A self-driving car is equipped with a large number of sensors and cameras. The data collected using these means form the basis for decisions the autonomous vehicle makes independently. This also raises all kinds of ethical issues. To answer these issues, Continental is working on a code of ethics for the use of artificial intelligence (AI).
AI is playing an increasingly important role in vehicles. Thus, more and more vehicles are equipped with autonomous technology, with the degree of autonomy varying. For example, some cars are equipped with driving assistance systems that ensure the vehicle stays in its lane or automatically brakes for obstacles. In other cases, a vehicle can drive autonomously to a significant degree, although drivers must be able to regain control of the vehicle at all times.
Ethical issues
However, the use of AI technology also raises ethical issues. For instance, sometimes difficult trade-offs have to be made. An example is a situation where a vehicle can only avoid a collision with a pedestrian by drilling into the crash barrier, endangering the safety of its occupants. Or a situation where a collision with a pedestrian is inevitable and the vehicle has to choose between colliding with a person with a pram or an older person.
Continental believes that ethical guidelines are needed to make such choices. "AI can and should only be programmed and used in accordance with clear ethical principles," states Dirk Abendroth, chief technology officer (CTO) at Continental Automotive. Abendroth points to the major role of smart algorithms in the automotive industry, including in the field of autonomous driving. AI-based decision-making should never discriminate, Continental stresses.
Increasing transparency and data security
The focus of the new guideline is to increase the transparency of computer-based decisions and enhance data security. If computer systems take over important operations, Continental calls it a basic requirement that humans continue to understand how a self-learning system works. The following questions are central to this:
- What is the origin of the data?
- Which processing steps lead to which steps?
- How are the data stored?
- Are processes involving the use of AI in line with other guidelines and regulations, for example on equality?
Such questions are essential in shaping the mobility of the future, according to Continental. The directive the company is working on must be in line with European ethical guidelines for trusted AI. Ultimately, all of the organisation's sites must comply with the directive, as well as partners it works with.
Who is responsible for a fatal accident?
Attention has long been focused on the ethical aspects of AI-based decisions. For example, John Kington, Professor of Information Systems and Business Computing at the University of Brighton in the UK, investigated who is responsible if robots equipped with AI cause a fatal accident. Kington focused on three legal models set out by Gabriel Hallevy, a professor at the Faculty of Law at Ono Academic College in Israel. The question of who is responsible depends on the model chosen. Read more about this research here.
Moral Machine experiment
US university MIT developed the Moral Machine experiment in 2018. This is an experimental online platform designed to explore moral dilemmas associated with autonomous vehicles. The platform collected 40 million decisions in 10 languages from 2.3 million people worldwide. Central to this is the question of whether a person may be killed if it saves several others. This shows, among other things, that the vast majority of people prefer human life over animal life, as well as the safety of a group of people over the safety of one person.
At the same time, however, there are also marked differences between countries. Broadly speaking, a distinction can be made between three groups. The first group comprises North America and various European countries where Christianity has been the dominant religion. The second group includes countries such as Japan, Indonesia and Pakistan, which have strong Confucian or Islamic traditions. The third group comprises Central and South America, as well as France and the former French colonies. Among other things, the first group appears to be more willing than the second group to sacrifice the life of an older person for the sake of a younger person. Further information can be found here.
Author: Wouter Hoeffnagel
