Abstract
Many seem to think that AI-induced responsibility gaps are morally bad and therefore ought to be avoided. We argue, by contrast, that there is at least a pro tanto reason to welcome responsibility gaps. The central reason is that it can be bad for people to be responsible for wrongdoing. This, we argue, gives us one reason to prefer automated decision-making over human decision-making, especially in contexts where the risks of wrongdoing are high. While we are not the first to suggest that responsibility gaps should sometimes be welcomed, our argument is novel. Others have argued that responsibility gaps should sometimes be welcomed because they can reduce or eliminate the psychological burdens caused by tragic moral choice-situations. By contrast, our argument explains why responsibility gaps should sometimes be welcomed even in the absence of tragic moral choice-situations, and even in the absence of psychological burdens.
Similar content being viewed by others
Notes
Throughout the paper, we shall talk about ‘AI systems’, but we take this to include things like simple rule-based systems, machine learning systems, deep learning systems, etc.
A burgeoning literature discusses whether artificial, non-human agents can be held responsible under certain conditions (see for instance Sebastián 2021; List 2021). We set this discussion aside here, since even if automatons could aptly be held responsible, this would change nothing from the perspective of our argument.
See (Goetze, 2022) for discussion of the tracing back strategy.
See (Simpson and Müller 2016) for discussion of this response.
It has been argued that AI systems that make “social decisions” like the one in Decision-Procedure Designer are often extremely erroneous (Raji et al., 2022). We are not concerned with extant AI systems, but with AI systems that work as described above. We should also mention that many have argued that there are excellent reasons not to replace human decision-makers with AI systems for reasons unrelated to responsibility gaps. Finally, we should mention that it has been argued that AI systems are often no better than basic statistical techniques (Narayanan, 2019), making it less clear why we should be particularly concerned with replacing human decision-makers with AI systems per se. However, we shall set such worries aside for the sake of argument. We thank an anonymous reviewer for suggesting that we make these things explicit. observations and limpoint this out.
Of course, the system will also distribute undeserved benefits to some. But since it is harder to see that such cases are wrongings of any particular individuals, we will focus on the other type of errors here.
There are several different reasons for why it can be important to know who is responsible for erroneous decisions. First, it can be important because we sometimes need to know who to punish for erroneous decisions. Second, it can be important because it can help us avoid erroneous decisions in the future. Third, it can be important when we need to provide redress for the “victims” of erroneous decisions (Goetze, 2022); (Gotterbarn, 2001); (Nissenbaum, 1994). Fourth, it can be important when we need to provide explanations to victims of why errors were made (Coeckelbergh, 2021). Thanks to an anonymous reviewer for suggesting that we highlight these different reasons.
Himmelreich only intends this to pick out one type of responsibility gap, notice, so it is not meant as a complete account. See (Hindriks and Veluwenkamp 2023) for discussion of Himmelreich’s account and other ways of conceptualizing responsibility gaps. Hindriks and Veluwenkamp are skeptical of there being responsibility gaps, arguing that in the relevant range of cases, responsibility is always indirect or there is blameless harm (so there is no room for “gaps” in responsibility). As stated before, we remain neutral on the question of whether responsibility gaps exist, but notice that even if Hindriks and Veluwenkamp are right, our argument speaks to the desirability of cases of blameless harm.
There is a substantive question here about how “thick” the judgment that somebody is morally responsible for some outcome is. On the probably thinnest possible interpretation, A is morally responsible for some outcome O when A caused O. On a thicker notion, such as the one employed by Santoni de Sio and Mecassi (2021: 1062), responsibility for an outcome tracks blameworthiness (provided the outcome is one that warrants blame)—this they call “culpability”.
For our purposes we can understand the idea of being “held responsible” broadly. It may include activities such as blaming, punishing or harming.
This case is inspired by a similar case from (Tadros 2020).
Thanks to an anonymous reviewer for pressing us to discuss this objection.
See (Tessman, 2017) for discussion of moral dilemmas.
We thank an anonymous reviewer for asking us to elaborate on this point.
References
Alexander, L., & Ferzan, K. (2018). Reflections on crime and culpability: problems and puzzles. Cambridge: Cambridge University Press.
Baum, K., Mantel, S., & Schmidt, E., and Timo Speith (2022). From responsibility to reason-giving explainable Artificial Intelligence. Philosophy & Technology, 35(1), 12.
Bjerring, J. C., & Busch, J. (2021). Artificial Intelligence and patient-centered decision-making. Philosophy & Technology, 34, 349–371.
Coeckelbergh, M. (2021). AI Ethics. MIT Press.
Danaher, J. (2016). Robots, Law and the Retribution gap. Ethics and Information Technology, 18(4), 299–309.
Danaher, J. Tragic Choices and the Virtue of Techno-Responsibility Gaps. Philosophy and Technology, forthcoming.
De Jong, R. (2020). The retribution-gap and responsibility-loci related to robots and automated technologies: a reply to Nyholm. Science and Engineering Ethics, 26(2), 727–735.
Dworkin, G. (2020). Paternalism. In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, https://plato.stanford.edu/archives/fall2020/entries/paternalism/.
Feier, T., Gogoll, J., & Uhl, M. (2022). Hiding Behind Machines: Artificial Agents May Help to Evade Punishment.Science and Engineering Ethics,28.
Felder, R. (2021). Coming to terms with the Black Box Problem: how to justify AI Systems in Health Care. Hastings Center Report, 51(4), 38–45.
Fischer, J., & Tognazzini, N. A. (2009).The Truth about Tracing. Noûs, 43(3):531–56.
Goetze, T. (2022). Mind the Gap: Autonomous Systems, the Responsibility Gap, and Moral Entanglement. FAccT ‘22
Gotterbarn, D. (2001). Informatics and professional responsibility. Science and Engineering Ethics, 7, 221–230.
Hanson, F. A. (2009). Beyond the skin bag: on the moral responsibility of extended agencies. Ethics and Information Technology, 11(1), 91–99.
Himmelreich, J. (2019). Responsibility for Killer Robots. Ethical Theory and Moral Practice, 22(3), 731–747.
Hindriks, F., & Veluwenkamp, H. (2023). The risks of autonomous machines: from responsibility gaps to control gaps. Synthese, 201, 21.
Kraaijeveld, S. R. (2020). Debunking (the) retribution (gap). Science and Engineering Ethics, 26, 1315–1328.
Kiener, M. (2022). Can we Bridge AI’s responsibility gap at Will? Ethical Theory and Moral Practice, 25, 575–593.
Königs, P. (2022). Artificial intelligence and responsibility gaps: what is the problem?Ethics and Information Technology, 24(36).
Langer, M., Cornelius, J., & König, and Andromachi Fitili (2018). Information as a double-edged Sword: the role of computer experience and information on applicant reactions towards Novel Technologies for Personnel Selection. Computers in Human Behavior, 81, 19–30.
Levinson, J., Askeland, J., Becker, J., Dolson, J., Held, D., Kammel, S., Zico, J., Kolter (2011). Towards Fully Autonomous Driving: Systems and Algorithms. IEEE Intelligent Vehicles Symposium (IV), 163–68.
List, C. (2021). Group Agency and Artificial Intelligence. Philosophy & Technology, 34, 1213–1242.
Matthias, A. (2004). The responsibility gap: ascribing responsibility for the actions of learning automata. Ethics and Information Technology, 6, 175–183.
Narayanan, A. (2019). How to Recognize AI Snake Oil. Arthur Miller lecture on science and ethics, Massachusetts Institute of Technology, http://www.cs.princeton.edu/~arvindn/talks.
Nissenbaum, H. (1994). Computing and accountability. Communications of the ACM, 37(1), 72–80.
Pagallo, U. (2011). Killers, fridges, and slaves: a legal journey in robotics. AI & Society, 26, 347–354.
Rubel, A., Castro, C., & Pham, A. (2019). Agency laundering and Information Technologies. Ethical Theory and Moral Practice, 22(4), 1017–1041.
Raji, I., Elizabeth Kumar, A., Horowitz, & Selbst, A. (2022). The Fallacy of AI Functionality. FAccT ‘22.
Santoni de Sio, F., & Mecacci, G. Four Responsibility Gaps with Artificial Intelligence: Why they Matter and How to Address them.Philosophy & Technology, 34:1057–1084.
Sebastián, M. (2021). First-person representations and responsible Agency in AI. Synthese, 199(3), 7061–7079.
Simpson, T., Vincent, C., & Müller. Just war and robot’s killings.The Philosophical Quarterly, 66(263):302–22.
Sparrow, R. (2007). Killer Robots. Journal of Applied Philosophy, 24(1), 62–77.
Tadros, V. (2020). Distributing responsibility. Philosophy & Public Affairs, 48(3), 223–261.
Tadros, V. (2011). The ends of harm: the Moral Foundations of Criminal Law. Oxford: Oxford University Press.
Tessman, L. (2017). When doing the right thing is impossible. Oxford University Press.
Tigard, D. (2021). There is no techno-responsibility gap. Philosophy & Technology, 34, 589–607.
Topol, E. (2019). High-performance medicine: the convergence of human and Artificial Intelligence. Nature Medicine, 25(1), 44–56.
Walen, A. (2021). Retributive Justice, The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed.), https://plato.stanford.edu/archives/sum2021/entries/justice-retributive/.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Munch, L., Mainz, J. & Bjerring, J.C. The value of responsibility gaps in algorithmic decision-making. Ethics Inf Technol 25, 21 (2023). https://doi.org/10.1007/s10676-023-09699-6
Published:
DOI: https://doi.org/10.1007/s10676-023-09699-6