What has the Trolley Dilemma ever done for us (and what will it do in the future)? On some recent debates about the ethics of self-driving cars

Per November 2016; see http://​perma.​cc/​F9N2-N6QS.

This claim can be understood in more than one way, as one reviewer rightly pointed out. One reading would see a triviality in claiming that usefulness depends on use. Usefulness, however, is different from use since the former, but not the latter, is an evaluative term. Here, however, I want to express the idea that particularly with regard to SDCs, their promise to reduce traffic fatalities depends on their widespread use because we cannot reap the SDCs full benefits if they are not used by everyone. The reason is that only to the extent that SDCs interact with other SDCs on the road without the interference of humans who happen to make mistakes can they work according to their purpose (which is: to make traffic safer). The same holds true with regard to vaccination, for example: only if a majority of people are vaccinated, the usefulness of vaccination is fully implemented.”

Some of the voices discussed below were not raised in an academic journal, but, for example, in publicly accessible blogs. Nevertheless, they deserve serious attention since they reflect (and probably influence) the public opinion about SDCs.

This is not to say that just because humans happen to reason that way, it is the correct way, as one reviewer remarked. However, the argument pursued here goes a step further and argues that there is some reason behind the fact that humans sometimes judge and act in a utilitarian way.

Thanks to an anonymous reviewer for making this point.

One reason to assume the truth of this thesis is the fact that we are able to talk about morality, rightness and goodness, even in the face of widespread pluralism. The fact that we can and do discuss moral issues is explained by a shared concept of morality, as opposed to differing conceptions of morality. I want to thank an anonymous reviewer for pressing me to clarify this extremely important issue.

An argument somewhat similar to the one advanced by Nyholm and Smids can be found in Hevelke and Nida-Rümelin (2015). In their view, the ethics of SDCs is not accessible using the TD. They argue that a utilitarian algorithm can be ethically justified as long as the victim’s identity (i.e. those to whom the SDC directs its trajectory) is not fixed prior to the crash and the algorithms reduces the total number of victims. In this case, SDCs realize a general decrease of the risk of dying in an accident, without being biased towards one group of persons or an individual having a higher risk of getting hurt by a SDC. Hevelke and Nida-Rümelin argue that this situation differs from a Trolley Dilemma where the identity of the actual and possible victims are known, thus the ethics of SDCs should rather not make use of the Trolley Dilemma to come to an ethical conclusion about crash-optimizing algorithms. However, when a group decides about the rules guiding SDCs in their community, no one knows on which side s/he will end up when it comes to a crash: they can end up in the group consisting of five persons or they can be the single person on the track (or be the passenger of the SDC). This means that should we try to find a rule guiding the behavior of SDCs in advance, the TD can actually be used to describe that situation with the same results that Hevelke and Nida-Rümelin find. See also Wolkenstein (2017b) for a discussion of this kind of argument.

That the TD is important might seem like a weak conclusion, as a reviewer has pointed out. However, since there are voices that wish to downplay the role of and research into the TD, this article has provided arguments to support the TD. Moreover, since it has revealed a few insights into what ethical research can do, it has cleared the path towards locating ethics more properly in technology development (see below).

“Widespread acceptance” has to be understood in the context of the promise that SDCs make: They can reduce traffic-related fatalities to the extent that the human factor is eliminated and only (or mostly) SDCs are on the streets.