Abstract
This paper discusses the ethical implications of perverse incentives with regard to autonomous driving. We define perverse incentives as a feature of an action, technology, or social policy that invites behavior which negates the primary goal of the actors initiating the action, introducing a certain technology, or implementing a social policy. As a special form of means-end-irrationality, perverse incentives are to be avoided from a prudential standpoint, as they prove to be directly self-defeating: They are not just a form of unintended side effect that must be balanced against the main goal or value to be realized by an action, technology, or policy. Instead, they directly cause the primary goals of the actors—i.e., the goals that they ultimately pursue with the action, technology, or policy—to be “worse achieved” (Parfit). In this paper, we elaborate on this definition and distinguish three ideal-typical phases of adverse incentives, where only in the last one the threshold for a perverse incentive is crossed. In addition, we discuss different possible relevant actors and their goals in implementing autonomous vehicles. We conclude that even if some actors do not pursue traffic safety as their primary goal, as part of a responsibility network they incur the responsibility to act on the common primary goal of the network, which we argue to be traffic safety.
Similar content being viewed by others
Notes
We refer to partially autonomous vehicles as vehicles with an SAE automation level up to 4 (SAE International 2016). Fully autonomous driving vehicles, on the other hand, are equated with an SAE automation level of 5.
This is confident information from personal interviews. The sources have first-hand evidence of these mechanisms, but do not want their names to be disclosed. They are known to the authors. There exist no studies to date that systematically explore this particular phenomenon, although a variety of human factor surveys have been conducted so far (Tennant et al. 2016; Kyriakidis et al. 2017). In this paper, we are mainly interested in the possibility of such incentives and its ethical implications, and only secondarily concerned with the prevalence of the scenario.
Borenstein et al. (2017) list five different aspects that may contribute to over-trust: individual psychological dispositions, cultural peculiarities, positivity biases, the design and behavior of an autonomous system, and their consistency and predictability.
To our knowledge, there exists no specific philosophical literature on perverse incentives—with the exception of Parfit’s discussion of “directly self-defeating moral theories” (Parfit 1984, Part 1, esp. Sect. 2), which we will discuss shortly. We therefore draw on general descriptions of perverse incentives in the literature on risk assessment.
For better readability, in the following, we refer to “technology, action, or social policy” as “TAP,” while “implementing a technology, carrying out an action, or introducing a social policy” will be shortened to “actualizing a TAP.”
We are grateful to Hauke Behrendt for highlighting this point.
As we have said before, the primary goal does not have to be a goal for the participant adopting the detrimental—or even negating—behavior. Rather, it is prima facie only a goal for the actor actualizing the TAP. Therefore, it is per se not be irrational for the participant to adopt this kind of behavior.
While “phases” refer to a temporal succession, we do not claim that there is necessarily a path dependency in the sense that when phase one is initiated, phase three has to be reached eventually. Rather, we want to shed light on the fact that a perverse incentive oftentimes is a gradual as well as incremental phenomenon, which, when closely monitored, can be prevented in its early stages.
In this paper, we cannot engage in detail the objection that by introducing AVs we are “trading” lives, as “the identities of many (future) fatality victims would change” (Lin 2013). This might be impermissible, since by reducing the number of traffic fatalities we might be disregarding the fact that “suffering is not additive in this way” (Taurek 1977, p. 308). By way of a short answer, we agree with Patrick Lin that traffic safety is a statistical goal that is operationalized in accidents per driven road mile (NHTSA 2014). This means that there is no actual life at stake but rather the risk of losing one’s life in an accident. But, as Lin points out, no one has the moral right “not to be an accident victim” (Lin 2013).
Note that we are employing here a form of reason subjectivism. We are aware that Parfit himself subscribes to a form of objectivism with regard to reasons, what he calls “Reasons Fundamentalism.” For the notion of perverse incentives, however, it is important that by implementing a technology, the implementing actor negates her own conclusive reasons for implementing it, not what she has “most reason to do” (Scanlon 1998, pp. 33–34) from an objective standpoint.
Motorcycles present a special case, as the harm from traffic accidents mainly falls on the bikers themselves. Since liberal societies typically try to refrain from acting paternalistic, the rationale behind not banning motorcycles from the road can be tied to the fact that they on average harm themselves. However, this aspect does not touch on our general point here.
This seems to be the case, albeit not with regard to perverse incentives, when it comes to speed limits on the German Autobahn.
From the supply side, however, AVs will most likely also be marketed as increasing overall traffic safety, as, e.g., Tesla’s autopilot function is.
In addition, the idea of such networks may mitigate or even solve the “problem of many hands” (van de Poel et al. 2011), i.e., the problem of a failure to distribute responsibility adequately within a collective agent.
“Each of us” in this context refers to each instantiation of the kind of responsibility network that we have described here.
References
Aven, T. (2016). (2016). Risk assessment and risk management: Review of recent advances on their foundation. European Journal of Operational Research, 253, 1–13. https://doi.org/10.1016/j.ejor.2015.12.023.
Bonnefon, J.-F., Shariff, A., & Rahwan, I. (2016). The social dilemma of autonomous vehicles. Science, 352, 1573–1576. https://doi.org/10.1126/science.aaf2654.
Borenstein, J., Howard, A., & Wagner, A. (2017). Pediatric robotics and ethics: The robot is ready to see you now, but should it be trusted? In P. Lin, K. Abney, & R. Jenkins (Eds.), Robot ethics 2.0: New challenges in philosophy, law, and society (pp. 127–141). Oxford: Oxford Univ. Press.
Boudette, N. (2017). Tesla's self-driving system cleared in deadly crash. The new Yorck times. https://www.nytimes.com/2017/01/19/business/tesla-model-s-autopilot-fatal-crash.html?_r=0. Accessed 27 February 2017.
Bratman, M. (1999). Faces of Intention: Selected Essays on Intention and Agency. Cambridge: Cambridge Univ. Press.
Dennett, D. (1989). The intentional stance. Cambridge Mass: MIT Press.
ERTRAC. (2017). Automated Driving Roadmap: Version 7.0. http://www.ertrac.org/uploads/documentsearch/id48/ERTRAC_Automated_Driving_2017.pdf. Accessed 25 November 2017.
Frewer, L. J., Howard, C., & Shepherd, R. (1997). Public concerns in the United Kingdom about general and specific applications of genetic engineering: Risk, benefit, and ethics. Science Technology Hum n Values, 22, 98–124. https://doi.org/10.1177/016224399702200105.
Gibson, J. J. (1986). The ecological approach to visual perception. New York: Psychology Press.
Gilbert, M. (1990). Walking together: A paradigmatic social phenomenon. Midwest Studies in Philosophy, 15, 1–14.
Goddard, T., Dill, J., & Monsere, C. (2016). Driver Attitudes About Bicyclists: Negative Evaluations of Rule-Following and Predictability. Transportation Research Board 95th Annual Meeting.
Golson, J. (2017). Driver in fatal tesla autopilot crash had seven seconds to take action. The Verge. http://www.theverge.com/2017/1/19/14326604/tesla-autopilot-crash-driver-seven-seconds-inattentive-nhtsa. Accessed 23 March 2017.
Goodall, N. (2014). Machine Ethics and automated vehicles. In G. Meyer & S. Beiker (Eds.), Road vehicle automation (pp. 93–102, Lecture Notes in Mobility). Cham: Springer.
Goodrich, M. A., & Schultz, A. C. (2007). Human-robot interaction: A survey. Foundations and Trends in Human-Computer Interaction, 1, 203–275. https://doi.org/10.1561/1100000005.
Gosepath, S. (1992). Aufgeklärtes Eigeninteresse: Eine Theorie theoretischer und praktischer Rationalität. Frankfurt/Main: Suhrkamp.
Habermas, J. (1994). Three normative models of democracy. Constellations, 1(1), 1–10.
Hancock, P. (2014). (2014). Automation: How much is too much? Ergonomics, 57, 449–454. https://doi.org/10.1080/00140139.2013.816375.
Hansson, S. O. (2013). The ethics of risk: Ethical analysis in an uncertain world. New York: Palgrave Macmillan.
Jennings, N., Sycara, K., & Woolridge, M. (1998). A roadmap of agent Research and Development. Autonomous Agents and Multi-Agent Systems, 1, 7–38.
Kirkpatrick, J., Hahn, E., & Haufler, A. (2017). Trust and human-robot interactions. In P. Lin, K. Abney, & R. Jenkins (Eds.), Robot ethics 2.0: New challenges in philosophy, law, and society (pp. 142–156). Oxford: Oxford Univ. Press.
Kyriakidis, M., Happee, R., & de Winter, J. C. F. (2015). Public opinion on automated driving: Results of an international questionnaire among 5000 respondents. Transportation Research Part F: Traffic Psychology and Behaviour, 32, 127–140. https://doi.org/10.1016/j.trf.2015.04.014.
Kyriakidis, M., de Winter, J. C. F., Stanton, N., Bellet, T., van Arem, B., Brookhuis, K., et al. (2017). A human factors perspective on automated driving. Theoretical Issues in Ergonomics Science, 53, 1–27. https://doi.org/10.1080/1463922X.2017.1293187.
Levin, S., & Woolf, N. (2016, July 1). Tesla driver killed while using autopilot was watching Harry potter, witness says. The Guardian. https://www.theguardian.com/technology/2016/jul/01/tesla-driver-killed-autopilot-self-driving-car-harry-potter. Accessed 2 June 2018.
Lin, P. (2013). The ethics of saving lives with autonomous cars is far murkier than you think. Wired. https://www.wired.com/2013/07/the-surprising-ethics-of-robot-cars/. Accessed 16 August 2017.
Lin, P. (2014, August 18). Here’s a Terrible Idea: Robot Cars with Adjustable Ethics Settings. Wired. http://www.wired.com/2014/ 8/heres-a-terrible-idea-robotcars-with-adjustable-ethics-settings/. Accessed 30 November 2017.
Lin, P. (2016). Is tesla responsible for the deadly crash on auto-pilot? Maybe. Forbes. https://www.forbes.com/sites/patricklin/2016/07/01/is-tesla-responsible-for-the-deadly-crash-on-auto-pilot-maybe/#23ec768b1c07. Accessed 27 February 2017.
Lin, P., Abney, K., & Jenkins, R. (Eds.). (2017). Robot Ethics 2.0: New Challenges in Philosophy, Law, and Society. Oxford: Oxford Univ. Press.
List, C., & Pettit, P. (2011). Group agency: The possibility, design, and status of corporate agents. Oxford: Oxford Univ. Press.
Loh, J., & Loh, W. (2017). Autonomy and responsibility in hybrid systems: The example of autonomous cars. In P. Lin, K. Abney, & R. Jenkins (Eds.), Robot ethics 2.0: New challenges in philosophy, law, and society. Oxford UK: Oxford Univ. Press.
Madigan, R., Louw, T., Dziennus, M., Graindorge, T., Ortega, E., Graindorge, M., et al. (2016). Acceptance of automated road transport systems (ARTS): An adaptation of the UTAUT model. Transportation Research Procedia, 14, 2217–2226. https://doi.org/10.1016/j.trpro.2016.05.237.
Merton, R. (1936). The unanticipated consequences of purposive social action. American Sociological Review, 1(6), 894–904.
Meyer, G., & Beiker, S. (Eds.). (2016). Road Vehicle Automation 3, Lecture notes in mobility. Cham: Springer International Publishing; Imprint; Springer.
Misselhorn, C. (2013). Robots as Moral Agents. In F. Rövekamp & B. Friederike (Eds.), Ethics in Science and Society: German and Japanese Views (pp. 30–42). München: Iudicum.
Misselhorn, C. (2015). Collective agency and cooperation in natural and artificial systems. In C. Misselhorn (Ed.), Collective agency and cooperation in natural and artificial systems: Explanation, implementation and simulation (pp. 3–24). London: Springer.
Neuhäuser, C. (2015). Some Sceptical remarks regarding robot responsibility and a way forward. In C. Misselhorn (Ed.), Collective agency and cooperation in natural and artificial systems: Explanation, implementation and simulation (pp. 131–146). London: Springer.
NHTSA. (2014). Fatalities by State and Road Function Class. https://www-fars.nhtsa.dot.gov/States/StatesCrashesAndAllVictims.aspx. Accessed 23 April 2017.
Norcross, A. (1998). Great harms from small benefits grow: How death can be outweighed by headaches. Analysis, 58(2), 152–158.
Norman, D. A. (1988). The psychology of everyday things. New York, NY: Basic Books.
NTMVSA. (1966). National Traffic and motor vehicle safety act: Public law 89–563.
Nyholm, S. (2017). Attributing agency to automated systems: Reflections on human–robot collaborations and responsibility-loci. Science and Engineering Ethics.
Nyholm, S., & Smids, J. (2016). The ethics of accident-algorithms for self-driving cars: An applied trolley problem? Ethical Theory and Moral Practice, 19, 1275–1289n.d.. https://doi.org/10.1007/s10677-016-9745-2.
Nyholm, S., & Smids, J. (2018). Automated cars meet human drivers: Responsible human-robot coordination and the ethics of mixed traffic. Ethics and Information Technology, 9, 332. https://doi.org/10.1007/s10676-018-9445-9.
Parasuraman, R., & Riley, V. (1997). Humans and automation: Use, misuse, disuse, abuse. Human Factors, 39, 230–253. https://doi.org/10.1518/001872097778543886.
Parfit, D. (1984). Reasons and persons. Oxford: Oxford Univ. Press.
Parfit, D. (2011). On what matters. Oxford: Oxford University Press.
Peden, M., Scurfield, R., Sleet, D., Mohan, D., Hyder, A., Jarawan, E., et al. (2004). World report on road traffic injury prevention. Geneva. http://cdrwww.who.int/violence_injury_prevention/publications/road_traffic/world_report/intro.pdf. Accessed 30 November 2017.
Rayna, T., Striukova, L., & Landau, S. (2009). Crossing the chasm or being crossed out: The case of digital audio players. International Journal of Actor-Network Theory and Technological Innovation, 1(3), 36–54.
Raz, J. (2005). The myth of instrumental rationality. Journal of Ethics and Social Philosophy, 1, 1–28. https://doi.org/10.26556/jesp.v1i1.1.
Renn, O. (2008). Risk governance: Coping with uncertainty in a complex world. London u.A.: Earthscan.
Robinette, P., Li, W., Allen, R., Howard, A. M., & Wagner, A. R. (2016). Overtrust of Robots in Emergency Evacuation Scenarios (pp. 101–108). Piscataway NJ: IEEE Press.
Roeser, S., Hillerbrand, R., Sandin, P., & Peterson, M. (Eds.). (2012). Handbook of risk theory. Dordrecht: Springer Netherlands.
Rogers, E. M. (1962). Diffusion of innovations. New York: Free Press of Glencoe.
SAE International. (2016). Taxonomy and definitions for terms related to driving automation systems for on-road motor vehicles. Washington DC: SAE International.
Santoni de Sio, F. (2016). Ethics and self-driving cars: A white paper on responsible innovation in automated driving systems. Dutch Ministry of Infrastructure and Environment Rijkswaterstaat.
Scanlon, T. (1998). What we owe to each other. Cambridge MA: Belknap Press of Harvard Univ. Press.
Scanlon, T. (2007). Structural irrationality. In G. Brennan, R. Goodin, F. Jackson, & M. Smith (Eds.), Common minds: Themes from the philosophy of Philip Pettit (pp. 84–103). Oxford: Clarendon.
Searle, J. (1990). Collective intentions and actions. In P. Cohen, J. Morgan, & M. Pollack (Eds.), Intentions in communication (pp. 401–415). Cambridge MA: MIT Press.
Siebert, H. (2001). Der Kobra-Effekt: Wie man Irrwege der Wirtschaftspolitik vermeidet. München: Deutsche Verlags-Anstalt.
Smith, M. (1994). The moral problem, Philosophical theory. Oxford: Blackwell.
Stilgoe, J. (2017). Tesla crash report blames human error - this is a missed opportunity. The Guardian. https://www.theguardian.com/science/political-science/2017/jan/21/tesla-crash-report-blames-human-error-this-is-a-missed-opportunity. Accessed 23 March 2017.
Taebi, B. (2017). Bridging the gap between social acceptance and ethical acceptability. Risk Analysis, 37, 1817–1827. https://doi.org/10.1111/risa.12734.
Taurek, J. (1977). Should the numbers count? Philosophy and Public Affairs, 6(4), 293–316.
Tennant, C., Howard, S., Franks, B., Bauer, M., Stares, S., Pansegrau, P., et al. (2016). Autnonomous Vehicles - Negotiating a Place on the Road: A study on how drivers feel about Interacting with Autonomous Vehicles on the road. http://www.lse.ac.uk/website-archive/newsAndMedia/PDF/AVs-negociating-a-place-on-the-road-1110.pdf. Accessed 16 August 2017.
Tesla Motors. (2016). A Tragic Loss. https://www.tesla.com/blog/tragic-loss. Accessed 23 March 2017.
van de Poel, I., Fahlquist, J. N., Doorn, N., Zwart, S., & Royakkers, L. (2011). The problem of many hands: Climate change as an example. Science and Engineering Ethics, 18(1), 49–67.
Voorhoeve, A. (2014). How should we aggregate competing claims? Ethics, 125(1), 64–87.
Wallach, W., & Allen, C. (2009). Moral Machines: Teaching Robots Right from Wrong. Oxford, New York: Oxford University Press.
Weber, M. (1978). Economy and Society. Berkeley CA: University of California Press.
Winter, J., Happee, R., Martens, M. H., & Stanton, N. A. (2014). Effects of adaptive cruise control and highly automated driving on workload and situation awareness: A review of the empirical evidence. Transportation Research Part F: Traffic Psychology and Behaviour, 27, 196–217. https://doi.org/10.1016/j.trf.2014.06.016.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Loh, W., Misselhorn, C. Autonomous Driving and Perverse Incentives. Philos. Technol. 32, 575–590 (2019). https://doi.org/10.1007/s13347-018-0322-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13347-018-0322-6