German Ethics Commission and Autonomous Driving

It took some time, but now autonomous cars are also being eagerly discussed in Germany. While in California alone, 36 companies now have a license from the traffic authorities to test self-driving cars on public roads, and senators have six principles for the regulation and development of autonomous vehicles, Mercedes, for example, only received permission to test self-driving cars in Stuttgart just before Christmas last year.

But now this issue is also being taken seriously in Germany. The latest indication of this is the report presented a few days ago by the German Ethics Commission on Automated and Connected Driving. And the mere fact that this report exists shows us two things: that the issue is being addressed and that priorities are being set.

First of all, with this report, Germany is leading the way in how autonomous systems should be handled in public and how they should be viewed. This naturally includes how these systems should respond in the event of accidents that could result in personal injury.

But then again, it shows the focus and approach in Germany: and this is primarily driven by fear and risk aversion. In Germany in particular and Europe in general, the dangers of new technologies are seen first and foremost. Even very abstract and purely theoretical marginal problems such as the trolley problem are discussed in detail that is disproportionate to their frequency of occurrence. And in doing so, they create the actual ethical problems.

Five key topics were addressed in working groups for the report. These were:

1. Working group 1: “Unavoidable damage situations”
2. Working group 2: “Data availability, data security, data economy” – questions concerning the data generated by automated and connected driving
3. Working group 3: “Interaction conditions for humans and machines” – interface between humans and technology
4. Working group 4: “Ethical context analysis beyond road traffic” – automated and connected driving technology in the context of other (connected) technologies
5. Working group 5: “Scope of responsibility for software and infrastructure” – responsibilities for open development systems

The commission was composed of representatives from industry, government agencies, religious groups, interest groups, and universities.

The result is 20 ethical rules for automated and connected vehicle traffic, which serve as guidelines for further action.

1. Partially and fully automated transport systems serve first and foremost to improve safety for all road users.
2. The protection of people takes precedence over all other considerations of utility. The aim is to reduce damage to the point of complete avoidance.
3. The public sector is responsible for ensuring the introduction and approval of automated and connected systems in public transport.
4. People’s independent decision-making is an expression of a society in which the individual, with their right to self-development and their need for protection, is at the center.
5. Automated and connected technology should prevent accidents as much as practically possible.
6. The introduction of higher automated driving systems, especially those with the ability to automatically avoid collisions, may be socially and ethically necessary if they can be used to exploit existing potential for damage reduction.
7. In dangerous situations that prove unavoidable despite all technical precautions, the protection of human life has the highest priority in the weighing of legal interests.
8. Genuine dilemmatic decisions, such as the decision between life and life, depend on the specific actual situation, including the “unpredictable” behavior of those affected. They therefore cannot be clearly standardized or programmed in an ethically unambiguous manner.
9. In unavoidable accident situations, any qualification based on personal characteristics (age, gender, physical or mental constitution) is strictly prohibited.
10. With automated and connected driving systems, the responsibility reserved for humans shifts from the driver to the manufacturers and operators of the technical systems and the infrastructural, political, and legal decision-making bodies.
11. The same principles apply to liability for damage caused by activated automated driving systems as to other product liability.
12. The public has a right to sufficiently differentiated information about new technologies and their use.
13. It is not possible to assess today whether complete networking and central control of all motor vehicles in the context of a digital transport infrastructure, similar to rail and air transport, will be possible and sensible in the future.
14. Automated driving is only justifiable to the extent that conceivable attacks, in particular manipulation of the IT system or inherent system weaknesses, do not lead to damage that would permanently undermine confidence in road traffic..
15. Permitted business models that make use of data generated by automated and connected driving, whether significant or insignificant for vehicle control, are limited by the autonomy and data sovereignty of road users.
16. It must be clearly distinguishable whether a driverless system is being used or whether a driver retains responsibility with the option of “overruling.”
17. The software and technology of highly automated vehicles must be designed in such a way that the need for an abrupt transfer of control to the driver (“emergency situation”) is practically impossible.
18. Learning systems and self-learning systems in vehicle operation, as well as their connection to central scenario databases, may be ethically permissible if and to the extent that they achieve safety gains.
19. In emergency situations, the vehicle must autonomously, i.e., without human assistance, reach a “safe state.”
20. The proper use of automated systems should already be part of general digital education.

These points show that the Commission did not fall into the trolley trap. Such situations should be avoided from the outset. Even if human self-determination is paramount, according to this report, self-driving technology can be mandated by lawmakers for ethical reasons, thereby implicitly prohibiting manual driving. Since the protection of human life is the highest priority, legislators could even find themselves forced to do so very quickly. This is especially true once the first comparative data is available and automated systems prove to be many times safer than human drivers.

Furthermore, such systems must not discriminate between people, i.e., who is given preferential protection or who is harmed in an accident scenario. The principle of equality applies. The question of fault shifts from the driver to the manufacturer. For regulatory authorities, this means that they must acquire the skills to evaluate such cases and conduct forensic analysis.

Cybersecurity and the security of such systems are becoming increasingly important. In light of the recurring cyberattacks against companies, political parties, and institutions, this is an important area in which we must acquire expertise.

The data generated by such vehicles should remain under the data sovereignty of road users. This is an issue that could hinder us economically in advance. Germany has missed the boat when it comes to the digital revolution. Apart from SAP, it has not produced any relevant companies in this industry. The five largest American companies all come from the digital sector and are worth twice as much as all DAX 30 companies combined. This means there is a risk that Germany will not participate in a business sector that McKinsey predicts will be worth $750 billion by 2030. If the data cannot be used, this also prevents the systems from being improved. Germany is falling behind technologically.

Vehicles that require human intervention should also be clearly marked, and if such intervention is necessary, it should be communicated to the person in a timely manner rather than abruptly.

The last point, which addresses digital education, is also interesting. Correctly interpreted, people should learn how to use digital systems and machines during their training so that they can assess them. Given the heated debate today about whether programming, for example, should be introduced as a school subject, this is a real milestone.

The report did not address what the next steps should be. How should regulatory authorities proceed, what criteria should be used to grant test licenses, and how can the balance between safety and technical progress be achieved? The report also does not address the issue of standards, such as the exchange of safety-related data between manufacturers.

Conclusion

This is the first report on autonomous driving that provides guidelines and answers questions, or at least raises them. Specifically, there is a lack of guidelines on how to proceed and how manufacturers and the public should deal with this issue. This should be done quickly if we want to influence technological progress and not let California and the US dictate it, for example.

AND: I would also like to see such an ethics commission for electric vehicles and combustion engines. The present report focuses strongly on human safety. The discussion about banning diesel vehicles and combustion engines in general should also consider this perspective. Can we still justify combustion engines on ethical grounds?

This article was also published in German.