The 4th Meeting of the Sub-Working Group on Social Rules for automated driving vehicles in age of AI

Overview

Date and Time: March 29, 2024 (Friday) from 13:00 to 15:00
Location: Online
Agenda:
1. Opening
2. Proceedings
  1. Explanation by the Secretariat (Main opinions of the Third age of AI Sub-Working Group on the Review of Social Rules for automated driving vehicles and Possible Issues Based on the Opinions)
  2. Automated driving and Civil Liability
  3. Summary of Judicial Precedents on Negligent Offenses - Focusing on Automobile Accident Cases
  4. Issues raised by members
3. Adjournment

Materials

Minutes

Councilor Kodama: First of all, I would like to make an administrative contact. Today's meeting will be held completely online. Members are requested to turn on the camera during the meeting and to unmute the microphone when they speak. If anyone else is speaking, please mute them. In addition, spectators are requested to turn off the camera and microphone.

Next, I would like to check the materials. As stated in the agenda sent in advance, the materials include the agenda, the member list, the secretariat explanatory material, the member Goto explanatory material, the member Imai explanatory material, the member Takahashi explanatory material, the member submission material, and the list of attendees. If you do not have them, please contact the Teams chat function or the secretariat by email. As for the attendees today, due to time constraints, I would like to distribute the list of attendees in your possession instead of introducing them. In addition, I heard that the member Ochiai will leave the meeting for about 30 minutes.

In addition, as in the previous meeting, Mr. Ohno and Mr. Suganuma from DADC are attending today as guests. Please note that the materials and minutes of the plenary meeting will be made public at a later date.

Then, I would like to ask Mr. Kozuka to proceed from here. Mr. Kozuka, please.

Mr. Kozuka: Thank you very much for your visit to . Thank you for taking the time to attend our meeting at the end of the fiscal year. Then, I would like to proceed according to the agenda.

Today, first of all, Mr. Goto will talk about civil liability. After that, Mr. Imai will talk about criminal liability. This time, we will hear stories on completely different themes, so we will exchange opinions for each teacher's presentation. After that, Mr. Takahashi has submitted materials, so we will raise questions based on this.

Prior to the presentation by each member, I would like to receive an explanation from the Secretariat. Then, Counselor Suga, nice to meet you.

Counselor Suga:
(Hereinafter, explanation will be made based on "Material 3: Explanatory Material for the 4th Meeting of the Secretariat")
Material 3: Based on the materials of the 4th Secretariat, we explained the main opinions at the 3rd SWG and the expected issues based on the opinions at the 3rd SWG.

Mr. Kozuka: Thank you very much for your visit to . I feel that the assumed points of discussion have deepened thanks to your active discussions every time.

Let's move on to Agenda 2 - (2). We have materials prepared by Mr. Goto on how automated driving and civil liability should be. Mr. Goto, thank you very much.

Goto Members:
(Explanation based on "Material 4: automated driving and Civil Liability")
Thank you very much. I received a request from the Secretariat to give us an idea on the whole of civil liability, so I would like to talk to you.

Please show me the second page of the slide. Although there are some parts that are preaching to the Buddha, I think that it is important to confirm what the civil liability system is for and how it is design as a starting point when considering the civil liability system, so-called damage compensation law system design. The two purposes of the civil liability system are to deter accidents and illegal acts by internalizing the damage to the perpetrator and to recover the damage to the victim. In traditional civil law, the main purpose is to recover the damage, but based on the concept of economic analysis of the law, I think that at least the prevention of accidents can be regarded as an important purpose along with or even more than the recovery of the damage. I would like to explain not by saying which is more important but by understanding that both are important purposes.

In addition, neither the prevention of accidents nor the relief of victims necessarily has to be achieved through the civil liability system. Since there can be other means, I think it is important to consider their complementary relationship. Regarding the prevention of accidents, there are criminal penalties that have been discussed so far, and there is also a safety regulation by the government. In the case of automated driving vehicles, I believe that the spread of automated driving vehicles will have the effect of suppressing accidents by reducing the number of accidents caused by human error. Regarding compensation for damage to victims, even if it is not based on compensation for damage, there is a social security system such as the vaccinations Health Damage Relief System for side effects of vaccines, so it is possible to respond in this way. In addition, for traffic accidents, there is compulsory automobile liability insurance as a compulsory insurance under the Automobile Liability Insurance Act. Liability insurance includes not only compulsory but also voluntary insurance. As for automobile insurance, I think it is common to take out voluntary insurance in addition to compulsory automobile liability insurance. In addition to liability insurance for the wrongdoer, it is possible for the victim to purchase a first party type of non-life insurance. Covering the damage in this way performs the same function in terms of damage recovery, although there is a question of whether it is justified or not. In light of the fact that there are other such options, I believe that the policy goal of the design will be to simultaneously achieve two objectives. One is how to provide incentives to manufacturers for automated driving vehicles development, safety assurance, and safety improvement. The other is how to recover the damage to victims of accidents that have occurred despite the expectation that the total number of accidents will decrease due to the spread of automated driving vehicles. I believe that the policy goal of the Issue will be to simultaneously achieve two objectives. Next page, please.

As you have already been informed, I would like to confirm what will happen if the current law on civil liability relating to automated driving is applied, dividing it into loss to persons and property damage. In the event of damage such as the death or injury of a person, the liability of the operator under Article 3 of the Automobile Liability Act will be applied. Unlike in Article 709 of the Civil Code, this is considered to be strict liability, and unless the operator can prove the so-called three conditions for exemption, the operator will be liable. In addition, regarding the liability of the operator, compulsory automobile liability insurance is mandatory, and in the event of the death of a person, up to 30 million yen will be paid out. As for voluntary automobile insurance to be added to this, according to statistics from the Non-Life Insurance Rating Organization, about 75% of liability insurance for bodily injury is covered by voluntary insurance, and 88.7% of liability insurance for bodily injury, including automobile mutual aid, is covered by voluntary insurance. In the case of voluntary insurance, the maximum amount of liability for bodily injury is unlimited at 99.6%, and automobile liability insurance is up to 30 million yen, but in the case of about 75%, or 88% or more depending on the perspective, the insurance will fully cover the loss to persons without limit. There is also product liability, but I would like to omit it because Mr. Fujita told me about it last time.

For property damage, the operator's liability does not apply, so there is only liability for tort as the liability for negligence under Article 709 of the Civil Code. In this case, the victim is responsible for proving the negligence of the perpetrator. In addition, there is no compulsory automobile liability insurance, so if the tort can be proven, it will be a problem whether or not you have a voluntary insurance. The participation rate of liability insurance for property damage is also about the same, and the maximum amount is unlimited in 96.2%, which is a little low, but in many cases, you have liability insurance without a maximum amount. The above is the system applied by the current law. Please show me the next page.

In terms of what kind of premise we will be considering in the future, first of all, we assume a situation in which a automated driving vehicles of level 4 or higher is assumed and the passengers cannot intervene. I think there will be various forms of use in the future. It is often discussed on the assumption that the so-called owner car will be replaced by a automated driving vehicles, but in reality, it is assumed that there will be a business use, for example, an unmanned taxi, an unmanned community bus, or a fleet of trucks, with the first car being driven by a person, followed by an unmanned truck. In the case of such business use, if it is an unmanned taxi or an unmanned bus, there is probably a remote monitoring center, and it seems that there is room for it to intervene in the event of something happening. Since it is a business operator, I think it can be pointed out that there is a possibility that it is mandatory to take out not only compulsory automobile liability insurance but also unlimited liability insurance for property and personal liability when operating under the Business Law. However, in the case of business use, if you think about it, it is possible that a business operator uses automated driving vehicles to transport its own goods. In that case, I think it will be arranged closer to the owner car, but in any case, I think it is necessary to be careful that there are two types. In any case, as an application of the Self-Compensation Act, whether it is level 4 or level 5, I would like to confirm here that there is an operator.

That being said, with regard to the operators and users of the automated driving vehicles, on the premise that there is no negligence in vehicle maintenance and software updates, I believe that when accidents occur, it is a malfunction of the vehicle. Last time, Dr. Fujita talked about how to understand the problems of automated driving vehicles. Overall, the number of accidents decreases, but in very rare cases, there are cases where accidents that would not be caused by humans occur. In such cases, what should be done was the focus of the discussion. I believe that this will probably be the most difficult case, but in reality, I believe that other problems may occur. For example, a simple programming mistake may result in an incorrect input, or a bug may not have been discovered at the time of creation, but it may appear in some case. In addition, as a result of deep learning, there may be various forms such as unexpected behavior of the system, so I think it is necessary to consider them. For example, in the case of overlooking a simple bug, I think from the previous talk from the Japan Automobile Manufacturers Association, we are making it so that such a thing does not happen. I think that is very desirable, but even so, mistakes will occur, so I think that we must think about that case. In addition, not only the story at the time of program creation, but also the possibility of update afterwards must be analyzed in more detail, but for the sake of time, I have made a simple explanation. Next page, please.

Therefore, in terms of what kind of civil liability system should be considered, three major options are shown. (1) is the idea that the cause of the accident is not the operator's responsibility side but the manufacturer's side, and that the operator's responsibility should be abolished or made negligent so that the manufacturer's responsibility becomes heavier. To be more specific, the burden of proof for defects in product liability is currently borne by the victims, but unless the manufacturer can prove that there were no defects, the manufacturer is responsible. On the other hand, doing so may hinder the manufacturer's development incentive. If the operator is not doing anything wrong and has achieved a certain standard, neither the operator nor the manufacturer will be responsible. Then, a public compensation fund system funded by the manufacturer, for example, in the case of vaccines, will be created, whereby the victims will be compensated, and the security of the automated driving vehicles will be secured by the security regulation of the administration. This is option 2. ③ is an approach in which the current law is maintained and the operator assumes responsibility first, and if the cause is on the vehicle side, the operator or its insurance company effectively makes a claim against the manufacturer based on the Product Liability Law. What was adopted in the report made by the Ministry of Land, Infrastructure, Transport and Tourism in 2018 was to make it the third plan tentatively. From the perspective of providing prompt relief to the victims, I believe that (iii) is the most effective.

There may be other variations, but regarding the above three, I have considered how the policy goal of preventing accidents, which I mentioned at the beginning, and the policy goal of compensating victims for damage, will be. First of all, regarding the prevention of accidents, option (1) is responsible for the manufacturer who can control the safety of the car the most, so I think it is the most solid motivation for the manufacturer to internalize the cost of the accident.

In (ii), as long as the manufacturer achieves a safe regulation, it will not be subject to civil liability. In this case, whether the prevention of accidents is sufficient depends on the content of the safe regulation by the administration. If it is too loose, it will be insufficient, I think, but on the other hand, if the safe regulation by the administration, in particular, the advance regulation is too strict, the development of the automated driving vehicles and the social implementation may be delayed, so it is difficult to adjust the amount. You mentioned a guideline, but I don't think I can say anything about it without looking at its content. Regarding the content of the guideline, for example, if it is abstract, such as preventing accidents, there is a question whether it will be a meaningful guideline for the manufacturer. As a layman, I think that it cannot be judged unless it is a little more numerical, but I would like to learn about this while listening to the communication between the authorities and the manufacturer.

③ is an option that is currently being taken, but in the end, it depends on how effectively the vehicle operator and the insurance company will make a claim against the manufacturer. As I said in the middle of the meeting, considering an unmanned taxi, for example, as a mode of use, in the case where the vehicle operator and the manufacturer actually belong to the same corporate group, for example, Google Taxi, it is not necessary to consider the problem of the claim so seriously. This may be sufficient depending on the fact that the vehicle operator and the manufacturer are more closely connected, and the contract there may be able to handle it, and it may be possible to supplement it with a safe regulation.

Next, with regard to the relief for the victim, as it is now, the operator is responsible for (3), and both compulsory automobile liability insurance and voluntary insurance are provided for that, so I think quite a wide range of protection will be provided. In contrast, in the case of (1), the manufacturer will be responsible. The manufacturer will probably be covered by a solid liability insurance, so the level of protection will probably not be very different. However, the possibility that the absence of the defect will be proven is more likely when the manufacturer proves it than when the operator tries to prove it. Therefore, (1) may be slightly more likely to be free of liability. However, I cannot say that point clearly yet.

From the perspective of compensation for damage to victims, the option (ii) is problematic. At present, in addition to automobile liability insurance, there is voluntary insurance, and in many cases the maximum amount is unlimited. In the case of (ii), you will rely on the public compensation fund system, but in the case of the public compensation fund system, I think the maximum amount is generally set. In the case of vaccines, I think it would be 40 million yen in the case of death. Compared with the current liability insurance for bodily injury, which does not have a maximum amount, which fully covers damage with insurance money, the level of relief for victims will be reduced, and in that regard, there is a concern that it will give a negative impression on the social acceptability of automated driving vehicles.

In that case, (2) is quite difficult to read, and I think that (1) will be a considerable change in the system, so (3) was also taken by the Ministry of Land, Infrastructure, Transport and Tourism's study group. However, in the case of (3), there is a problem of how to think about the problem of assuming product liability by the so-called operator. How to continue to provide incentives to manufacturers who should be responsible for product liability can be replaced by the story of the reimbursement and the safety regulation that I mentioned earlier, but there is still a story of how to think about the operator being subject to civil liability. In this regard, there may be a sense of burden that you can sue, but I think that the operator will not pay out of pocket if you join the voluntary insurance, so in that case, I think it will be a matter of paying the insurance premiums. In terms of paying the insurance premiums, if the introduction of the automated driving vehicles reduces the number of accidents, for example, if there is a merit that the insurance premiums are significantly lower when using the automated driving vehicles, I think it is possible that the burden is not so heavy as to impair the sense of satisfaction when compared in total. This is only a possibility, partly because it depends on the amount of money, but I think it is necessary to consider it together.

In this way, it is difficult to reach a conclusion, but in addition to what I have mentioned today, I believe that it is difficult, but impossible, to make a difference in the way of responsibility depending on, for example, the form of use of the automated driving vehicles, whether it is an owner's car or a business use. In addition, I have previously said that there are some difficulties when considering the case of overseas manufacturers. However, I believe that this is a decisive story because it will remain regardless of which option is taken.

So far, I have considered only the case of loss of life. However, I would like to briefly point out at the end the problem of what to do in the case of property damage. In the case of property damage, there is no liability of the operator under the current Automobile Liability Act in the first place, so it is only liability for tort. In that case, the users of the automated driving vehicles are not at fault, so the victims have no choice but to sue the manufacturer based on the PL Act. It will be difficult, so what will the victims do in that case? If the PL Act is not strengthened, the victims may have to rely on their own non-life insurance to cover the damage. If this will be a big problem, I may be mistaken, but in many cases, the property damage in a traffic accident will be other cars or buildings near the road. In that case, I don't think it is unlikely that the victims will take out non-life insurance. If it is other cars, I think they have my own automobile insurance, and if it is buildings, I think they have fire insurance, so I think that can be covered. On the other hand, in the case of loss of life, it is difficult to expect that pedestrians will take out non-life insurance by themselves in case they are hit by a car, so I think that is the big difference between loss of life and property damage.

I am sorry for the inconclusive story, but I have announced what I have thought so far. That's all.

Mr. Kozuka: Thank you very much for your visit to

Yokota Member: I'm Yokota , thank you for your nomination. Also, Mr. Goto, thank you for your explanation.

I would like to explain in accordance with the materials. On page 2, there is the content of the 2018 MLIT Review Meeting, which Mr. Goto also mentioned. As described in blue letters, I believe that Proposal 1 is appropriate, which is to consider a mechanism to ensure the effectiveness of the exercise of the right to obtain reimbursement from automobile manufacturers, etc. by insurance companies, etc., while maintaining the conventional operator liability for damage caused by accidents during the use of the automated driving system. This falls under Proposal 3 in Mr. Goto's explanation. Regarding proposals 2 and 3, I believe that they call for a certain burden on manufacturers, etc., and they fall under proposals 1 and 2 in Mr. Goto's explanation. Please go to the next page.

This is a summary of the contents of my remarks at the previous Sub-Working Group up to the third meeting based on the discussions up to now. (1) is about clarifying the compensation body and building a compensation system without omission. I understand that if the prior and general approach discussed at the third Sub-Working Group is adopted, there is a possibility that no one can be held accountable, although it is limited, for what is considered to be negligence if it is a human driver. In that case, I believe it is necessary to clarify how the compensation system will be built. For example, I believe that (1) the national government, (2) providers of automated driving vehicles (automated driving service), or (3) users of automated driving vehicles (automated driving service) are candidates.

The other point is from the perspective of ensuring prompt relief for victims, and I was speaking because I thought speed was also important. First of all, in the case of manual driving, we may ask drivers and others about the situation of accidents, but in the case of automated driving, there is a possibility that the payment of insurance proceeds may not be implemented or may be delayed because there are no drivers. From this perspective, in order to ensure prompt relief for victims, I believe that data recorded in automated driving vehicles is necessary. At present, for Level 3 automobile liability insurance, we are constructing a scheme in which automobile manufacturers provide data recorded in automated driving vehicles to insurance companies for the purpose of investigating the cause of accidents. If this system can be constructed at Level 4 or higher, I believe that prompt payment of insurance proceeds will be possible.

The third point is that when full-scale automated driving society arrives in the future, the number of accidents is expected to increase, but if it is impossible to determine whether or not there are defects unless all accidents are investigated, it will take time to provide relief to the victims as the number of cases increases. We believe that there is a possibility that we will not be able to make a practical decision unless we make a simple judgment on the presence or absence of defects based on certain guidelines and investigate only the parts where it is difficult to make a judgment.

Finally, we are currently making special contracts that respond to such automated driving society. In the past, bodily injury liability and property damage liability insurance covered cases in which the legally insured person was liable for traffic damage. However, this special contract for victim relief costs, although the name differs depending on the insurance company, is a special contract that provides prompt victim relief by covering the amount of damage caused to the victim first as an expense insurance in the event that the insured person was not legally liable for damage. Please also refer to the reference materials on page 5 of the materials. Please go to the next page.

The fourth slide is a summary of how the Non-Life Insurance Association of Japan views Dr. Goto's three proposals. The perspective of accident prevention is not listed here, but it is an evaluation of the perspective of victim relief. With regard to Proposal (3) and Proposal (1) by the Ministry of Land, Infrastructure, Transport and Tourism, I believe that there is no sense of retrogression from the On the other hand, I believe that Proposals 1 and 2 have some problems, as described here. It has been pointed out in the practical part that in the case of converting the burden of proof of the defect of the product liability of Proposal 1 to the manufacturer side, the victim side may have a little more burden when responding to the incident because it will not be possible to provide a detailed response in the conventional automobile insurance, for example, an out-of-court settlement agency. In addition, from the perspective of speed, it may take time to determine whether or not there is a defect. Regarding the introduction of the public compensation fund system in Proposal 2, I think it will depend on the mechanism, but I am a little concerned that there may be a delay in the timing of payment compared to the current system, and as Mr. Goto said, if we include bodily injury liability insurance, there is a possibility that the relief for victims may be reduced while many current policyholders are unlimited. In the column of "Understanding and questions about the proposal" in the center of the document, the parts marked with * are considered to be points of contention when Proposal 1 and Proposal 2 are adopted. With regard to Proposal 1, I believe that the issue is how to switch to a different legal system at Level 4 or higher while continuing to use the Self-Compensation Act and the PL Act in places with low levels in automated driving. Regarding Proposal 2, I believe that the issues will be how to conduct damage assessment, especially in the form of actual loss payment, which is currently conducted in the insurance industry, how to offset negligence if the victim is negligent, and how to determine the amount of investment by the manufacturer. Regarding Proposal 3, Mr. Goto said in his explanation that he would be able to obtain a sense of satisfaction, but I believe that the responsibility of the operator continues to arise for the owners and users even in the world of automated driving, so I do not believe that it is fundamentally impossible to obtain a sense of satisfaction. In addition, you mentioned insurance premiums, but in terms of whether it is inside or outside so-called ODD, the burden of insurance premiums in the outside world should be basically the same from Proposal 1 to Proposal 3, and since the number of accidents is expected to decrease in ODD, I believe that the burden of insurance premiums in that part will decrease and become more appropriate.

Finally, since there was also a reference to property damage, I would like to briefly touch on it. At present, even for damage for which the perpetrator is unknown, as Mr. Goto mentioned, we are taking certain self-defense measures such as vehicle insurance. In addition, given that the legal arrangements for loss to persons and property damage are currently different, while clarifying the legal liability for property damage, for example, I believe that it is an option that could be considered to expand the mandatory vehicle liability insurance, which is currently applied only to bodily injury, to property damage as an insurance for the perpetrator. However, since the system for this will be significantly different from the current system, I believe that it is necessary to carefully consider it based on the consistency with the arrangements so far. That is all.

Mr. Kozuka: Thank you very much for your visit to Yokota. We have many professors who would like to speak. Mr. Takahashi, Mr. Inadani, Mr. Nakahara, Mr. Sato, in that order, please. First of all, Mr. Takahashi, nice to meet you.

Takahashi Member: , I think that the public compensation system in (2) is a little bit bad. In the past, at the time of the Former Tomorrow's Meeting, there was a movement to expand the benefit System for Crime Victims and create a new Compensation System for Victims. At that time, a bureaucrat said that why the budget should be used to fully compensate only crime victims is always a problem. For example, it was pointed out that the government does not fully compensate those who are born with disabilities or those who were damaged by natural disasters, but why it is unfair to fully compensate only crime victims. I think that a similar problem will occur when the public compensation system is introduced. I think that the question will be why only the victims of automobile automated driving must receive full compensation from the government. Then, what will happen if a compensation fund is established? This time, it will be no more than a condolence payment, so it will not be a relief for the victims, and I think that this is also a problem.

Regarding (iii), from the viewpoint of the operator, even though it is a automated driving and does not control anything about the operation, even though it has an insurance company, I think that it is formal, but why should the operator be the accused and be sued? I think that is a bit of a drawback. However, from the viewpoint of compensation for victims, (iii) is the most effective. That is all.

Mr. Kozuka: Thank you very much for your visit to Takahashi. Then, Mr. Inadani, nice to meet you.

I think there are some other points. For example, as one of the hard cases that was not mentioned this time, considering the future implementation of things like connected cars, I think it is necessary to think about how to think about the failure of cooperation with road infrastructure and how to conduct the investigation. It is not the issue of civil liability itself that you talked about this time, but I have listed some hard cases, so I have listed them.

In addition, in the material I received from Mr. Yokota, control of operation is also mentioned, but in the case of automated driving Motors, I feel that it is a substantial problem that it is necessary to explain again in what sense the passengers can be said to be controlling the operation, especially from the perspective of internalizing negative externalities by giving incentives to risk managers for risk management.

On the other hand, as I said earlier, if the problem of the effectiveness of the accident investigation is solved to some extent, if product liability, etc. is expanded, there is a possibility that the lawsuit will be prolonged compared to the case where the reimbursement system is strengthened, and I think that the problem may not be so big. Of course, the discussion on how to ensure the effectiveness of the accident investigation will affect the conclusion, but I think that the difficulty of (1) may not be as big as you pointed out.

With regard to the issue of information collection for the appropriate operation of civil liability and insurance, especially the issue of how to obtain investigation cooperation from foreign companies, there is a point of contention on how to make extraterritorial application effective. At the same time, there is also a point of contention on how to think about the scope of government access premised on DFFT. When a Japanese company causes an accident overseas, there are strong extraterritorial application provisions in the country or area, and if they are followed, government access will be granted. However, if Japan's extraterritorial application provisions are not effective, Japanese companies may be unilaterally taken data. It is necessary to advance these issues in a manner that can achieve a balance between Japan and overseas. From a global perspective, I am advocating the application of the Agreement on Deferment of Prosecution, which is a system to provide incentives for investigation cooperation while taking into consideration that only Japanese companies will not be disadvantaged in the aspect of investigation cooperation. I think it is better to consider developing such a system. Again, if the investigation system for accidents, etc. works well, I think that the difference will disappear to that extent for (1) and (3) in particular among the candidates mentioned here.

Finally, whether we are going to use manufacturing liability or automobile liability insurance in the future, I think the basic direction is to internalize negative externalities after setting minimum standards, so to speak, using the concept of defects and disabilities linked to safety performance standards. As I said earlier, it is necessary to update these concepts while being linked to the latest technology level, and to reduce the asymmetry of information. However, ultimately, in light of the speed of development and development of technology, and when it comes to competition in terms of safety itself, that is, to give incentives to improve safety, I think it is possible to consider the introduction of a system in which the degree of risk management's skill and the range of responsibility are proportional, or the introduction of a simpler strict liability linked to the occurrence of accidents, from the viewpoint of promoting the direction of (1). In that case, it will be easier to pursue responsibility, so I think there is a possibility that problems such as compensation for victims will be solved cleanly if we proceed in the direction of developing (1). That is all.

Mr. Kozuka: Thank you very much for your visit to . Then, Mr. Nakahara, who specializes in civil law, please come in.

Nakahara Member: I will also talk about the public compensation system, and I think that when considering such a public compensation system, its position in relation to the liability system is important. I believe that the point discussed by Mr. Yokota this time was the compensation fund system in the sense that it supplements the liability system when the liability system does not affirm the responsibility. I believe that ② of the three options presented by Mr. Goto in his presentation is intended to substitute the liability system itself. When supplementing the liability system, the premise is how to build the liability system itself.

In the following, I would like to comment on Mr. Goto's presentation. I think the biggest advantage of replacing the liability system is that the victim does not have to go through court proceedings, and by applying for compensation at an administrative window, he can receive compensation easily and quickly. Of course, even in the current system of the Self-Compensation Law, there are a compulsory liability insurance system and a system to make direct claims to insurance companies, so if the amount is within the actual amount of insurance money, the victim can receive compensation easily and quickly, but if it exceeds that amount, it will be a problem. In establishing a compensation fund system like option 2, I think there are generally various problems, such as how to set the requirements for compensation, how to set the level of compensation, especially if it exceeds that, whether to leave the way for compensation under the liability system, who will make decisions on whether or not the requirements for compensation are satisfied, who will manage the fund in the first place, who will contribute the source of compensation, and whether to establish a mechanism for claiming compensation from the person who caused the damage. In Professor Goto's presentation, it was assumed that it would be a set with immunity or the prevention of accidents by administrative regulation, but it is possible to consider the option of making it mutually complementary with liability systems or setting up a mechanism for claiming compensation for the person who caused the accident. Therefore, we understand that it is only an example. In any case, when establishing a public compensation fund system, it is necessary to build a specific system by referring to existing systems such as industrial accident compensation, pollution-related health damage compensation, Relief System for Sufferers from Adverse Drug Reactions, Crime Victim Benefit System, and vaccinations Health Damage Relief System, as well as various systems in foreign countries. For the time being, if we ignore the cost aspect, I think that a somewhat free system design will be possible. However, there are two important points, and one point is from the viewpoint of preventing accidents pointed out by Dr. Goto. It is conceivable that even if the liability system is abolished or reduced, the operator of the compensation fund will compensate the cause by claiming the cost, but it is easy to imagine that the problem is that it is difficult to ask the operator to have such expertise and workload. Another point is that, as Dr. Takahashi mentioned, it is necessary to explain why such a fund system is established only in automated driving vehicles. It is not simply because there are difficulties in pursuing liability, but there are countless such cases. In particular, considering AI, it is typically assumed that it is difficult to prove using technology using AI, so it is not a self-evident problem, and I think there are difficulties. That's all.

Mr. Kozuka: Thank you very much for your visit to . Then, Mr. Sato, please go ahead.

Member: Mr. Goto , I would like to thank you for your presentation. I have three or four questions. First of all, I believe that how to consider the premise of the system examination will be important in the future examination. In the discussion of the study group of the Ministry of Land, Infrastructure, Transport and Tourism, the discussion was based on the premise of the transitional period in which automated driving vehicles has gradually entered. Even if we look at the current situation, it is not necessarily the case that only automated driving vehicles is operated, and it is quite a long time before all cars and vehicles become automated driving vehicles. Therefore, until then, I thought that we should discuss with the extent to which it is necessary to coexist the old system centered on operator responsibility and the new system in addition to that in mind.

Dr. Goto presented three approaches. Regarding the perspective of (1) manufacturer's liability, I understand that the current system of design places great emphasis on prompt relief for victims. I believe that how to realize that within the system of (1) is an extremely difficult issue. I understand that some current automobile manufacturers do not necessarily have product liability insurance, and it is assumed that they will fight thoroughly. If that is the case, I believe that there is a concern that prompt relief for victims may not be realized even if the burden of proof, etc. is necessarily shifted.

Regarding (ii), which was also discussed in the Study Group, I believe that it was pointed out in the discussion of the Study Group of the Ministry of Land, Infrastructure, Transport and Tourism that how to collect insurance premiums fairly will be a problem because the safety and accident rate differ depending on the manufacturer in consideration of the manufacturer's contribution, and how to collect insurance premiums fairly and whether they can be collected reliably including overseas manufacturers.

Regarding whether or not it is good to continue to maintain the conventional approach of (iii), whether or not it is acceptable to be the responsibility of the operator in the first place, that is, whether or not the operator has control over the operation, as you pointed out, has been discussed in the report of the Study Group, and it is said that it is possible to sort out that there is control over the operation at least. For example, in the Level 4 unmanned automated driving service, which we are now considering, I think that a certain degree of arrangement has been made that the service provider has control over the operation, although I think there are shades of gray. In addition, regarding the point of ensuring the effectiveness of the reimbursement from the operator to the manufacturer, etc., which was mentioned as a problem in the conventional approach of (iii), I think that it may be possible to cover to some extent by discussing how to make the system for PL responsibility and ensuring the effectiveness, as described in (I). In other words, as described in (iii), we believe that it may be possible to overcome the problem by maintaining the operator responsibility itself and adding a certain modification in the PL responsibility. That is all.

Mr. Kozuka: Thank you very much for your visit to Next, Mr. Ochiai and Mr. Hatano are raising their hands. Since time has come, I would like to pause once after Mr. Ochiai and Mr. Hatano have made their statements. Mr. Ochiai, nice to meet you.

Ochiai Member: I would like to say a few points that I have noticed. In your presentation today, Dr. Goto once again pointed out the importance of sorting out the premises. In considering the legal system and scientific validity, what kind of premise is placed and what is required will be the premise of various discussions and arguments. While sorting out this point, I would like to reiterate that it is very important that a specific civil legal system also needs to be discussed. From that perspective, based on Dr. Goto's remarks, I would like to make one point. I believe that the self-driving car this time is not self-driving by itself. I believe there is something like the connected car that Dr. Inadani mentioned, but I believe that the data for operation is operated in cooperation with infrastructure such as the road side and traffic signals. What other infrastructure actors are responsible for such control, and what autonomous vehicles should do. Conversely, I think it is the same as the argument based on the premise, but if the theory of responsibility is not organized, even if the automated driving vehicles is operated, the manufacturer does not actually decide what to prepare, and if it is not decided, I think that preparing everything is an option. In that sense, I believe it is important to discuss the sharing of responsibilities in relation to infrastructure.

Second, I believe that you pointed out the point of digitization regarding the ideal way of developing advance regulation. I believe that it is quite often impossible to actually rewrite legal norms into numbers as they are. However, I believe that it is possible to write down what is written in the performance requirements, such as what is how many meters, as they are, but basically, when performing mechanical processing, the system side also has requirements entered in numerical values, and I believe that it is likely that the system will comply with them. I also believe that it is a basic matter as a social system related to autonomous driving to sort out the way of thinking about what is required when law evaluates in numerical values. In that sense, I believe that it is important to sort out what kind of response is required when digital processing is performed to the extent possible, whether it is possible or not, or whether it will be wide or narrow, including cases where performance is required by approval standards, etc. for automobile vehicles, and rules related to infrastructure development and operation.

The third point is that there are several options arranged by Dr. Goto, but first of all, I think Dr. Goto pointed out the difference between Dr. Fujita's preliminary and general approach and what machines and people can do in it. I think there was also a point about the relationship between the theories of responsibility. I think it is very significant that you deepened your analysis of responses to malfunctions and functional failures in automated driving vehicles, and I think that resolving this point will be a very important perspective as we consider and develop guidelines.

In addition to that point, I received three proposals, but Proposal 3 is a system and option that is currently being taken, but on the other hand, I think that there is an actual unnaturalness for passengers as Mr. Takahashi said. Legally and canonically, it may be possible to read that we are controlling the operation, but there are some parts that are not automatic driving, but when we consider the responsibility components in the case of using other AI, there is a concern that if we recognize the concept that has been expanded too much, especially in automatic driving, which is important in the legal system for AI, it may have adverse effects on other parts. In that sense, I think that the discussion of 3) should be revised. Regarding 2), I think that it is necessary to consider how to support internal performance compensation activities for foreign countries, in addition to the point of how much compensation can be provided under the legal relief system by insurance.

In that case, I think the one proposed in Proposal 1 is good. On the other hand, not only is the operator's liability made to be the liability for negligence, but also there is a separate discussion on product liability, so I think it is important to consider the realization of damage compensation and the way to secure financial resources as discussed today, and to be able to combine them.

Mr. Kozuka: Thank you very much for your visit to . Now that we have talked about the premise, could Mr. Hatano make a statement? I would like to conclude with that.

Hatano Member: . This is Hatano of the Japan Automobile Manufacturers Association. Regarding the several approaches presented by Dr. Goto, while I can understand them as hypothetical questions and answers, what I would like to confirm a little is that I had the impression that it was a little unclear which issue was mainly discussed, whether it was an argument about a so-called operator who has liability for damage even in the absence of negligence for prompt victim relief in the first place, or an argument about the certification of negligence to clarify the liability for warranty against defects of products. If it is the former, I think that the level 3 automated driving has already been implementation in society, and under the Road Traffic Law, it is arranged that the driver uses the automatic traveling device, and under civil law, it is arranged that this driver continues to be the operator. Due to the revision and enforcement of the Road Traffic Law in 2023, although the driver is absent at Level 4, it is possible to understand that the specified automatic operation implementer is not the driver, but is an alternative to the driver, considering that it is specified that the specified automatic operation implementer operates the vehicle after taking measures to comply with the Road Traffic Law on behalf of the driver. In that case, considering that the control and benefit of operation continue to be assumed by the specified operator in lieu of the driver, the concept of the operator does not seem to change significantly even if the level of automated driving changes. Given this point, I feel that the story in which the manufacturing side, which is not defined as an operator at all, is involved in this discussion is somewhat strange.

On the other hand, from the perspective of proving defects and malfunctions of products, which is called certification of negligence, it is additionally required to record operation during automated driving, which is imposed on Level 3 by the revision of the law in 2020, and it is actually not difficult for owners and users to request disclosure of this data. Assuming that this will be treated in the same manner even at Level 4, I think that there is actually no such difference in understanding that there is a significant difference in the investigation of errors or defects due to the overwhelming difference between conventional products and automated driving. Rather, considering that automated driving is legally guaranteed a mechanism to record data and provide it to owners, I think it is possible to say that automated driving is making progress in investigating the cause. Therefore, I feel that it is better to discuss what is necessary for the manufacturer after sorting out the relationship between the discussion of defects and negligence and the story of the operator who is controlling the operation and receiving benefits. That is all.

Mr. Kozuka: Thank you very much for your visit to

Basically, I think that among the functions of civil liability that Mr. Goto said, the first, that is, the change of behavior by whom and what kind of behavior will be changed by the existence of the liability system, is significant. Basically, it is desirable to have a system in which there is a strong incentive to ask the manufacturer to improve and respond in the end to prevent accidents, regardless of whether it is fault or accountability. At the same time, I think it was desirable to have a person who can be involved in reducing accidents in some way, even if he or she is not the person behind the wheel, and to have that person change his or her behavior. In terms of what will happen in the future, in the case of the provision of services and automated taxis that Professor Goto mentioned, even if we can say the same thing as the operator, when the owner car is Level 5 and the person who is riding in the car does not know what to do just because he owns the car, there seems to be a problem of whether it is meaningful to place responsibility on him. On the other hand, in relation to the relief of victims, Japan's accident system up to now is not just civil liability, but insurance companies have played an extremely large role, and know-how and technology for victim relief have been accumulated there. I believe that there may be an issue of what to do about it in the future. To put it more broadly, I think it is a problem of which system and which choice are desirable for society as a whole, considering that behavioral changes such as reducing the incentive to purchase a private car and an owner car when it becomes a automated driving vehicles can actually occur or may occur by selecting a product liability concept, a vehicle operator concept, or a fund concept. If I were to draw attention to the question raised by Mr. Hatano, I heard that Mr. Goto raised a big question about how these systems create and change the premise rather than what kind of premise is used for discussion.

I would still like to discuss it, but I would also like to discuss criminal law later today, so I will stop here for now.

Regarding agenda 2 - (3), Mr. Imai has prepared the content of "Outline of Judicial Precedents on Negligent Offenses - Focusing on Automobile Accident Cases." Then, Mr. Imai, thank you for your cooperation.

Member Imai:
(The following explanation is based on "Material 5: Summary of Judicial Precedents on Negligent Offenses - Focusing on Automobile Accident Cases")

Thank you very much. I will share it with you in Word format and explain it. The task assigned to me was not to talk about theories, but to provide an overview of what judgments are being made about negligent offenders in the courts, so I created this memo.

First of all, in relation to accidents involving conventional automobiles, that is, automobiles that are not automated driving vehicles, what kind of articles are problematic as negligent offenders are roughly divided into the two written here. Article 5 of the automated driving Act on the Punishment of Acts Resulting in Death or Injury, which is the crime of negligent driving resulting in death or injury, is often seen in newspapers. This article and Article 211 of the Penal Code, which is the crime of causing injury through negligence in the pursuit of social activities, are included. It can be interpreted that the person responsible for the crime of negligent driving resulting in death or injury is limited. It is written that the person is negligent in taking the necessary precautions when driving a car, so it is natural to understand that the person responsible is the driver of the car. However, there is a problem as pointed out by Mr. Hatano earlier. Mr. Hatano pointed out that it is very important to know whether or not the person in charge of the specified automatic operation takes the place of the driver. Unfortunately, there is still little interest in this point in the Society of Criminal Law, and it is not known who the driver is. Conversely, Mr. Hatano's opinion is that saying driverless driving is itself a false indication, and it is probably a global trend in the future that the SAE standard itself is strange. However, it is beyond the scope of this discussion, so I will not cover it today. In that case, for example, I think the current concerns of manufacturers are that even if the crime of negligent driving resulting in death or injury is not applied, the crime of professional negligence resulting in death or injury under Article 211 of the Penal Code may be applied. Therefore, regarding the establishment of this crime, the phrase "failure to exercise necessary care" in the second bullet point immediately below means so-called negligence, but next to that, "due to" means that there is a causal relationship, and if an illegal result occurs that "a person is killed or injured," the success or failure of the criminal negligence will be contested. Furthermore, as a premise, what is generally said in Japan as a requirement for the establishment of a crime is to discuss it in three stages: there is an act that falls under the constituent requirements, and it is illegal and culpable. Here, in particular, teachers other than those in the Faculty of Law may have questions about what the constituent elements are, but they are colorless and have a certain meaning. To be colorless is to say that a constituent element is nothing more than what constitutes an act, which is a translation of the German word Tatbestand, and it is almost a colorless concept. It is sometimes used in civil law, but in criminal law, it is supposed to confirm whether or not it is a constituent element in order to determine the outer frame of the scope of punishment. The important point is the following illegality, which is the infringement of legal interests or its risk, and the liability of what kind of psychological state it was. I would like to proceed with this subject. Please take the next page.

It is the meaning of the second fault. I will return to the judicial precedents later, but they only show judgments in individual specific cases and are significant in that respect. Therefore, I do not think that the judicial precedents show a certain position toward negligent offenders. We can say the trend of the case law in general, but it is difficult to judge which specific theory is relied on. However, I will try to explain from what point of view the understanding can be divided. There is a point that Dr. Goto is able to understand. First of all, I would like to talk about the basic perspective that is shared. What criminal law is for is not described here, but it can be said that it should serve as a backup function for the tort system, although it is historically opposite. Criminal law does not teach a person to behave in a certain way, but exists to protect the interests of the law. I am thinking in the same way as the chairman and Mr. Goto, and I am considering the criminal law system from the perspective of what is necessary to deter the infringement of legal interests. What I emphasize in doing so is that when I think that my legal interests will be violated or will be in danger of being violated in the near future due to my action, I am required to have another motivation to stop taking this action, and if I do not have another motivation, I will be evaluated as a willful act. On the other hand, if you regret that you would not have hit a person or killed a person if you had stepped on the brake there, but if you could not have foreseen the future infringement of legal benefits at the time of the act, or if you could have foreseen it from the perspective of the court after the fact, you can be evaluated to have been responsible, although it is one step lower than intentional, because if you had been a little more careful, you could have foreseen it and made a different choice of action by using what you foreseen as motivation. This is the explanation when the liability element of negligence is accepted. Please take the next page.

The idea I just mentioned is the so-called old faultism. In other words, it is the position that whether it is intentional or negligent, when a car runs at a speed of 40 kilometers per hour, for example, and hits a pedestrian and kills him or her, whether it was intentional or careless, there is no difference in the fact that the person died at a specific date, time, place, and in a specific manner, so whether it was intentional or negligent has nothing to do with illegality and is merely a matter of liability. This is the understanding of old faultism. In addition to this, in criminal jurisprudence, there is a new theory of negligence, which takes the idea that the degree of badness is different between the case where the person hits the person without stepping on the brake, but at a legal speed of 40 kilometers, and the case where the person hits the person while driving at 40 kilometers without stepping on the brake inadvertently, and the person's feelings have meaning in the evaluation of illegality. According to that idea, if you violate the Road Traffic Act and drive in violation of the legal speed limit, it will be explained that the illegality is affirmed just by that. I think the explanation is difficult to understand for people who heard it for the first time. Furthermore, in the 1965 s and' 50s, around the time of rapid economic growth, when pollution incidents occurred, the following new negligence theory, the fear theory, appeared. This is not an old story but a new one. As you all know, the issue of violation of the Food Sanitation Law is still being fought. It may be a little scary, but if you use it thinking that everyone is using it, and there is a story that kidney dysfunction has occurred, what should be considered to be negligence is that there is a company that allows you to buy a large amount of goods at a lower price than other companies, and you are worried if it is okay to buy from them, but if you sell goods thinking that if you can make them quickly and sell them at a low price, you will increase your profit, you have been too careless about unknown risks, and if you have thought of a vague risk that may occur in the future, you should have dealt with it. This view is called Xin Xin Negligence or Uneasiness Theory. This understanding was adopted by the Tokushima District Court in the Morinaga Arsenical Milk case, but other court cases do not follow this understanding. I think the reality is that it is rejected in criminal trials, where careful judgment is required, to go this far and admit negligence widely. In detail, some recent theories reevaluate the Xin Xin Negligence theory, but I will omit that story because it is different from today's discussion Issue. At present, many theories are based on the modified old theory of negligence described below. In short, since the latter half of the 1975 s, there has been a theory that reminds us that we can only recognize negligence in a manner that conforms to intention by carefully considering the specific predictability of the occurrence of the result on a case-by-case basis. I think it is a common theory up to the present. Please take the next page.

As for points to note, in all theories, what is important is that, for example, in the case of an automobile accident, on the premise that there is an illegal result that a person was hit and killed by a delay in stepping on the brake and that the legal interest of human life was violated, whether or not there was a causal relationship between the result and the inaction that the brake operation was delayed is recognized, and if there was a causal relationship, whether or not it could have been foreseen is judged, and if it could have been foreseen, the brake should have been stepped on a little earlier. In this way, predictability is considered first, and there is an obligation to foresee a foreseeable result, and further, as long as there is a possibility of avoiding the result, an obligation to avoid the result is also imposed. When there is a violation of any of the obligations, that is, a violation of the obligation to foresee or the obligation to avoid the result, negligence is affirmed. I would like to repeat this, but I believe that you can understand it if you consider it in the steps of tracing back from the result, confirming the illegality, determining the causal relationship, and finally judging the feelings of people.

Now, what are the judicial precedents? I will omit the general review because it is almost the same as the points to be noted that I just mentioned, but to supplement it a little, as I said at the beginning, in the judicial precedents, the presence or absence of negligence is merely a judgment in individual and specific cases. The judgment is not indicated by clearly setting general judgment standards. In that case, I think it would be desirable for the compilation of judicial precedents to be a law, but at present, I think it is quite difficult to write the definition of negligence as the greatest common divisor of judicial precedents in the law. This is still debated even in places where the concept of diligence is determined from case law, such as in the United Kingdom and the United States. In addition, the concept of the CCD (Competent and Careful Driver) standard proposed by Member Hatano is also widely interpreted and has not been decided. I think the discussion of negligence is similar to that. What I will talk about below lacks uniformity, but I will introduce that the judicial precedents are like this. Please turn to the next page.

To re-summarize the thinking process of the case, whether a car hit a person and resulted in injury or death, and whether the result could have been avoided, I have not talked much about this this time, but for example, when a victim who wants to commit suicide runs out and runs over a person at the legal speed, it is not in time to brake just before. In such a case, since there is no possibility of avoiding the result, it should not be discussed in the criminal law in the first place, and the theory is to deny the causal relationship or to deny negligence, but in the current theory, many people deny the causal relationship. However, the case law does not make that clear, so today, I will talk about the relationship between causality and negligence regarding the possibility of avoiding consequences. If the causality is affirmed, all the objective aspects of the crime are satisfied because objectively there is a causal relationship between the inaction and the result that one's braking was delayed. The rest is the subjective aspect of the crime, and if it is not intentional, it is less than intentional negligence. Then, what is negligence at that time? If you could predict the result, please predict the result. If you could predict the result, could you take evasive action such as operating the brake quickly, and if you could, shouldn't you have done it? Using these steps, negligence is finally recognized. At that time, as I will say in Case No. 2 later, regarding the meaning of concreteness in the specific predictability of the occurrence of the result, the meaning of concreteness is not clarified in the theory. Since the judgment is made based on the evidence that appears in criminal court cases, there is a limit to the evidence that can be raised after the accident, for example, by the police investigation, so there is a limit to the scope that can be revealed. As long as there was as much specific anticipation of the outcome as possible, I think that both the judicial precedents and the prosecution practice are trying to realize predictability. However, there are some judicial precedents, such as Precedent 2, which I will see later, are quite rough, and people who see them for the first time wonder if this is OK. Please take the next page.

Then there's the word "principle of trust." You may hear that people in science also talk to lawyers, but especially in criminal law, it is a word that you often hear. It was imported from the German Penal Code, and it is a concept or idea that spread in Japan in the 1965 s but is now on the wane. The principle of trust is, for example, when you are driving at the legal speed and following the traffic lights, as in the case law mentioned later, when you are driving recklessly and ignore the traffic lights or do not follow the right turn method, so to speak, when you are in an accident and the victim dies, you are doing all the actions you should be acting as you are, and you did not expect the other party to do anything strange, and it is unfortunate to punish them, so the principle of trust comes out, and when it applies, negligence is denied. If you trust the actions of other people involved in traffic, there are three requirements for negligence. One is that the specific predictability of the occurrence of the result is denied, the other is that the possibility of avoidance of the result seems to be denied, and the other is that the possibility of both is probably denied. As mentioned in the judicial precedents mentioned today, in a relatively large number of cases, when specific predictability is denied based on the principle of trust, there is no need to question the possibility of avoidance of the consequences. It should be noted that since the "principle of something" is an extrinsic argument, it cannot be generally or abstractly accepted, and when looking at the case, there is no choice but to examine whether there was concrete predictability. I would like to emphasize this point once again.

I would like to look at precedents. There is the Hokkaido University Electrocautery Surgery case on predictability. This is an incident in which a new physician was treating a head nurse with an electrocautery knife with veteran patients and nurses, and the positive and negative of the electrocautery knife were applied in reverse. As a result, the head nurse's leg was seriously burned and he had to have his leg amputated. At that time, the head nurse was a veteran, so he was found guilty of professional negligence resulting in injury. However, it is said that the new physician was following the order of the patient and did not expect serious injury, and his negligence was denied. The conclusion is quite critical. However, even if he was a physician at that time, a person who was still inexperienced would not know how much specific checking should have been done to prevent the replacement of the positive and negative. Therefore, it is still worth being a precedent as an example in which there was no specific predictability of the occurrence of the result. In short, it can be seen that the precedent does not take the fear theory here.

Next, regarding the second case, this is a slightly different case. The defendant was driving a truck, and there was a passenger. When he thought that it would be too heavy to move forward even if he stepped on the accelerator today, the defendant was also running away, so he turned the steering wheel to the left to avoid contact with an oncoming car and crashed into a utility pole. The passenger was also injured, but suddenly a person fell from the truck and died. The defendant did not think that there were two people behind his car, so he argued that it was, so to speak, a gift accident and he was not responsible, but the Supreme Court did not admit it. This is a rather strange judgment, and it is criticized in theory. If you were more careful, you might have found out that there was a person behind him in the rearview mirror. If you were more careful, you might have found out that there was a person behind him in the rearview mirror. If you were to drive recklessly, you should be careful not to see the person, but to be aware that you might injure a person who might be jumping in. This is a terrible logic, but it is a case that shows that if you take the abstract legal code theory, which says that even if you kill someone who you did not see, there is intention, negligence will also spread.

In addition, the third case is characterized by the fact that it was not a case prosecuted by the public prosecutor. Since the case was prosecuted based on the decision to prosecute by the Committee for Inquest of Prosecution after the examination, it can be said that the case was originally considered to be impossible to prove negligence by the public prosecutor, but the Supreme Court also admits that idea. Since the accident was extremely tragic, it is strongly hoped that the results will be resolved by a method other than the application of the Penal Code, but I believe that this will also be the case from the perspective of a criminal law researcher. It may be necessary to install ATS in many places, but we do not know when and where such an accident will occur. It is impossible for the management in the headquarters building of JR West to predict where an accident is likely to occur. If we are talking about specific predictability, I believe that the Supreme Court also recognized that it is impossible to punish the person for negligence in this case.

I would like to move on to case No. 4. From here on, I will list court cases that also mention the possibility of avoiding the consequences relatively, but I would like to note that, for example, case Nos. 4 and 6 also raise the issue of the possibility of avoiding the consequences first and deny it, although they do not state this. Case No. 4 is a bit old, the Osaka High Court in 1967, and it is an idea that is close to the principle of trust, but it is not that you should not drive unless you eliminate the possibility of all kinds of accidents, but it is that you can deny negligence under the Criminal Code if you predict what kinds of accidents are possible on the road and deal with them. I think it should be a judgment like this, probably based on common sense.

After No. 5, there are various cases of accident types. In the case of No. 5, when the defendant was in the car and was about to turn right, the victim's car came running at high speed from behind on the right. It is said that there are usually no such people, and as written in the second half of the third line of this judgment, unless there are special circumstances, it is held that it is sufficient to drive by trusting that another vehicle coming from the right side will observe traffic laws and take appropriate actions to avoid a collision with one's own car. This still has precedent value as the Supreme Court de facto recognized the principle of trust in 1966. Reading out this judgment denies the predictability of the result, but as I said at the beginning, the court, in order to make a more certain judgment, also determines that the result was not avoidable, as in this case, so I would like you to read both of them as such.

No. 6 is also a case from 1967. Around this time, motorization occurred in Japan, and in order to avoid becoming a 100 million citizen ex-con, for example, a traffic fine system was established. At the same time, in the interpretation of the Penal Code, such a principle of trust was included and negligence was denied. This time, it is a similar story. This time, it is a story about a motorcycle, but when the defendant was about to make a right turn without making a so-called two step right turn, the victim's vehicle came running behind him on the right, and the accident occurred. As written in this judgment, the defendant violated the Road Traffic Act at the time, but it is also significant in that the violation of the Road Traffic Act does not immediately affirm his negligence under the Penal Code. At this point, it is highly likely that the Supreme Court has not adopted the new theory of negligence. In the form of the new theory of negligence, it is easy to reach a judgment that it is difficult to deny the responsibility for negligence under the Penal Code because he is behaving badly if he is violating the Road Traffic Act, but the Supreme Court has not adopted it.

An intermediate summary of the principles of trust is written below, as I mentioned earlier. In each case, various road conditions, and even on the same road, the slip situation varies depending on the weather and the date and time. The principles of trust are that if you could predict the results to what extent in the specific road conditions of each case, and if you thought that other vehicles would not fly at that time, you do not need to be any more careful. It is briefly mentioned at the end that how this affects automated driving vehicles, but ADS in automated driving vehicles probably acts according to the rule. If it is set at the legal speed, you cannot run on the flow of traffic, so you are running at the legal speed, and if you do so, you will be flapped and driven into it, or if you accelerate to get on the flow, you will be caught by a police car instantaneously. Based on Case 6, the principles of trust seem to apply to drivers of cars running at actual speeds different from the legal speed, but whether it will be used for automated driving remains a major problem. To solve this problem, I personally think that it is better to avoid the occurrence of the difficult problem of the principles of automated driving vehicles and trust by physically dividing the ODD setting area and driving only automated driving vehicles in it, but I think this will be a future discussion here.

The cases after No. 7 will become more detailed in the Heisei, so I would like everyone to read them. No. 7 is a case in which the victim vehicle was found at the last minute, so it is easy to deny the possibility of avoidance as a result. I think that the causal relationship could have been denied if the defense counsel argued, but the negligence was denied.

No. 8 is an incident in which a taxi driver ignored the red light and hit a car that ran into him under the influence of alcohol at 70 kilometers, ignoring the yellow flashing light ahead and the red light on the intersecting road from the left, and the taxi passenger was killed. The Supreme Court said that as a taxi driver, you do not need to think that there is a person who ignores the red light on the intersecting road and runs into you recklessly. I think this also uses the so-called principled idea of trust in order to deny predictability. However, it also introduces the possibility of evasion.

In the case of No. 9, the traffic light in the direction of his travel changed from green to yellow, and the defendant began to turn right before it turned red. When he did so, the oncoming vehicle crashed into him because it was still green due to the staggered traffic light, but when the defendant started to turn right without knowing it, he collided with the oncoming vehicle and the other party died. At that time, the Supreme Court also recognized his negligence, saying that it was imprudent to think that the other party would turn red just because his own traffic light turned red, and that he should not be trusted.

Finally, I would like to point out a point that seems to be a Issue that should be examined here in the future based on conventional cases of criminal negligence. In relation to the predictability of the result, it is necessary to consider the fact that the content of the predictability of the result differs depending on the situation, such as what kind of car models are running in what kind of physical situation, that is, whether it is only in automated driving vehicles, whether it is a conventional car, or whether it is a place where speeding and traffic law violations are common. As I said at the beginning, the criteria for determining the predictability have not yet been indicated in the Penal Code. However, in making a judgment on a case-by-case basis, it is important to set such road conditions. The possibility of avoidance of the result is a little easier to cut, and it is already too late when a person jumps out at the last minute, or when, for example, another automatic driving vehicle jumps out at the last minute due to a defect in the ADS. Therefore, the possibility of avoidance of the result will be eliminated, and the causal relationship will be eliminated, or the element of negligence will be eliminated, and the defendant will be acquitted. In addition, although it is not the title of today, as I said at the beginning and as Mr. Hatano said, I think that the point whether the specified automatic driving implementer is the driver should be carefully discussed. That is all.

Mr. Kozuka: Thank you very much for your visit to , thank you very much for your very detailed explanation. Since time is pressing a little, we would like to proceed as follows. We would like to discuss this until around 15:00, and after that, we would like to allow for an extension of about 15 minutes due to the materials that Mr. Takahashi has submitted.

I would like to express my opinion freely, but there is something that bothers me, and I would like to express it first. At the Sub-Working Group, we have already seen a number of graphs in the four quadrants of the graph from the Japan Automobile Manufacturers Association (JAMA), directly from the Hatano Group. Toward the end of the third session of the Sub-Working Group, some members pointed out how it would be applied in specific terms, which was a matter of concern. As the Chief Inspector, I had a look at Mr. Imai's materials in advance, so I thought it would be okay to make some revisions based on that, so I asked the Secretariat to write the drawing. It is the right side that you are projecting now. First of all, at least when considering criminal liability, although negligence may be categorized to some extent, it seems to be concrete predictability and not an abstract sense of fear. Therefore, reasonable anticipation is not a legal term, and I think it is better not to be so. We can simply consider whether or not there is specific predictability. It is not my intention to demand more than the current precedent in terms of specific predictions. If there is no specific predictability, it may be effective for the absence of negligence or the possibility of evasion as a result, but in any case, criminal liability cannot be established, and therefore, the lower part of the horizontal line that separates the presence or absence of specific predictability may be considered to be criminal liability cannot be established. In that case, in consideration of the part above the horizontal line, if it is avoidable and unavoidable, as indicated by the Japan Automobile Manufacturers Association, there is room for criminal liability to be established in the event of an incident even in automated driving vehicles. There may be a question of whose criminal responsibility it is, but I think there is a possibility that a subject who had specific predictability will be criminally responsible. If it is unavoidable, there is no possibility of evasion as a result, and apart from whether it is an element of negligence or an element of causality, I do not think criminal liability can be established here either. As long as there is specific predictability, incentives based on the Criminal Law do not work well. Therefore, although accidents may occur, we thought that it would be necessary to conduct technical development to reduce damage as much as possible and to follow up on areas beyond the reach of the law. Above all, I believe that it is important to specify the specific predictability of where this horizontal line is. This is my opinion, so I would like to hear your frank opinions.

I would like to ask the teachers who are raising their hands to speak, but at the same time, Mr. Hatano also has a question, so could you please raise your hands? Then, Mr. Takahashi, Mr. Inadani, Mr. Fujita, Mr. Hatano, Mr. Ochiai, in that order, please. First, Mr. Takahashi, please.

Takahashi Member: . I was also one of the criminal defense lawyers in the case of the last ruling, the time difference signal on July 13, 2004. He was found not guilty in the first trial, but he was found guilty in the second and third trials. In addition to the Supreme Court precedents mentioned by Mr. Imai, in the Kintetsu Ikoma Tunnel Fire Case on December 20, 2000, the Akashi Okura Coastline Sand Beach Cave-in Case on July 22, 2014, and the conviction of the first instance in the Karuizawa Ski Bus Fall Case, which I was in charge of, all of these cases, the violation of the duty to avoid consequences and predictability have been linked to each other. In other words, we adopt a balanced approach in which if there is a serious violation of the obligation to avoid consequences, the predictability is widely recognized and the conviction is tried to be made as much as possible, and if there is a violation of the obligation to avoid consequences, but the degree of the violation is slight, the predictability is severe and the innocence is tried to be made. In that case, as the Chair mentioned earlier, assuming that there will be no criminal immunity, if there is no predictability, or conversely, if there is, when we try to specifically consider what kind of case it is in the case of automated driving, I think the end result will be whether or not the duty of avoidance has been fulfilled, and to what extent it has been fulfilled or not. I believe that the Study Group should consider these matters. When programming is made, if operation in accordance with the Road Traffic Law is not properly secured and an accident occurs, it will be judged that the person has not fulfilled the duty to avoid as a result, and by extension, it will be judged to be predictable and guilty. The question is, are there any exceptions? In other words, we did not make programming in accordance with the Road Traffic Law, and in that regard, we certainly did not fulfill our duty to avoid consequences, and we are criticized, but even so, I think it is necessary to consider the point that there may be cases where it is said that there is no predictability or possibility of avoiding consequences due to the peculiarities of automated driving. On the contrary, if an accident occurs despite the fact that the operation in accordance with the Road Traffic Act has been properly secured in terms of programming, this is in principle the fulfillment of the obligation to avoid the consequences, but even so, I think it is necessary to consider whether there are cases in which it is possible to predict the accident and avoid the consequences. In other words, although we worked hard to create programming in accordance with the Road Traffic Law, as long as the incident still occurred, I think it should be examined that there may have been predictability and possibility of avoiding the consequences from the viewpoint of the peculiarities of automated driving vehicles. Since the Supreme Court does not necessarily regard the Road Traffic Law as a violation of the duty of care in business or driving, I think it is necessary to consider such exceptions. In addition, later on, I would like to talk about the most problematic point in traffic law due to the duty to avoid consequences. That's all.

Mr. Kozuka: Thank you very much for your visit to . Mr. Inadani, please.

Member Inadani: Thank you very much for your comments. Rather than the specific examples I received today, I understand that the question of how to grasp the relationship between the current laws and regulations and the program that governs the behavior of automated driving, which is related to Mr. Fujita's talk earlier, will be a very big issue in the end, and I understand that this is one of the issues raised. Therefore, from such a perspective, I would like to make a few remarks in the form of a slightly larger discussion. Thank you very much for your comments. Mr. , thank you very much for your wonderful report, which is very humorous and rich in meaning. Although the reason is different from Dr. Takahashi's, I also think that it is better to think mainly about the violation of the duty to avoid consequences. In your report, Mr. Imai probably mentioned the perspective of understanding the Penal Code from a functional perspective. In other words, in relation to the setting of the issue this time, as the chairman talked about, I think the focus of the discussion is what form of sanctions should be used based on the awareness of the issue of what obligations are most desirable to be performed by automobile manufacturers and, in some cases, operators from the perspective of optimizing risk management. I believe that this issue is also raised at the end of Professor Imai's materials this time. Regarding quantitative safety assurance, how to consider the probability of a certain accident occurring within a certain ODD is a problem, and I understand that Professor Imai has organized this issue in terms of the relationship with predictability under the Penal Code. As Mr. Imai said, there is a considerable variety of predictability when looking at precedents, court cases, and theories, and I think it is quite difficult to draw a line between what cases can be said to have had specific predictability. On the other hand, if we consider the duty to avoid consequences in the sense of reducing the probability of the occurrence of consequences to a certain level or more, I think it is easy to discuss it in connection with the story of the civil liability side that has been discussed so far, that is, the idea of design in a form that links the safety performance standard with defects and obstacles, and updating the standard in a timely manner. In that case, I think it would be better to consider liability for negligence in connection with the duty to avoid as a result of reducing the probability of an accident to a certain level or less. The idea of permitted danger that I mentioned in the beginning is also based on this idea, but in any case, I think it is better to consider the violation of the obligation to avoid consequences than to cut off the possibility of prediction. In particular, if deep learning is included, it will be unstable whether or not it could have been specifically predicted depending on which case law on predictability is taken seriously, and I think that this itself is not a problem. Therefore, given that software behaves stochastically, I thought that it would be good to consider a mechanism to quantitatively set a standard at the point of consequence avoidance violation.

In addition, in Mr. Takahashi's comment, you mentioned the relationship with the Road Traffic Law, but I would like to comment separately on how to realize the automated driving and Road Traffic Law or the Penal Code after Mr. Takahashi's report.

Finally, in considering the duty to avoid consequences, as Dr. Ochiai and I have mentioned in relation to the previous report and the present story, I think that the relationship with infrastructure and the relationship with the actions of other actors, which are closely related to infrastructure, must be considered as a set. Based on these considerations, I think that it is good to be able to organize the remaining issues, particularly how to do with passing. That is all.

Mr. Kozuka: Thank you very much for your visit to . Then, Mr. Fujita, please go ahead.

Member: , thank you for your detailed report. It will be a similar statement to Dr. Takahashi and Dr. Inadani in conclusion, but it will be a statement that the criminal liability of automated driving will be focused on the possibility of avoiding the consequences. I am not specialized in criminal law, so it may be a matter of course for experts, but I would like to make some comments. In relation to the crime of negligent driving causing death or injury, even if we take the conventional framework for automobile accidents in which the presence or absence of negligence is considered based on the possibility of avoiding the consequences and the predictability that is the premise, even if the framework itself is the same, I think it is necessary to carefully consider what differences and characteristics will emerge from the case of normal human driving. In the case of human driving, control of a moving vehicle, in particular, actions immediately before an accident becomes a problem, so judgment is made mainly on the predictability and the possibility of avoiding the consequences at that time. On the other hand, the programming of the AI of the automated driving equipment is judged at the time of design, so the content, roles, and positioning of predictability and result avoidance are likely to change. As a result, Dr. Takahashi and Dr. Inadani's remarks led to the emphasis on the possibility of avoidance. If the events subject to predictability, which are problematic at the time of AI design, mean various situations that may occur during design, which should be assumed at the time of program automated driving, there are various things in the content. In the case of human driving, human cognitive abilities are extremely limited, so the objects that can be predicted just before an accident are also limited and specific. However, at the design stage, it is possible to consider and assume over time, so the scenarios subject to predictability are extremely wide. As a result, even if liability is denied on the grounds that there was no negligence, the number of cases in which such accident scenarios are considered to have been completely unforeseeable will decrease significantly, and as a result, the number of cases in which liability is denied on the grounds of possibility of avoidance or other factors will increase. A good illustration of this is the example of the sudden jump in the four quadrants of today. I don't think there are many people who think that there is criminal responsibility for development and manufacturing a system that cannot prevent 100% of this, but the problem is why. In the case of a sudden rush, it may be possible to explain that there is no possibility of avoidance in the case of a car driven by a human, or it may be possible to say that such a rush could not have been foreseen in the first place. However, since the possibility of a sudden jump itself can be easily assumed as a situation at the programming stage, it is impossible to say that there is no predictability about the possibility of such a sudden jump. If that is the case, the reason why there is no liability is that there is no possibility of evasion as a result in the sense that it is impossible to create a program that can respond to such a runaway, and that it is a value judgment that it is okay to run it even in a automated driving vehicles that cannot respond to such a runaway. If I were to put it another way, I would say that it is permissible to make a design as if such a rush did not occur, or in other words, that there is no responsibility based on the "principles of trust in design." This can be said to be the principle of trust as a concept that defines the limit and range of the possibility of evading the result. The fact that it is not possible to 100% respond to runaway is relatively objectively determined mainly from a technical point of view, but whether it is OK to drive such a automated driving vehicles or not to respond to it at the time of programming is a comprehensive judgment that includes what kind of action people should take in a car society, so it is a highly normative value judgment. These are all issues that should be discussed in a completely different dimension from predictability, and I think that most of the errors that become problems when programming automated driving vehicles are our problems. In that sense, the conclusion is very similar to what Dr. Takahashi and Dr. Inadani said.

In light of the above, it will be necessary to clarify two specific points in the future. One is the limit of the possibility of avoiding the consequences due to the limit of program creation, and the other is the range of automated driving vehicles operations that cannot be handled by the program within a certain range. As mentioned earlier, it is relatively easy to say that the person who jumped out is bad, such as when the person suddenly jumped out in front of the car. However, in the case where the victim is not bad, but it is impossible to create a program to prevent accidents in advance, I think that there will be a hard case when it comes to the question of whether automated driving vehicles can accept it as an unavoidable social risk. I think that we will have to consider this matter by citing specific examples individually and create a basic market view.

Finally, I should comment after hearing Dr. Takahashi's report, but I am not sure if I will be able to attend until the end due to time constraints, so I think it is true that the program must comply with the obligations under the Road Traffic Act. However, I think it is possible to discuss this in relation to negligence, the possibility of avoiding consequences, and predictability, or to position it as a separate dimension of programming constraints, and to validation the possibility of avoiding consequences for cases that cannot be covered by it on the premise that it is naturally observed. I think it is necessary to consider which one to choose. In addition, the provisions of the Road Traffic Act when I say "observe the Road Traffic Act" are not abstract or general obligations such as the safe driving obligation in Article 70, but are limited to those that clearly show more specific and individual behavioral norms. Regarding such specific behavioral norms, I would like to add the pros and cons of sorting them out as constraints on program creation separately from the risk of causing accidents. That is all.

Mr. Kozuka: Thank you very much for your visit to . I would like to ask Mr. Imai what you think about the fact that it is impossible to predict the situation of specific accidents at the time of programming. In particular, I would like to ask you later what will happen in relation to the 1989 judgment of the person who was riding on the carrier that you talked about today. I have heard that Mr. Hatano has a question about the crime of dangerous driving resulting in death or injury. In addition, if you have any comments on my proposal to revise the materials of the Japan Automobile Manufacturers Association, please let me know.

Hatano Member: Thank you for your nomination. Regarding the four quadrants presented today, although they are slightly different in meaning, I felt that they would be rearranged in consideration of liability under the Penal Code, and the industry would generally agree with them. The left side is a case of predictable accidents, and the vertical axis is based on the image of the number of cases. Since the revised material is simply a classification, the quantity does not matter, and probably the uniform area on the four quadrants has an impression that it is closer to this meaning, but I think that the content has been changed to make it more understandable.

On the other hand, in the presentation from the Japan Automobile Manufacturers Association at the 3rd Sub-Working Group, I mentioned that the yellow area in the figure is a question about whether the crime of negligent driving causing death or injury or the crime of dangerous driving causing death or injury under the Act on Punishment of Acts Related to Motor Vehicle Driving applies, considering that in the first place, a person is not involved while driving in automated driving under the Penal Code. Last time, the Ministry of Justice explained that it does not fall under the crime of negligent driving causing death or injury, but there was no mention of the crime of dangerous driving causing death or injury, so I would like to confirm this if I have the opportunity. Since we do not have time, I will do so once.

Mr. Kozuka: Thank you very much for your visit to . I would like to revise this four quadrant chart. May I ask the Criminal Affairs Bureau of the Ministry of Justice to answer some questions you have just asked?

Director Seki: I am Seki, Director of the Criminal Affairs Division, Criminal Affairs Bureau, Ministry of Justice, . With regard to the point you just asked, when I look at the constituent elements of the crime of dangerous driving causing death or injury, depending on the type, it is premised that the person is driving or driving. Since Mr. Hatano mentioned the difference between Level 4 and Level 3 in terms of the degree of control over driving, it is inevitable that the evidence and judgment on the situation in which the driver is not present in your question will be made on a case-by-case basis, and it is somewhat difficult to give a definite answer uniformly. However, basically, as the Ministry of Justice answered earlier, if we consider the premise that there is no driver and no one to drive the car, the crime of negligent driving causing death or injury or the crime of dangerous driving causing death or injury will not be applied in such cases of fatal or injury accidents, and I believe that it is generally possible to say that the crime of negligent driving causing death or injury will be a problem in business. Is that all?

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

Mr. Kozuka: Thank you very much for your visit to . Mr. Ochiai, I'm sorry to cut off your time, but could you summarize it in a minute or two?

Ochiai Member: I would like to make a comment after I tell you a little about the part that was cut off earlier. Regarding foreign corporations, I think it is important to ensure equal footing in law responses, including foreign corporations, including administrative laws and regulations, taking into account the development of licenses, corporate registration for foreign businesses, and the development of corporate sanctions in some cases, which are considered to be carried out not only by the Civil Code but also by administrative laws and regulations, financial laws and regulations, and the Telecommunications Business Act.

Regarding Mr. Imai's presentation, I agree with the emphasis on the point of the duty to avoid consequences as mentioned by each teacher. In the course of the discussion, what kind of role will be sought in the program? Among the differences between what people can actually do and what programs can do, there are things that people can see but cannot do in the program, and on top of that, I think how to evaluate the fact that the program as a whole can be operated more safely than what people are doing will be a theme that is probably meaningful to discuss. While emphasizing such a perspective, I would like to agree with the opinions of the teachers so far. That is all.

Mr. Kozuka: Thank you very much for your visit to . Mr. Imai, could you please add some additional questions, including the one I mentioned earlier?

Member Imai: Yes, I would like to make a brief comment. First of all, regarding the arrangement of the figures presented by Member Hatano of the Japan Automobile Manufacturers Association, which you mentioned, I think it would be better for you to point out. When we say reasonable predictability, the word reasonable is particularly often used in British and American Law, but in fact, the meaning of reasonable is included in the interpretation of predictable, so I think it is better to separate them by whether they are predictable or not for the sake of simplicity. In addition, it is self-evident that criminal liability cannot be imposed when it is impossible to predict or avoid, so I think it would be better to make a table like this, and I think it is consistent with the purpose of Member Hatano, so I also agree with this picture.

In addition, Mr. Takahashi, Mr. Inadani, and Mr. Fujita gave us very constructive opinions. The topic of this time was to explain what the precedents are, so I explained it like this. In particular, Mr. Inadani may be familiar with it, but I personally think about how much predictability is necessary in the relationship between the severity of the result and the probability of the result. I think that the idea of understanding the possibility of avoiding the result and the predictability in conjunction with each other in this way will lead to future discussions and Issue. I think that the direction is close to the idea that Mr. Fujita summarized very well. I think that it is possible to consider the level of the fact that it is impossible to predict the occurrence of the accident at the time of the design when the accident occurs several years after the creation of the program. I also think that predictability should be organized in relation to such avoidance measures while standing on the latter viewpoint. From this understanding, negligence under the Penal Code should be quite close to negligence in tort under the Civil Code. Therefore, as Mr. Goto said at the beginning, I also said that we can have a unified understanding from the viewpoint of the use of punishment for deterrence. That is all.

Mr. Kozuka: Thank you very much for your visit to . I think the discussion has been organized. Now, let's move on to the issue raised by the members. I would like Mr. Takahashi to explain the materials. Thank you very much.

Takahashi Member:
(Hereinafter, we will explain based on "Material 6: automated driving and the so-called Article 38, Paragraph 1 of the Road Traffic Act")

Yes, as we have talked about in the past discussions, I believe that everything boils down to how to understand the duty to avoid consequences. The biggest problem with the duty to avoid consequences, as I would like to tell the media, is the so-called Article 38, Paragraph 1 problem. I would like the JAMA to be fully aware of this point. Please go to the next page.

There is a first sentence and a second sentence. The first part defines the scene when a pedestrian has not yet crossed the crosswalk and is about to do so. Therefore, the pedestrian is not yet on the crosswalk, but on the sidewalk. Then you must proceed at a speed that allows you to stop, unless it is obvious that you are not on the sidewalk at the end of the crosswalk. The speed at which this stop can be made is specifically less than 5 kilometers. In other words, it means the speed at which you can stop naturally just by applying a normal brake without applying a sudden brake. The idea is that you can trust that a car will approach you at a stopping speed, slowly and gradually slowing down at the stop line or just before the crosswalk, as long as it is a crosswalk, and it is to protect that expectation. In other words, even if there is no traffic light, it is a regulation that thoroughly protects the safety of crossing pedestrians by imposing on the vehicle side that the vehicle should be driven at a speed that can be stopped even if a person or the like runs out as long as it is a crosswalk. You ask me if this kind of accident really happens, but actually this kind of accident happens a lot. Where there are many of them are elementary and junior high school students on the school route and during school hours. There are many children who run out. However, even if you jump out, if you cause death with this, it will always be filed, and it will almost certainly be tried. You will be put on trial, and there is a possibility that you will be sentenced to imprisonment if you are on the school route, during school hours, or in elementary school. This is the defense range of the first sentence of Article 38, Paragraph 1. It says that it is obvious that there is no pedestrian, but this is because, for example, in the case where there is a telephone pole at the end of the pedestrian crossing, and a pedestrian is standing behind the telephone pole from the viewpoint of the driver, and cannot be found, or in the case where another car is stopped just before the pedestrian crossing in the first traffic lane in the traffic lane with two lanes, and the vehicle is driving in the second traffic lane and approaches the pedestrian crossing, the person on the pedestrian crossing or at the end of the pedestrian crossing cannot be seen. In this case, you have to drive at a speed that you can stop without fail. In fact, I myself have conducted experiments in what I call automated driving. When I was hiding behind a telephone pole and standing at the end of the sidewalk on the crosswalk side, as expected, a car called automated driving passed in front of me. The first sentence of this paragraph is very important. If it is a sudden jump out, it does not mean that there is no predictability and no possibility of avoiding the consequences. At a crosswalk, you must anticipate a sudden rush. And you have to run at a speed where you don't collide even if there is a sudden rush. As a result, you have a duty of avoidance. I would like you to be careful here. Please take the next page.

The latter stage is when the perpetrator recognizes a pedestrian who is actually crossing or trying to cross. In this case, you must stop without fail, not drive at a speed that allows you to stop. In addition, you must not block the pedestrian's progress. In other words, it is a violation of the Road Traffic Act if a pedestrian passes in front of you and starts moving immediately after you pass. This is because a pedestrian may remember something and return while crossing. Therefore, you must continue to stop until the distance from the pedestrian is a safe distance from your vehicle. In fact, the prosecutor's indictment is written in such a way. It is written that there is a duty of care to proceed while confirming the safety of the pedestrian by waiting for the pedestrian to pass. Here, too, it is a violation of the Road Traffic Act if you do not create a program that allows you to continue to stop until the distance from the pedestrian is sufficient. Please go to the next page.

If so, some people may think that it is really necessary to assume that a person hiding from a shadow suddenly jumps out or turns back. However, as long as it is a crosswalk, it must be assumed. This is because children often behave like that. If you cause a death on a crosswalk, you will be tried in most cases, and a fine will not be enough. If you are on a school route, during school hours, or are an elementary or junior high school student, you will be sentenced to imprisonment in almost all cases, and in some cases, you will be sentenced to imprisonment. Therefore, if a program is created without assuming these two patterns, the person in charge of technical development in automated driving vehicles may be sentenced to criminal punishment for violating the duty to avoid consequences, which was predictable under the current road traffic law. Please go to the next page.

The first sentence of Article 38, Paragraph 1 of the Road Traffic Act says that the speed is such that you can stop unless it is clear that there is no person. However, I believe that the current automated driving system is based on the premise that it is okay to make a temporary stop only when it is clear that there is a person, or to proceed at a speed that allows you to make a temporary stop. In other words, it is based on a completely opposite recognition of the situation. Therefore, when it comes to what to do when creating a automated driving program in the future, it is necessary to focus on two points. The perspective of what to do in order to avoid accidents at hand is important, but it is not enough to just devote the best of science to it. What should be done to ensure that accidents do not occur? In particular, unlike adults, there are traffic weak people such as elementary school students and the elderly who are inferior in attention, judgment, and prediction. Based on this premise, we are making every effort to consider what should be done to ensure that accidents do not occur. Therefore, for the implementation of automated driving, we should first learn the road traffic law thoroughly and invest resources in creating a program that follows the road traffic law.

Finally, jumping out on a crosswalk does not violate the principle of trust. Rather, you have to drive on the premise that you will jump out on a crosswalk. That's all.

Mr. Kozuka: Thank you very much for your visit to . I understand that Mr. Fujita will be attending the meeting in about five minutes. Do you have any comments?

Member: As I have already mentioned, regarding compliance with traffic laws and regulations that have specific contents and are given specific norms of behavior, will it be discussed within the framework of the duty to avoid consequences, or will it be easier to understand as a system if it is imposed as a constraint on programming that a program that does not comply with traffic laws and regulations should not be created in the first place? I would like you to separately consider in which form it will be organized in the future. With regard to the specific example that Commissioner Takahashi mentioned, it is exactly as you said that a program that allows such things should not be created, but there is room for discussion on where it should be positioned in the legislation, so I would like you to consider it in the future.

Mr. Kozuka: Thank you very much for your visit to . Then, Mr. Inadani, please go ahead.

Basically, whether it is the Road Traffic Law, the Penal Code, or other laws related to traffic safety systems, I think the aim is to realize a safer traffic system. When the current Road Traffic Law, Penal Code, and other related laws and regulations are applied to human beings, I think there are places where it is not clear whether they are in the best condition from the viewpoint of improving safety. However, originally, from the perspective of regulation digitalization, the Digital Ad Hoc Working Group has been consistently engaged in it, but from the perspective of regulation digitalization, I think there is a general theory that we must look at it from the perspective of how we can realize a safer and more efficient society by substituting AI and robots for the tasks assigned by regulation, which are cognition, judgment, and execution performed by humans. Since the strong point of human and the strong point of AI are basically different, it is better to keep in mind that if AI and robots are required to move in the same way as humans, safety and efficiency may deteriorate. In that case, how do the current regulations on traffic system safety, such as the Road Traffic Law, the Penal Code, and the Road Transport Vehicle Law, function to improve safety? What is the point of mechanical substitution in that function? In order to do so, we must consider how the laws and regulations themselves will change. Although this is not the case with this proposal, if it is simply said that it is good to have AI and robots observe the current laws and regulations, unnecessary things such as automatic stamp presses will be created, and the scope for improving security and efficiency through AI and robot activities will be narrowed. I think it is necessary to pay attention to the fact that it may eventually go against the argument that we should improve the security of the entire transportation system by using the automated driving system well and increase the profits of society as a whole. As you stated at the beginning of this Sub-Working Group, the introduction of the automated driving system is an indispensable Issue for maintaining the basic infrastructure of Japanese society and the standard of living of many people. Given this, I would like to say that it would be better to deepen discussions in the direction of improving the security and efficiency of the traffic safety system from the perspective of appropriately utilizing the automated driving system, after conducting a more in-depth analysis with the intent I just mentioned, including the relationship between relevant laws and regulations, the programming and design of the automated driving system, and the road environments surrounding it. That's all.

Mr. Kozuka: Thank you very much for your visit to . It is difficult to proceed with the discussion any further because the closing time of the meeting has already passed, but I feel the same way about what Mr. Inadani said at the end. At present, all laws, including the Road Traffic Law, are written in characters, and in the end, humans are supposed to read, interpret, and apply them. However, in the case of automated driving, it must be made into a program, and I think that this is, in a sense, a translation. Human language, in the Japanese case, is translated into a program language, and I think that how to translate it is how to make the society in which automated driving has been put into practical use better, in other words, how to make the society as accident-free as possible. I also feel the need to discuss how to translate it in the future. As Mr. Inadani pointed out, I would like to propose it as a sub-working group.

In addition, several professors pointed out the relationship with infrastructure today, which has not been discussed in this sub-working group so far, so I would like to consult with the Secretariat about what kind of method can be taken up.

Now, I would like to summarize today's sub-working group. I would like to ask Mr. Yoshihisa Hasui, Deputy Director-General of the Economy, Trade and Industry Ministry in Digital Agency, to summarize. Thank you very much.

Deputy Director-General: Once again today, thank you very much for your very enthusiastic discussion, despite your busy schedule. I think the discussion may have been a little too packed. I am sorry that it always takes a long time.

Today, after discussing the nature of civil liability, we received various opinions and suggestions, including the significance of negligence in criminal liability, clarification of the scope of liability, and Issue in the Road Traffic Act, which was proposed by Mr. Takahashi. Regarding civil liability, Mr. Goto summarized the components to be considered in an easy-to-understand manner. He presented three options that could be considered. In response, Mr. Yokota also received opinions mainly from the perspective of rapid recovery from damage, and I think he has deepened his understanding of the factors to be considered in the detailed system design in the future. We also received various opinions from the teachers.

With regard to criminal liability, Mr. Imai gave an overview of the judicial precedents on criminal negligence, and I think that the predictability and avoidability of the examination Issue and the results that are extracted were very much discussed. In addition, I summarized the way of responsibility in the form of correcting the four quadrants created by the Japan Automobile Manufacturers Association. I think that this sub-working group consisting of people from various backgrounds has advanced a common understanding of the concept of criminal liability. This was pointed out by Chief Kozuka. At the same time, Mr. Takahashi also pointed out how we can translate the Issue of Article 38, Paragraph 1 of the Road Traffic Act in a form tailored to automated driving, AI, etc., while it is extremely difficult to comply with it.

As mentioned above, very important points were raised. I believe that there was also a discussion of the balance between predictability and avoidability, but I believe that it is also important to set qualitative and quantitative standards in safety standards and guidelines regarding the content summarized today and how automobile operators should approach compliance with the Road Traffic Act.

I am wondering if we can make a certain summary in May, including the themes of civil liability, criminal liability, and the Road Traffic Act that you discussed today, and we will continue to consider them based on the opinions you gave us today. At the next 5th Sub-Working Group, the Secretariat will present a certain summary based on the discussions we have made so far. I would like to ask for your continued support to the members and relevant ministries and agencies. Thank you very much. Thank you very much for your time today.

Mr. Kozuka: Thank you very much for your visit to . As we were very limited in our time, we would appreciate it if you could send any additional comments to the secretariat no later than next Friday.

In addition, as usual, the materials for the plenary session will be announced on the Digital Agency website at a later date. The minutes will be announced on the Digital Agency website after confirmation by the professors.

The next 5th Sub-Working Group meeting, which Mr. Hasui also mentioned, is scheduled for late April. We are currently considering how to structure the agenda, so we will contact you later.

I would like to conclude today's fourth session of the Sub-Working Group. Thank you very much.