The use of artificial intelligence (AI) offers new perspectives not only for the operation of autonomous vehicles, but also for the aviation industry. While the autopilot has been standard for decades, scientists are currently developing simplified command input via voice recognition or even ideas for electronic copilots. This raises not only technical questions, but also ethical ones: “Basic ethical rights and social norms and values have to be respected, if the possibilities of artificial intelligence simultaneously reduce the user’s control,” emphasizes moral theologian Prof. Dr. Alexis Fritz of the Catholic University of Eichstätt-Ingolstadt (KU).
Basic ethical rights and social norms and values have to be respected.Prof. Dr. Alexis Fritz, Chair of Moral Theology, KU
Reflecting on an ethical framework for AI offers the opportunity to fully exploit the potential of this innovative technology. Legal standards or guidelines would have to be in line with ethical ones. Fritz and his team are now researching the fundamentals of this within the framework of the project “Artificial Intelligence European Certification under Industry 4.0” (KIEZ), which is led by Airbus Defence and Space GmbH and funded by the Federal Ministry of Economics. In addition to Airbus, other technical partners, such as several Fraunhofer Institutes, are involved in the three-year project—as are DFS Deutsche Flugsicherung GmbH, and the European Union Aviation Safety Agency.
A fundamental challenge for the use of artificial intelligence in aviation is the mandatory certification of AI technologies. Aircraft components are safety-critical and are therefore subject to high quality standards, as errors can have fatal consequences. However, complex AI cannot currently be used, because there is a lack of corresponding certification procedures. Currently, regulations only exist for systems that are designed deterministically. It is therefore necessary to create standards for technologies that are not rigidly programmed but react flexibly in real time. “Certification methods have the goal of sufficiently reviewing whether the systems are reliable. In addition, they are indispensable for convincing the public of this reliability,” says Fritz. Due to the risk of harm in the aviation sector, the aim is not to minimize risk, but to exclude it.
The debate about the use of AI is particularly characterized by the specific interaction between human and machine and the question of how responsibility can be assigned when humans and computers interact with each other. As Fritz describes it: “Strictly speaking, AI cannot make decisions or act ethically. It can therefore not bear any responsibility. Nevertheless, in the network of human-machine interaction, there is a danger that it is no longer clear who is responsible for what when tasks are delegated.” Therefore, the attribution of personal abilities and characteristics to AI systems, like all attempts to humanize technology, should be questioned critically.
“In our research network, we are all concerned with the ethically valid certification of AI technologies as a contribution to responsible use. This involves thinking about more than human dignity or discrimination. The robustness and security of AI, and even data quality, are ethical goals,” Fritz emphasizes. The aim of his project is therefore to investigate how and on what fundamental premises basic ethical rights and social values can play a key role in the design and use of AI in aviation in the future.
The research association offers the opportunity to introduce moral-ethical questions already in the development phase—in close exchange with technical and legal partners.Prof. Dr. Alexis Fritz, Chair of Moral Theology, KU
As a result of a discussion about the expectations and fears surrounding artificial intelligence, a large number of governmental bodies and renowned non-governmental organizations at the national, European, and international level have now developed their own ethical criteria for the development and use of AI. Together with his project collaborator Felix Steinbrecher and a team of students, Fritz has meanwhile carved out 380 guidelines for AI. These draw on a broad range of sources: from a catalog written by a Latin American feminist organization, to international stakeholders like Google and Microsoft. For the KIEZ project, the EU initiative “Trustworthy AI” is particularly important.
The KU has a strong profile in the social sciences, economics, and the humanities and an ethical model based on the Christian image of humanity. Against this background it has a particular responsibility to relate technical progress to social change, to identify potential risks and challenges, and to make a contribution from a scientific perspective to a digital society that is oriented toward people.
“Due to the dynamic nature of the project, the continuous systematic recording, stocktaking, review, and comparison of all these initiatives is very time consuming and costly, but also very fruitful and important,” says Fritz. As a first step, the project team is currently researching the relevant regulations worldwide in order to review and compare them. In addition, it is important to find a common language for central terms such as “values,” “goals,” or “criteria” that engineers, lawyers, theologians, and philosophers alike can understand.
Subsequently, the researchers plan to develop an independent conceptual design, which contains ethical-technological criteria, principles, and guidelines. These are also being developed in an ongoing exchange with the technical and legal partners on the overall project. “Legal standards must be in line with ethical standards. Therefore, we want to compare ethical criteria with existing legal regulations and provisions in this field and point out where further developments are necessary. The research association offers the opportunity to introduce moral-ethical questions even in the development phase—in close exchange with technical and legal partners. This is what makes the project so exciting for us,” emphasizes Professor Fritz.