The “SocAIty” study 2021: The most important insights
Law: In evolutionary interplay with progress
Technical developments are often a few steps ahead of legal frameworks. These advances not infrequently provoke uncertainty. That also applies to autonomous driving, even though self-driving cars and buses are rarely on the road in public traffic today. The systems and vehicles are either at the prototype stage or can only drive autonomously under specific conditions. So the technology is not yet at the level that is discussed in theory. For that reason, legal systems around the world face the challenge of thinking in future scenarios. Precedent-setting cases will only follow in the course of the coming years. From the perspective of many experts who participated in the study, legal enactments for autonomous driving are currently primarily a matter of regulating technology on a large scale, rather than individual vehicles or systems.
That requires a new, evolutionary approach to regulation and legislation. That approach has to proactively demand that legislators have a technologically open-minded understanding and be safeguarded over the long term through continuous exchange with economics and research. Benefits: Legislation does not regulate ahead of developments or the market, which in turn supports industrial openness to innovation and investment and builds confidence among users. One possible approach would be what is known as the trial mechanism: this starts from an assumption of an “imperfect status quo.” In a transparent process, it instead strives for the greatest possible safety and acceptance.
The study additionally addresses the central question of liability: Who is liable in the event of an accident that involves an autonomous vehicle? Is it the driver, the owner, or even the manufacturer? Most of the experts believe it is likely that, even in the future, every person who occupies an autonomous vehicle as a driver or who owns an autonomous vehicle will bear a high degree of responsibility and have to be liable for his or her own errors. This is also important for the industry.
“Ultimately, liability cannot fall unilaterally at the expense of one party. There have to be incentives for everyone involved to behave carefully. Only then will manufacturers likewise be motivated to continue doing research and regularly bring new innovations to the market.”
The industry is also interested in a harmonization of regulatory frameworks worldwide. Ultimately, cars as well as other systems will be sold around the world. “We need an aligned European legal framework that deals with approving these vehicles. Because we won’t get very far with a national-level patchwork,” says Richard Goebelt, Member of the Board of Management & Head of Vehicle and Mobility at the TÜV-Association. And Automotive Engineer / Advisor Sandy Munro of Munro & Associates adds that “for the moment, I would be happy if there were a uniform regulatory field for Europe. It is also a big market and then you would have a bargaining tool for talking to other countries and saying that if you acknowledge our vehicles and regulations, then we will recognize yours.” With respect to regulations and jurisprudence, Germany is a positive example, according to the study, because a legal framework has been established here that can be used as a model internationally.
Ethics: Relationship of trust between human and machine
The speed with which autonomous driving becomes widespread in everyday life depends above all on social acceptance. According to experts, there are still many challenges that have to be overcome in which ethical and moral aspects will play an important role. Overall, there is a societal tendency to demand zero tolerance for errors with respect to autonomous driving. Additionally, one central aspect is the controversy over what is known as the “moral dilemma.” Applied to autonomous driving, what we mean is decision making scenarios in which there are multiple possibilities that always lead to an undesirable outcome. One frequently used example is that of a hypothetical evasive maneuver in which people are injured or even killed no matter which way one swerves. This raises hypothetical questions like: Who deserves to be avoided more? Which life is worth more? An Ethics Committee appointed by Germany’s Federal Ministry of Transport took up these issues in 2017.
The debate around this question is often emotional and, in the eyes of some experts, ideologized. “Who should we drive around first? If we continue to set our agenda this way, we will never move forward,” points out Christoph Lütge, Director of the Institute for Ethics in Artificial Intelligence at the Technical University of Munich. “We can’t solve these dilemmas, nor will we have a perfect approach to them.” That is why most of the experts recommended a shift away from theoretical discussions about unsolvable moral quandaries and toward a more solution-oriented approach to avoiding accidents. Manufacturers and safety technology developers are assigned a crucial role in this. In order to shape the discourse further and build social acceptance, most experts support clear guidelines and principles that are formulated by committees or societal representatives, generally accepted, and continuously refined through practice.
Data: The tension between performance and data protection in autonomous driving
In the future, data will be collected and processed in larger quantities for autonomous driving than in almost any other field. Above all, this includes environmental image data and movement data for individual vehicles as well as that of other road users. Data collection raises questions and uncertainties: Will the passengers of tomorrow be “glass passengers” or will they be able to decide how their data are used? With respect to public autonomous services of the near future, such as shuttle services and fleets, experts see data ownership or at least predominant data usage rights as clearly belonging to the service providers. Likewise in the case of “robotaxis,” operators and providers will use that data for their own purposes. Ultimately, as fleet operators, they have to offer the service and improve it in their own interest and they need data in order to do that. On top of that, there is also the infrastructure for data collection and analysis.
However, some experts believe that a model in which data belongs to the users is conceivable. Manufacturers and operators would then be permitted to use that data for specific purposes, similar to how this currently happens with respect to smart phones. The objective should be to grant users a degree of responsibility with respect to their own data. Some experts emphasize that ostensibly clearly applicable legal frameworks and laws are often annulled with the click of the “Accept” button on the General Terms and Conditions.
Effective protection of personal information and the user’s private sphere is one of the central themes concerning social acceptance of autonomous driving. That is because the digital and network infrastructure are also targets for manipulation, for instance by hackers. That is why it is important to develop appropriate security standards so as not to compromise users’ confidence. Protection requires anonymization and encryption technology for data. Auto manufacturers and mobility providers must additionally develop and implement efficient cybersecurity concepts. According to the experts, manufacturers are ultimately challenged to account for and integrate data protection and data security solutions from the start as they develop autonomous vehicles.
The experts agree that a quick and comprehensive implementation of and penetration with autonomous vehicles will require the existence of a corresponding digital infrastructure. For that reason, they recommend that a 5G standard be established and implemented as quickly as possible at the international level. Another cornerstone for an autonomous future is comprehensive cloud edge computing or onboard edge computing. Apart from the enormous potential for a more efficient and therefore also a more ecologically sustainable future, networked and data-driven mobility concepts can also have an enormous social impact. That includes building new infrastructures and services that are oriented to human needs. Ideally, that would lead to a new form of inclusive and social mobility. For instance, people could be mobile if they do not have a drivers license or are unable to drive a car on their own for health reasons. To make this into a reality, the experts wish for visionary pioneers in economics, science, and policy as well as social trust in data’s efficacy and its abilities.
74 pages, EN