as a reference, it will not be interesting. It will not make much sense. It will just be a website that collects and summarizes information. The Digital Agency is a government-led organization, and I hope that it will be a catalog that can publish more and more information on the digitalization of technology related to government regulations, rather than being led by the private sector.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

While there are several exits, it is necessary to be fully aware that demand must be utilized. In this environment, the government must take the lead in showing that it can change.

Nemoto-san, please.

Basic Member: I think I will be saying the same thing as Professor Nakamura, but it is important to clarify the purpose of the procurement itself on the procurement side, and I think it is necessary to make a catalog that shows what kind of things can be secured using such technology.

On page 11 of the material that Mr. Suga explained at the beginning, there was an expression about disadvantages and restrictions, but in fact, I do not understand this matter well. The government should have issued the requirement, and I think it is sufficient to say that it can be achieved. I think it is too much to ask that everything can be done. That is my first point.

There are a number of issues that need to be taken into account. One of them is how to introduce technology into areas where monopolies exist. There are a considerable number of monopolies, and this is also done as a form of regulation. New technology can be used in that area, but I think the basic stance of the current administrative agency is that only those who have a XXXX license can handle it. If that happens, it will be a form of declaring from the beginning that we will not introduce technology into areas where qualified people are operating. I would like Digital Agency to work hard on this part to a considerable extent.

In addition, as mentioned in the previous discussion, there is a tendency to require a 100-percent guarantee only when the machine is doing it, and as mentioned earlier, whether there is a function to prevent dozing at the time of testing was mentioned as an example. Even if a human being is a test supervisor, dozing cannot be prevented 100 percent. I felt that the catalog would not be completed unless the level of recognition was properly included in the requirement.

That's all.

Chairman EZAKI: Are there any other members who did not speak, including , do you feel any reaction?

The Secretariat (Suga): The next Committee meeting in Indeed, it is true that demerit information is necessary in addition to the essential requirements established through dialogue with the regulatory authorities. I think we need to take a closer look at what demerits were discussed. The authorities will first say that what is considered essential is essential. On the other hand, Toyota-san also mentioned in the chat earlier, but I would like to discuss the possibility that it is not essential in the first place, and that is my first concern. When assuming that the essential requirements are sufficiently broad, I think it is certainly a point whether further demerit information is necessary in addition to that.

Also, regarding business monopoly, I wrote on page 15 that there is no regulation. Am I correct in assuming that you meant that there are cases where there is de facto regulation?

Basic Member: To put it bluntly, if there are gray-zone answers such as legal tech that may be illegal, it means that the administrative agency is saying that it should not be done with machines. They are afraid that the theory that it is no good to replace with machines in all areas where business monopoly is conducted will become more and more.

The Secretariat (Suga): The next Committee meeting in I think the case you just mentioned is in the upper left of this figure. In other words, it may be necessary to confirm whether there is technology or not, but at least in terms of regulations, there are analog regulations that only humans can do and only qualified people can do, and it falls into the category that restrictions on the use of technology are very tight, so I think it will be considered as a subject.

Basic Member: Nice to meet you.

Chairman EZAKI: Are there any other members who did not speak, including If we do not properly organize the terminology around this, we must be careful because what is meant by monopolization differs from person to person. Also, when I say disadvantages, it may be written in the sense of points to note or something like that. I feel that it could be used in a very bad way.

, please.

Member SHIMADA: , this is exactly what I was trying to present earlier, and I wanted to say that there are things to keep in mind rather than disadvantages. You shouldn't think of AI as a magic wand, AI is a cute child like a human being, so you can't use it effectively unless you educate it properly, or something like that.

The Secretariat (Suga): The next Committee meeting in Group, proposed, when discussing the disadvantages of the technology, it may be necessary to write a note of caution about the technology in question, such as the user's knowledge that he or she must read this document before using it, or the need to prepare a disclaimer that says that it may not be possible to do this, and that there is a possibility that errors will occur when using it.

Member SHIMADA: That's right. I think most new digital technologies in the future will use AI in some way, so we need to first remove the expectation that you can use it as it is once you buy it.

Also, I would like to say that it is a kind of process when we apply them. It is a guideline, and we should study it thoroughly. When it comes to operation, there should be an operation card, and we should start to guarantee the quality as a whole. I think all of you here already know this. By carefully promoting this, ordinary people will be able to use it. This is my assertion.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

As for the operation of the guidelines and guidebooks, Kato-san wrote that the guidelines and guidebooks should be used properly.

Then, Dr. Kawahara will be here until around eleven twenty five. Thank you very much.

Member Kawahara: , when I myself was involved in various DXs, the most troubling case was that while I was thinking that the rules are written like this and should be implemented in coding as they are, in fact, there are many exceptions that are judged comprehensively in the end, and even if I try to do it according to the written rules, it does not work that way. So, although it can be automated, in the end, it is easy to decide what needs to be fine-tuned. In this way, it can coexist with the AI. Rather, I felt that what I really want to know is the best practice of how to cleanly handle ideally determined and flexible operations in the system. This is my first point.

The other point is the creation of incentives. As mentioned in Professor Someya's comment, I believe there will be more than a few disadvantages in the short term, but in the long term, there will be some advantages if we all work hard to bear the costs. I believe that is exactly the case with vaccine certifications. I believe that the creation of incentives is to create rules with strong leadership for things that can be used overseas and for other purposes, and for which the government is confident that everyone will bear the costs.

That's all from me.

Chairman EZAKI: Are there any other members who did not speak, including Thank you very much.

Incentives are a very difficult issue, and this will be an opportunity for administrative officials to show their skills, but I hope we can do a good job of collecting such information here.

In the chat, we need to set up a guideline to read the guidebook, and ask them to prepare a guidebook, including its operation and incentives.

Member KATO: When I saw this, I was the one who said I shouldn't do it because I don't know which is better, guidebooks or guidelines, so I think the naming is important. I think ordinary people don't know the difference. Guidelines and guidebooks, logic are like that.

Chairman EZAKI: Are there any other members who did not speak, including If we organize them in advance, it is very important to display the web pages and outputs created here by category.

Member KATO: If all Japanese understood the difference between guidelines and guidebooks, I think Japan would become a very strong country.

Chairman EZAKI: Are there any other members who did not speak, including Are there any other opinions?

Mr. Nakagaki, please.

Member NAKAGAKI: . I'm currently working with NITE on something called Smart Safety, and I have some concerns about it.

I think you are right about the integration of the operation of the AI. One major issue is who owns the data in competitive and non-competitive areas. For example, when it comes to security and maintenance, what is covered is some kind of asset. If the owner of the asset has a public organization, such as civil engineering or electrical equipment, that is fine. However, if it is owned by an individual company, it is the data owned by that company.

The more difficult thing is that it will be the operators who provide maintenance services. It is often the case, but I think the company that delivered the plant will take over. In that case, they want to use that data to keep it in their hands and secure their superiority. In that case, if they disclose that data or expand it horizontally, it will benefit others in the end, so I think they want to ensure confidentiality. They don't have money or time, and I think the purpose of saving as much labor as possible is to raise the level of the whole, but on the other hand, I think they should also secure the business potential of the company.

For example, wind power, which is a particularly good example, is supplied by overseas manufacturers. For example, when we try to conduct condition-based maintenance and collect data from them, the overseas manufacturers have all the data, and even the asset owners cannot access all the data. In such a situation, the data does not spread, and this is a problem for us.

I made a statement not because I raised an issue, but because of the challenges we currently face.

That's all.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

This may be what I am talking about, but there are probably some cases where this problem is shared and solved by this method in other industries. For example, in the case of the computer industry, it is said that all logs are taken by users, and in the medical industry, rules were revised so that individuals can take all the information that hospitals had in the beginning. When this happens in a global area, the problem is how to create a global consensus through cooperation between regulatory agencies and industry.

Then, as an output of the digital transformation, the output is to do it properly, and this is the output. In other words, in the consideration of advancing the actual digitalization, one of the purposes of this committee may be to escalate the work that must be overcome by any means and to focus on the policy. That is what Professor Nakagaki probably mentioned.

What do you think, Ms. Suga? You may not want to do that.

The Secretariat (Suga): The next Committee meeting in From the beginning, when I asked Professor Nakagaki to serve as a member of this committee, I thought it was reasonable that you pointed out that point as a concern.

We will ask the digital ad hoc group to make a regulatory reform so that certain technologies can be introduced. In addition to allowing the use of technology, if, for example, real-time monitoring technology can be adopted, on the other hand, monitoring data can be fully accessed by facility owners, and if necessary, the authorities can also access it. I think it is possible to show this as a set of guidelines. I would like to receive comments and advice on this as well.

Chairman EZAKI: Are there any other members who did not speak, including Do you have any opinions? Personally, I think we should do it.

Member NAKAGAKI: If there are any other examples that go beyond these areas, please let me know. I myself am very worried about this area, so I would like to learn about it.

Chairman EZAKI: Are there any other members who did not speak, including It would be good if that sort of thing could be shared, and solutions could be found in the guidebook rather than in the guidelines, in other words, there would be such policies and practices, so it would be good if we could move things forward in concrete terms.

Next, Mr. Nakamura asked a question in the chat session about whether information such as best practices and operational processes would be included in the map this time. Today, Mr. Shimada and I have both expressed the same opinion, but I believe there are multiple opinions that this should be done.

Mr. Suga, how about this?

The Secretariat (Suga): The next Committee meeting in Yes. First of all, regarding best practices, as I wrote specifically in the discussion points, I would like to post it in a way that it does not become a barrier to entry for startups.

In addition, I understood that it is necessary to provide information on the operation process in a firmly linked manner, which was also proposed by Mr. Shimada.

On the other hand, I would also like to ask for advice on how and by whom guidebooks should be compiled, and what questions should be asked at the stage of public call.

Chairman EZAKI: Are there any other members who did not speak, including , please.

Member SHIMADA: After all, I think that it is very important to be an independent organization. In short, I don't think that some people should write guidelines or guidebooks to suit their own convenience. In that sense, I think that there is a motivation for everyone to be able to use it by skillfully incorporating organizations that cut across industries. I think that a system in which experts look at it and the government can recognize it as a guidebook or issue guidelines would be good.

The Secretariat (Suga): The next Committee meeting in For example, I think that the process of public call for technology exploration will continue in the future. In that case, I would like to see a guidebook or some kind of document that clarifies the operation process set up as a column to be referred to, and those who respond to the public call are first asked to propose it. If more than one document comes out, I would like to see a process in which we assume that it may be a very neutral document that is worth referencing across industries and services, and look at it carefully.

Member SHIMADA: is a kind of industry organization that has set up a QA4AI, and I think it would be a good idea to use it. If we were to put it out in public call, it would be more like a cross-industry organization. Rather, it would be individual companies. They would make various proposals that are convenient for them, with the aim of increasing their sales. I wonder if this would really be a good thing for everyone.

Chairman EZAKI: Are there any other members who did not speak, including In that sense, in the area of industrial cybersecurity that I am involved in at METI, we are proceeding with a system in which we first create some sort of common guidelines for a certain area or industry, and then create a guidebook while looking at the guidelines for each industry. Then, if it is a good guidebook, it will be promoted to a guidebook published by the government. Do you mean it would be good if you could do something like that?

Member SHIMADA: That's right.

Chairman EZAKI: Are there any other members who did not speak, including This may be one of the best practices.
It will be very good if this kind of thing happens in various industries. At that time, there will be another point where we will not forget to take a global perspective. The story of the Michelin Guide is written in the post now.

Mr. Kawabata, would you like to talk about something?

Prof. Kawabata: We need to have some really high-level discussions here, and as an example, the Michelin Guide provides information to the users in this way, it provides information that benefits the users, it's a guidebook, and the guidelines say, for example, there are very easy-to-understand rules for one to three stars.

At the same time, there are very strict internal standards for selecting restaurants. The training of people who make the standards is exactly the same as the AI I mentioned earlier, but the Michelin Guide shouldn't be mistaken. For restaurant people, all over the world, if a restaurant falls from three stars, it's worth dying. In that case, they are required to be as strict as machines, and the training of those people is as strict as setting global standards and training them all by employees. I think I can imitate that. I think it's great that it has been going on for 100 years.

And the important thing is that Michelin, which is the main body of it, has nothing to do with the restaurant operation. When asked about the purpose of what it is made of, Michelin says that it is making the guidebook with the aim of having mobility used in safe and secure, because if mobility is used in safe and secure, it will drive a lot of distances and go far away, and as a result, it will be good because tires will be sold.

What I said earlier about the difficulty of setting incentives is here at all. The motivation to build such an advanced restaurant or a very comfortable hotel, the incentive is to be evaluated by Michelin in the end. It will appear in the guidebook. Michelin hopes that its tires will eventually sell. It is a very long route, but I think it has created a fairly correct route. There are quite a few good points in guidebooks and accompanying guidelines that have been around for a long time for this kind of incentive design, so I was asking if the incentive design will proceed smoothly when the discussions so far have become realistic.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

If we put them together, it may become a document that Kasumigaseki or local government officials refer to the most. Since incentive design is a very important part, it may be good if various examples are essentially organized.

Prof. Kawabata: There are quite a lot of examples like this, and I have covered quite a lot of successful systems in the world, so although it is not digital, the way to create the system is actually the same, and I think it would be good to make it digital so that it can be done quickly.

Chairman EZAKI: Are there any other members who did not speak, including , please.

Member SHIMADA: I talked about incentives. In terms of actual incentives, for example, the amount of money a company is paid is already limited, which is often the case in regulated industries. The system is designed so that even if we reduce costs ourselves, it does not allow us to reinvest in our company, for example.

Chairman EZAKI: Are there any other members who did not speak, including Of course, there are various multi-faceted incentives, so I would like to ask you how you are solving them. There are some know-how as a company that cannot be disclosed, but if there is something that can be shared, I wonder if it would make everyone happy if I could refer to it well as a successful example.

Ogawa-san, please.

Member Ogawa: I'm Keiko Ogawa. Nice to meet you.

I think there was a lot of very useful information. There was some discussion about guidelines, but I would like to talk a little bit about the trust collateral framework that was in the last paper. When thinking about this framework, I think it is very important to clarify the roles of Digital Agency and each regulatory authority from a role and responsibility perspective.

For example, as a framework to secure trust, Digital Agency will establish rules for upper-level governance policies. This will include monitoring and verification. Each ministry and agency will also have clearly defined control responsibilities in accordance with these upper-level rules. For example, each ministry and agency will identify specific risks by following the basic policies for risk identification established by Digital Agency. I think it is very important to build a governance system that clarifies each department's roles.

What are the risks at that time? I think we need to organize them so that there is no confusion. For example, as for risks from a technical perspective, as Mr. Shimada mentioned earlier, which is very helpful, there are issues such as black-boxing of AI or biased data. In addition, there are issues of RPA and robotics, and third party risks if cloud is used. Even if it is perfect to identify and capture all the risks unique to the latest technology itself, if we lack the identification of risks that hinder the original purpose, I think the final purpose of the regulation will not be achieved.

Therefore, how to organize the definition of risk is very important. In addition, we need to make decisions on how to respond to that risk. I think it is very difficult to completely reduce the risk to zero. This is especially true for responses to fraud, but we are always in a state of cat and mouse with our responses to fraud. Therefore, in the end, other committee members mentioned earlier that clarifying rules and processes is very important, and I believe that this is also very important in fulfilling accountability in the end.

A general risk approach is to comprehensively identify the risks that are expected, analyze the degree of importance if they are actualized, and depending on the significance of the risk, take a risk that has not been dealt with so far, prepare controls to reduce the risk, avoid the risk, or transfer the risk. Make sure that you can visualize and explain such a risk approach. In addition, the situation changes dynamically all the time, so I think it is necessary to have a framework to respond to agile with PDCA.

What we need to consider this time in particular is that small and medium-sized companies and new startups will be included. In that case, the issue of cost will also come up. I think there was a discussion about incentives, but I think there will also be enough discussion about feasibility, such as whether we can deal with risks in the first place. It will be important to what extent we can take risks in balance with risks and fulfill our accountability to the people. Therefore, I think it may be possible to prioritize from fields where risks are not so high and risk response costs are not so high.

Also, I think there was a discussion on security earlier, but there are two types of risks: general risks and risks specific to each project. For example, there are generally related risks such as Minister in charge of Administrative Reform Okada issues including the supply chain, cyber security risks, and information risks, which I mentioned earlier, and project-specific risks, such as cheating on this project. The nature of these risks is different. In particular, general risks need to be dealt with in a cross-sectional manner so that there is no overlap.

As the committee members mentioned earlier, I am aware that various measures are being taken in other areas as well. Since various measures such as Zero Trust Architecture in June will be taken, I believe that it will be necessary to proceed with overall risks in a manner that does not overlap or conflict with discussions on such issues.

Also, this time, one concern is data. I saw the standard guidelines for the promotion of digital government, and there are data strategies. The primary objectives of this time are the automation and digitalization of analog administration, but on the other hand, the collected data, of course, besides the issue of who owns it, I think we cannot ignore the organization of data strategies, such as how the government will utilize it.

In the business world, everyone is starting to put up data strategies with DX, but even if they want to do it, there are so many places where there is no usable data in the first place or compatibility is not guaranteed. As you know, in terms of AI, everyone is currently taking on a lot of challenges, but despite the fact that there are very few AI human resources in the market, in reality, about 75 to 80% of the work of AI human resources is data collection and cleansing. Therefore, data strategy is not the scope of this time, it can be done by others, it can be done later, and when data strategy is done later, it can easily become a very heavy task. Therefore, for example, the compatibility of the generated data, interoperability, standardization, and which data should be targeted, I think it should be considered to some extent.

Lastly, for example, when conducting verifications or audits, we need to pay attention to the fact that there will be a huge backlash later, such as when audit trails or digital trails are not available even though digitalization has actually been made. There may be cases where audit trail data is lost after a certain period of time, time logs are not available for a sufficient period of time, they cannot be retrieved through searches, or verification cannot be conducted ex-post facto, even though trails are available. I believe it will be important to design conditions, including for subsequent monitoring.

Chairman EZAKI: Are there any other members who did not speak, including audit.

Some people have not yet spoken, but I believe that Professor Saito and Professor Matsuo of IPA will also be present, and if they would like to speak, please do so.

Prof. Matsuo: Regarding the AI that Mr. Shimada mentioned, I think that's exactly right. I believe that it is extremely important to firmly grasp these points in order to utilize AI.

Regarding the Technology Map, Mr. Suga organized it very neatly, and I understood it very well. In order to make it successful, we need the cooperation of many people, and since it has a strong social aspect, how to make it into a system that can be used, and how to turn it into a service, are probably important perspectives. In that sense, I thought it would be very important to start running the PDCA cycle as soon as possible, and to accumulate ideas so that many people can use and participate in the Technology Map through various trials.

That's all.

Chairman EZAKI: Are there any other members who did not speak, including .

Saito-san from IPA, please.

Member Saito: , I have heard a lot about you from earlier, and I would like to tell you something that I am a little concerned about. Basically, this time, the Digital Council, which is active in Digital Agency, has talked about creating an ecosystem in which guidelines and guidebooks will be created and circulated. In a sense, from a platformer's point of view, I think it is good to talk about how to activate it by giving incentives.

When Mr. Shimada brought up the issue of quality assurance, he went so far as to say that a framework should be created and implemented properly. For example, when a new technology is applied to a place where normal regulations are being implemented, he went so far as to say that under what kind of framework it should actually be implemented. In other words, he said that it should be created within this committee.

Then, in a sense, I came to have an image that the ecosystem itself has to consider the framework and process of how to handle data such as the current quality assurance. And, according to what Ogawa-san said earlier, the discussion has developed to the point where we are thinking about how to collect data here and how to handle it, so I need to confirm once again where Digital Agency is thinking now, otherwise I don't know where I stand. I don't know how far I should consider, so I would like to confirm a little. May I?

Chairman EZAKI: Are there any other members who did not speak, including , the first thing I would like to answer is that if there are so-called missing areas, there are probably ministries and agencies in charge, and I think the action is to properly work on them. Digital Agency will create them if they are under the jurisdiction of Digital Agency, but for other areas, the supervising ministries and agencies or the industry will create them properly. I have introduced the topic of industrial cybersecurity, and when such guidelines or guidebooks are created, I think the guidebooks will include the repository or technology map prepared by the Digi-Agency, or the name has not been decided yet, but I think it will be possible to refer to them in one-stop.

Member Saito: I would like to add one more thing based on what Mr. Ezaki just said. Then, for example, when we actually started operating it, each ministry and agency looked at some guidelines, and when we operated it, some trouble occurred, some incident occurred, at that time, on the contrary, there was talk of coming back here and correcting the so-called process or framework, or not?

Therefore, regarding the question of how much the Agency for Digital QA4AI will do, in a sense, I think it would be good if the Agency for Digital is responsible for the common aspects of the overall management and final use of the system, and if we are responsible for such aspects ourselves. This time, assuming the so-called life cycle after the creation, if we need to develop the framework itself in a more appropriate manner, in addition to the operation and maintenance of the current guidelines and guidebooks, if it is also done by the Agency for Digital TV, I think there will be a discussion about entrusting the operating organization itself to an external organization, as I mentioned earlier. I imagined that it would probably be necessary to talk about retaining the ownership of the Agency for Digital TV while entrusting such matters to an external organization. I wondered what to do about this.

Chairman EZAKI: Are there any other members who did not speak, including , please.

The Secretariat (Suga): The next Committee meeting in At this meeting, I believe that you are holding discussions that will create a function that does not yet exist. First of all, I will thoroughly discuss the need for such a function in the government. I will also discuss the need for the parties responsible for the function to take on these responsibilities as a set. I believe that discussions can only take place in the order of who can take on these responsibilities.

However, we cannot simply ask for the introduction of technology for nothing and leave the rest up to us. At this point, I think we need to take responsibility and consider the matter here to some extent.

Member Saito: Understood.

Chairman EZAKI: Are there any other members who did not speak, including I think that Mr. Saito pointed out that we need to solve very difficult problems properly, and in the chat, Mr. Noboru, who is related to this, said that in a very strong organizational structure, people with diversity will not move, so unless we create a structure that can move more independently, it will be dangerous to create a very large or very firm framework.

Finally, I think we received a very difficult assignment in Digital Agency.

Mr. Noboru, please.

T: I posted it on the Chat, but this second meeting was also a project that I myself worked on at IPA and METI in the past, and there are examples of situations that looked bad at first but turned out to be very successful. What is important is that we have a very diverse and valuable group of members. And I think it is a wonderful gathering of people who are not only willing to give their expert opinions, but who are probably willing to put their hands to implement everything that was said.

At the same time, I believe that there will never be a convergence of ideas and fields of expertise. If we proceed on the basis of planning based on the consensus of all, we will end up implementing things on the basis of the lowest common denominator, and there is a risk that we will not be able to make the most of the abilities of each of us and that we will not be able to create any noteworthy social value. In general, I was not sure about this, but I think that if we continue to hold meetings like this for two to three months, and hold these interesting meetings, we will naturally be able to determine the most common range of fields. If we consider this to be a single company, I think that we should also establish a code of conduct within the scope of what we can do, but not to the extent that we can do it.

After that, after about three months, after that, for one to two years, I think it would be good for each person to independently proceed with what I mentioned earlier at their own discretion and responsibility. Once the framework is decided, there is an agreement that it is okay to do most of it comprehensively, so I think it would be good for each person to do it as they wish, without getting permission and agreement one by one or making adjustments because of disagreements with other people. This is the stage (2).

With regard to (4), I believe that if we do so, you will be able to produce a wealth of results. This may be inconsistent because it is not based on planning, but I believe that the results will cover a wide range of areas in depth. And since everyone has a natural ability to cooperate, I believe that in fact, the results that seem to be inconsistent will naturally combine and organically move to produce excellent results as the Digital TV Agency. Based on my own experience, I thought that this way would have a higher probability of success than the method of proceeding through consensus building.

That's all.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

In that sense, as I believe some members of the Council have said that it is important to move things first, I think we will move what we can without waiting, and we will revise things as we move them. In this process, as Mr. Saito mentioned, we will pay attention to what kind of large framework the ecosystem will be built within.

It's hard for you, Hitoshi, but let's do our best together with your cooperation.

The Secretariat (Suga): The next Committee meeting in I was a little happy because I thought that you had just made a proposal in the direction of easing the secretariat. THANK YOU!

Chairman EZAKI: Are there any other members who did not speak, including ?

The Secretariat (Suga): The next Committee meeting in Toyoda members via chat.

H. Toyota: In , it may be a little different from the current discussion. It may lead to meta, but rather than a big project, this time, for example, we will have a trial public call, but it will be an individual application or something that can be implemented by each company with a small granularity, so I feel that the granularity is not easily connected to the current discussion.

Also, from the perspective of those who provide information, if each company lists what it can provide, it will end up just listing what it can do now, and new industries and technologies will not be created. Therefore, unless a mechanism for sharing information as a social sub-layer or a mechanism for sharing various shares as hardware rather than just information on the environmental side is probably created in this area, true information using it, reediting of the industrial layer, and re-valorization will not essentially be created.

This is also chicken and egg. By doing what is called a technology map, if we arrange them as a map on a single basin, the granularity will not be uniform and they will be arranged randomly. Therefore, I thought that it would be necessary to collect them firmly from the perspective of using them to create vertical layers as subcategories. If possible, I thought that it would be necessary to have a two way system in which we can present such things as forecasts at the beginning and have them submit requests for those things.

That's all.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

In that sense, it may be that this committee needs to define and create not only the repository mentioned in the chat, but also something like a mountain extracted from the repository strategically or empirically. As an opinion, it is a common opinion that a mere repository is no good.

The Secretariat (Suga): The next Committee meeting in Secretariat, I wish I could write that we are starting to do that on page 17, but I am afraid I have not been able to explain it well. I do not want to talk about individual application layers such as training and exams, but when I include them and make them abstract, I am motivated to think that it would be good if we could structure the technology that plays a function in what it is talking about.

Chairman EZAKI: Are there any other members who did not speak, including Thank you.

There is still a lot to discuss, and I think it would be a big deal if we discuss while drinking alcohol. However, the time we have received is up, so I will conclude today's agenda.

Finally, I would like to ask the secretariat to explain the next meeting of the Committee.

The Secretariat (Suga): The next Committee meeting in will be the last of the year. It is scheduled for December 1 from 10:00 to 12:00. The Secretariat will contact you again for further details.

The proceedings of today's meeting will be announced on our website at a later date after the secretariat has confirmed the draft minutes with the attendees. In addition, if there are no particular objections to today's materials, we would like to make all of them available on the digital TC's website.

It would be greatly appreciated if you could join us for a few moments for such a high-level discussion. Thank you very much for attending the committee meeting today.

Chairman EZAKI: Are there any other members who did not speak, including Thank you very much.

Now, that's all for today's meeting. Thank you for your hard work.