Autonomous vehicles
they knew that in a crash, they had no choice in who would get hit by the car. In this case, the public would put immense pressure on governments and manufacturers to give individuals the chance to implement their own morals. On the other hand, this could prove complicated and would require a whole new approach to designing the code and algorithms to the self-driving car. Moreover, there would be a much more definite ethical system if the morals of the vehicle were determined and fixed by an institute such as the manufacturers or the government (Patrick 2020). Another major consideration is how the person responsible for deciding the vehicle’s ethics acts, or more s pecifically: ‘What if the person who decides how the vehicle should act in certain situations is under moral uncertainty?’ Moral uncertainty is when someone has all the non -moral facts (such as the details of a crash scenario) but remains uncertain about what morality requires of them (Lin 2020). For example, if the person must choose whether the vehicle hits Pedestrian A or Pedestrian B, they would have to divide their belief in what is the correct answer between 2 options (in this case). In other words, they would compare their belief that hitting Pedestrian A is the right answer, to their belief that hitting Pedestrian B is the right answer. Prescriptive uncertainty (uncertainty created by fixed factors) would also have to be a considered when programming the vehicle. Law, culture, and religion would all create prescriptive uncertainty. For example, in certain societies, it may be frowned upon to hit an animal. This is assuming, however, that there even is a correct answer to these ethical dilemmas, ethics is relative, so how would a manufacturer create an acceptable vehicle that is in accordance with different moral views? In addition, let us assume that there was a ‘correct’ answer between the vehicle hitting Pedestrian A or Pedestrian B, and the vehicle chooses the ‘incorrect’ outcome: who is held liable? Is it the manufacturer, the individual responsible for the morals of the vehicle, or someone else, and what would be the consequences? (Patrick 2020). A proposed solution is ‘ethics settings’ . These are essentially a selection of fixed settings determined by engineers or designed using machine learning algorithms. These predetermined settings allow the driver to decide their own outcome in a crash situation, and although this may seem like a morality test for an individual, this would eliminate the search for someone responsible for a crash, because even though the self-driving car took action, it was the driver who chose this outcome. But as discussed at the start of this section, this is a very impractical idea, and different people would have different moral views, so the settings would have to be extremely open and cover all possible eventualities. To add to this, a public poll conducted by Open Roboethics initiative (ORi) questioned people on the famous tunnel problem. This is a hypothetical dilemma where you are driving a car on a mountain road, quickly approaching the entrance to a tunnel. Suddenly, a child runs on the road and you cannot brake in time. There are 2 options: swerve into the side of the tunnel entrance which would result in your own death, or you hit the child and kill them. The poll found that 64% of people would save themselves and ultimately run over the child in the tunnel problem, and 12% of the respondents agreed that the manufacturers decide the moral decisions of an autonomous vehicle. This suggests that the vast majority of the population is against the engineers making moral decisions on their behalf. Will this force the autonomous industry to solely follow the path of ethics settings? These philosophical questions and points of views will not be resolved anytime soon and will have to be thoroughly investigated by governments and even courts for different parties (the manufacturers, the public, the government) to come to an agreement (Patrick 2020).
26
Made with FlippingBook interactive PDF creator