Ethics in Autonomous Driving

LGST 612 – Responsibility in Professional Services

Final Paper

Moral problems in autonomous driving

Prof. Alan Strudler

October 21, 2016

Matthias GERG

Table of Contents

1.) Introduction 2

2.) Legal basis for autonomous cars 3

3.) Self-driving cars and ethics 3

3.1 Non-identity theory 4

3.2 Utilitarian view (Consequentialist) 5

3.3 Kantian view (Deontological) 6

4.) Personal recommendation for robotic car programming 7

* Course concepts and theories covered in class in bold print

1.) Introduction

For decades, autonomous or self-driving cars seemed to be wishful thinking and totally unrealistic. However, the progress that has recently been made in that industry suggests that the motor industry will be disrupted “far sooner, faster and more powerfully than one might expect.” In May, BMW’s CEO Harald Kruger announced that they will launch a fully-autonomous car until 2021 and also tech giants like Google work on the same technology. By 2030, McKinsey & Company predicts up to 15% of new vehicles sold will be self-driving and thus a heated discussion started about the various implications this would have on driver safety and people’s everyday lives. First of all, experts project a decrease in traffic related deaths, the largest accidental killer of Americans, claiming 35,092 road fatalities in 2015. Even though safety standards have increased significantly, human beings, who control a vehicle are faulty and therefore accidents end fatal very often. Thus, “autonomous cars remove the human element from driving and by doing so, theoretically, the number of traffic related deaths caused by humans behind the wheel will drop.” However, accidents will still happen when humans and robots are sharing the road and therefore questions around liability and morality inevitably arise. Should self-driving cars in any event protect their owners or is there a trade-off to be made in order to achieve the best possible outcome in terms of casualties?

The following paper will discuss moral and ethical issues related to autonomous cars by analyzing different ethical viewpoints and theories. Finally, the paper will conclude by taking a clear stand on how car manufacturers ought to program and design self-driving cars of the future in order to act morally and avert damage from humans in the most “ethical” way.

2.) Legal basis for autonomous cars

In many fields, detailed legislative texts provide clear guidance and orientation. The subject matter of self-driving cars, however, is so groundbreaking that law simply hasn’t caught up yet and a “legal framework for autonomous vehicles does not yet exist (…)”. And even if – in road traffic, the law and being morally right often diverge. Just imagine breaking the law for speeding when taking your expecting wife to a nearby hospital. Should self-driving cars never disobey the law in autonomous mode? As Lin (2013) underlined, “our laws are ill-equipped to deal with the rise of these vehicles”6 and this is exactly where ethics come into play. As a legal basis has not been created yet, “we have the opportunity to build one that is informed by ethics.”6 Consequently, the following section will identify ethical “costs and benefits” in the context of robotic cars and then scrutinize and evaluate the matter from different angles of ethical theory.

3.) Self-driving cars and ethics

At first, there are various arguments that strongly favor the introduction of autonomous cars from an ethical perspective. They will enable disabled or physically impaired people, or also the elderly to drive in their own vehicles which would increase their mobility and quality of life. In addition, experts predict that self-driving cars will have a positive environmental footprint due to their rationality. Most importantly, however, self-driving cars promise to be safer and less faulty than human drivers and will therefore decrease the traffic-related death toll.

This, on the other hand, directly leads to the downsides that have to be considered from an ethical perspective. Passenger safety, which will be in the focus of this paper’s analysis, comes with the morally delicate topic of trading-off lives. Most likely, the victims of future crash scenarios with robotic cars will not be the same people who would have died in a regular car accident and vice versa. Before applying various ethical theories to this issue, one also has to mention another important downside of autonomous cars: privacy and data security concerns. If hackers could take over control of a self-driving car, which is far from unrealistic in times of cyber-crime, they could easily not only infringe privacy rights, but also purposefully put the lives of car owners at risk.

As stated before, the key matter of debate among industry experts and scientists currently is around how to program car computers and which algorithms would guide their decision making progress. Especially in worst case scenarios, when an accident seems inevitable, different ethical theories would offer different solutions.

3.1 Non-identity Problem

Oxford philosopher Derek Parfit introduced the non-identity problem in the early 1980s and referred to the conflict of current versus future generations’ well-being. In Parfit’s theory, there is a policy choice to be made of either depleting a natural resource or conserving it. Thus, the quality of life for the current generation is competing with the well-being of future generations. Lin (2013) relates this to the potential benefit of autonomous cars cutting fatality rates by half. This would still imply that, most likely, a new group of victims would be affected which is a troubling circumstance. So can we infer the policy of harming future people by enjoying the safety benefits autonomous cars can offer as being unethical?

Even though a non-identity problem exists in principle in this