Skip to main content

Never Mind the Trolley: The Ethics of Autonomous Vehicles in Mundane Situations

Abstract

Trolley cases are widely considered central to the ethics of autonomous vehicles. We caution against this by identifying four problems. (1) Trolley cases, given technical limitations, rest on assumptions that are in tension with one another. Furthermore, (2) trolley cases illuminate only a limited range of ethical issues insofar as they cohere with a certain design framework. Furthermore, (3) trolley cases seem to demand a moral answer when a political answer is called for. Finally, (4) trolley cases might be epistemically problematic in several ways. To put forward a positive proposal, we illustrate how ethical challenges arise from mundane driving situations. We argue that mundane situations are relevant because of the specificity they require and the scale they exhibit. We then illustrate some of the ethical challenges arising from optimizing for safety, balancing safety with other values such as mobility, and adjusting to incentives of legal frameworks.

This is a preview of subscription content, access via your institution.

Notes

  1. We define an “autonomous vehicle” as a motorized ground vehicle with the capability of highly or fully automated driving, what is sometimes called automation level 4 and 5 (SAE International 2016).

  2. Many contributions do not distinguish as strictly as we do between trolley cases and trolley problems. However, we think this distinction is important. We are grateful to anonymous reviewers for their encouragement to make this distinction clear upfront.

  3. We do not endorse this claim. For many authors this claim motivates trolley cases as relevant to the ethics of autonomous vehicles (see Nyholm and Smids (2016) for an overview).

  4. The objections that we discuss in this paper largely supplement objections discussed by Goodall (2016) and Nyholm and Smids (2016). Trolley cases, according to Goodall (2016), are problematic in that they (1) pose a false dilemma (in fact, there are more than two options), (2) assume certainty over outcomes, (3) assume certainty over the environment, and (4) are in fact rare. Nyholm and Smids (2016) argue that trolley cases are not perfectly analogous to the situations of autonomous vehicles. This is because (5) the decision problem is different (e.g. with respect to when a decision is taken, and the numbers of agents involved), (6) the issues of moral and legal responsibility are in fact relevant but neglected by trolley cases, and because (7) decisions in fact need to be made under uncertainty. We take on board the points about uncertainty, that is, point (2), (3), and (7) in our fourth objection. We also agree with points (5) and (6) as raised by Nyholm and Smids (2016) but we do not pursue these points in our paper in this way (but see note 5). Our positive proposal on mundane situations incorporates the proposal made by Goodall (2016) on the importance of risk-management but it also extends this proposal in that we highlight considerations beyond risk and safety.

  5. Specifically, we are not aware of a full discussion elsewhere of our first objection (that given technical restrictions, trolley cases rest on assumptions that are in tension with one another) and our second objection (that trolley cases cohere with a certain design framework). Our third objection (that trolley cases look for a moral answer when a political answer is called for), can be seen as a version of point (5) made by Nyholm and Smids (2016). However, we instead focus on a specific instance of their point highlighting a difference between moral and political philosophy. Our fourth objection (that trolley cases might be epistemically problematic) combines objections made by many others (Elster 2011; Fried 2012; Wood 2013; Kagan 2015).

  6. We restrict the discussion to collisions, given the context of autonomous vehicles. A more general definition would instead be formulated in terms of distributions of harms and benefits.

  7. Nyholm and Smids (2016) argue that each of these three assumptions is not met in the reality faced by autonomous vehicles and that trolley cases are therefore not a good analogy.

  8. For a helpful overview see Nyholm and Smids (2016: 1280).

  9. In the literature on autonomous vehicles, Lin (2014) argues that trolley problems are “meant to simplify the issues in order to isolate and study certain variables.”

  10. See http://moralmachine.mit.edu. For another example see Frison et al. (2016).

  11. Trolley cases in some ways resemble the party game of “would you rather” questions. Some of the reasons for which “would you rather” questions exert a certain attraction might also explain why trolley cases are captivating.

  12. However, it stands to reason to what extent this situation – a paradigmatic instance of a collective action problem in which individual incentives lead to an outcome that is overall worse – is actually typical of politics.

  13. That trolley cases are not contradictory in this way is supported by the fact that they are clearly conceivable. Their conceivability suggests that trolley cases are epistemically possible.

  14. Because this scenario raises not only the question of to whom the harms accrue but also the question of whether harms should be minimized, it fails to isolate two values. An intuition is hence no clear indication about relative importance of two values.

  15. Some argue against trolley cases on the basis that they are rare (Goodall 2016). We do not pursue this objection. Even if the situations that give rise to trolley cases are rare, they will occur with certainty over the long run. Moreover, regardless of whether these situations in fact occur, autonomous vehicles still need to be programmed to behave in one way or another to prepare for the eventuality of unavoidable collisions. In short, the low frequency of trolley cases is, as such, not yet an argument against their relevance for the ethics of autonomous vehicles.

  16. It should be noted that Nyholm and Smids (2016) discuss decision-making situations – such as the number of agents involved, and the information available. Nyholm and Smids do not discuss different design approaches in artificial intelligence.

  17. Despite these limitations, trolley cases here play their role as a didactical device.

  18. In this way the methodology of trolley cases differs significantly from that of trolley problems which aims at the formulation of moral principles. We thank an anonymous referee for pressing us to make this clear.

  19. Nyholm and Smids (2016: 1282) make a similar point in that they identify as a disanalogy between autonomous vehicles and the trolley problem the fact that the former is a decision-situation involving “multiple stakeholders” whereas the in latter “the morally relevant decision-making is done by a single agent.” However, their objection is much more general. They do not highlight this distinction between moral and political philosophical approaches.

  20. Judith Jarvis Thomson reminds us that “we should be troubled by the fact that so many people have tried, for so many years—well over a quarter of a century by now—and come up wanting.” (2008)

  21. This question is the starting point of Gogoll and Müller (2017) for their discussion of whether ethics settings should be mandatory or personal.

  22. Nevertheless, trolley cases can play a useful role as a didactical device by illustrating the issue of the ethics of user settings (Millar 2014; Gogoll and Müller 2017; Millar 2017).

  23. Other examples of relevant values are values of social justice, such as sustainability, privacy, and equality of access (Mladenovic and McPherson 2016).

  24. In most legislations, pedestrians’ responsibilities are higher when crossing the street outside of dedicated crossings. Unlike in crosswalks, drivers might not have to yield to pedestrians.

References

Download references

Acknowledgements

For their helpful comments and discussions, I am grateful to Chris Gerdes, Geoff Keeling, Patrick Lin, Jason Millar, Jesse Saloom, and two anonymous referees of this journal.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Johannes Himmelreich.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Himmelreich, J. Never Mind the Trolley: The Ethics of Autonomous Vehicles in Mundane Situations. Ethic Theory Moral Prac 21, 669–684 (2018). https://doi.org/10.1007/s10677-018-9896-4

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10677-018-9896-4

Keywords

  • Applied ethics
  • Ethics of autonomous vehicles
  • Ethics of technology
  • Driverless cars
  • Ethics of self-driving cars
  • Methodology
  • Thought experiments
  • Autonomous vehicles
  • Self-driving cars
  • Trolley problem