Artificial intelligence (AI) ethics is proposed as an emerging and interdisciplinary field concerned with addressing the ethical issues of AI, such as the issue of moral decision-making. The conflict between our intuitive moral judgments constitutes an inevitable obstacle to decision-making in AI ethics. This article outlines the Moral Relevance Approach, which could provide a considerable moral foundation for AI ethics. Taking moral relevance as the precondition of the consequentialist principles, the Moral Relevance Approach aims to plausibly consider individual moral claims. (...) It is not only the common ethical target shaping our moral consensus but also the inherent moral ability connecting others with us. (shrink)

In 'Rethinking the Good', Larry Temkin makes two core claims. First, the goodness of a distribution is sometimes ‘essentially comparative’ – it sometimes depends on which alternative distribution(s) it is compared to. Second, such cases threaten the transitivity of ‘all things considered better than’. I argue that the goodness of a distribution may indeed depend on what other distributions are feasible. But contrary to Temkin, I also argue that transitivity holds even when the goodness of a distribution depends on the (...) feasible set of which it is a member. (shrink)

When one faces competing claims of varying strength on public resources for health, which claims count? This paper proposes the following answer. One should count, or aggregate, a person’s claim just in case one could sympathize with her desire to prioritize her own claim over the strongest competing claim. It argues that this principle yields appealing case judgments and has a plausible grounding in both sympathetic identification with each person, taken separately, and respect for the person for whom most is (...) at stake. It also defends this principle against several heretofore unanswered objections. (shrink)

Tännsjö’s book Setting Health-Care Priorities defends the view that there are three main normative theories in the domain of distributive justice, and that these theories are both highly plausible in themselves, and practically convergent in their normative conclusions. All three theories point to a somewhat radical departure from the present distribution of medical resources: in particular, they suggest redirecting resources from marginal life extension to the care of mentally ill patients. In this paper I wish to argue, firstly, that prioritarianism (...) should not be considered as an amendment to utilitarianism, as it is in Tännsjö’s view, but as a distinctive fourth option. This can best be appreciated if we focus on a reading of the theory that emphasizes its derivation from egalitarianism and its attempt to develop an intermediate approach between utilitarian and egalitarian intuitions. Secondly, in response to Tännsjö’s central objection to prioritarianism, I will argue that the theory does not apply in intrapersonal cases but is only relevant for decisions regarding the interpersonal distribution of benefits. Finally, I will suggest that a practical convergence of the four theories on specific issues such as artificial reproduction or mood enhancement is far less likely than Tännsjö seems to believe. (shrink)

This paper presents a new kind of problem in the ethics of distribution. The problem takes the form of several “calibration dilemmas,” in which intuitively reasonable aversion to small-stakes inequalities requires leading theories of distribution to recommend intuitively unreasonable aversion to large-stakes inequalities. We first lay out a series of such dilemmas for prioritarian theories. We then consider a widely endorsed family of egalitarian views and show that they are subject to even more forceful calibration dilemmas than prioritarian theories. Finally, (...) we show that our results challenge common utilitarian accounts of the badness of inequalities in resources. (shrink)

Many of us believe (1) Saving a life is more important than averting any number of headaches. But what about risky cases? Surely: (2) In a single choice, if the risk of death is low enough, and the number of headaches at stake high enough, one should avert the headaches rather than avert the risk of death. And yet, if we will face enough iterations of cases like that in (2), in the long run some of those small risks of (...) serious harms will surely eventuate. And yet: (3) Isn't it still permissible for us to run these repeated risks, despite that knowledge? After all, if it were not, then many of the risky activities that we standardly think permissible would in fact be impermissible. Nobody has yet offered a principle that can accommodate all of 1-3. In this paper, I show that we can accommodate all of these judgements, by taking into account both ex ante and ex post perspectives. In doing so, I clear aside an important obstacle to a viable deontological decision theory. (shrink)

One widely used method for allocating health care resources involves the use of cost-effectiveness analysis (CEA) to rank treatments in terms of quality-adjusted life-years (QALYs) gained. CEA has been criticized for discriminating against people with disabilities by valuing their lives less than those of non-disabled people. Avoiding discrimination seems to lead to the ’QALY trap’: we cannot value saving lives equally and still value raising quality of life. This paper reviews existing responses to the QALY trap and argues that all (...) are problematic. Instead, we argue that adopting a moderate form of prioritarianism avoids the QALY trap and disability discrimination. (shrink)

This paper introduces a condition for rational choice that states that accepting decision methods and normative theories that sometimes entail that the act of choosing a maximal alternative renders this alternative non-maximal is irrational. The paper illustrates how certain distributive theories that ascribe importance to what the status quo is violate this condition and argues that they thereby should be rejected.

It is a common view that benefits to the worse off should be given priority when health benefits are distributed. This paper addresses how to understand who is worse off in this context when individuals are differently well off at different times. The paper argues that the view that this judgment about who is worse off should be based solely on how well off individuals are when their complete lives are considered (i.e. 'the complete lives view') is implausible in this (...) context. Instead, it is argued that a pluralistic stance toward this issue should be accepted. This pluralistic stance recognizes that also the view that only focuses on how well off individuals are now and in the future (i.e. 'the forward-looking view') is relevant. The argument is based on appeals to intuitive judgments concerning who is worse off in different cases and reference to various underlying reasons why priority to benefits to the worse off is justified. (shrink)

Principles of need are constantly referred to in health care priority setting. The common denominator for any principle of need is that it will ascribe some kind of special normative weight to people being worse off. However, this common ground does not answer the question how a plausible principle of need should relate to the aggregation of benefits across individuals. Principles of need are sometimes stated as being incompatible with aggregation and sometimes characterized as accepting aggregation in much the same (...) way as utilitarians do. In this paper we argue that if one wants to take principles of need seriously both of these positions have unreasonable implications. We then characterize and defend a principle of need consisting of sufficientarian elements as well as prioritarian which avoids these unreasonable implications. (shrink)

Distributive theories evaluate distributions of goods based on candidate recipients’ characteristics, e.g. how well off candidates are, how deserving they are, and whether they fare below sufficiency. But such characteristics vary across possible worlds, so distributive theories may differ in terms of the world which for them settles candidates’ characteristics. This paper examines how distributive theories differ in terms of whether candidate recipients’ relevant characteristics are grounded in the possible world that would take place if the distributor does not intervene (...) (call it the “input” world) or if they are grounded in each possible world that the distributor can bring about through different decisions (call each an “output” world). We illustrate the importance of this distinction in relation to one distributive theory, prioritarianism. As we show, both input and output versions of prioritarianism are plausible interpretations of the literature, and there are good reasons to take input prioritarianism seriously. Ultimately, however, we argue that input prioritarianism should be rejected. (shrink)

Should we allow grave harm to befall one individual so as to prevent minor harms befalling sufficiently many other individuals? This is a question of aggregation. Can many small harms ‘add up’, so that, collectively, they morally outweigh a greater harm? The ‘Close Enough View’ supports a moderate position: aggregation is permissible when, and only when, the conflicting harms are sufficiently similar, or ‘close enough’, to each other. This paper surveys a range of formally precise interpretations of this view, and (...) reveals some of the problems they face. It also proposes a novel interpretation which avoids these problems. (shrink)

Entre les états (2, 3) et (2, 4), le Principe de Différence (PD) choisit-t-il le premier, le second, ou est-il indifférent? Cette dernière interprétation est confortée par le choix par Rawls des courbes en L pour illustrer le PD (Rawls 1999 [1971], § 13, fig. 5), formellement analogue au Maximin en théorie de la décision, et qui admettrait des courbes d’indifférence ; une fois le plus mal loti maximisé, on est indifférent entre les états des mieux lotis ; le Leximin (...) enjoint lui de choisir le second état, puisqu’il réitère le PD sur le dernier plus mal loti. La plupart des commentateurs choisissent le Leximin ; or ce principe est inégalitaire, car il peut laisser s’enrichir les riches alors que les pauvres stagnent. Le PD sélectionne en fait le premier état. Si l’on peut passer du premier état au second, on doit pouvoir passer du premier à un état (2+n, 3-n). Le PD est univoque, différent du Maximin et du Leximin, qui peut favoriser des sociétés duales, où seuls les riches s’enrichissent. On n’a pu généraliser le résultat à plus de deux classes, même en tenant compte de la « connexion en chaîne », laquelle exprime l’idée fondamentale de Rawls : la solidarité via la coopération.Code JEL : I13. (shrink)

This thesis discusses the possibility of limited moral trade-offs between different people’s welfare. In chapter 2, I introduce the two central limited trade-off conditions. First, according to minimal infinite superiority, significantly benefiting one person matters more than slightly benefiting each of any number of better-off people. Second, according to minimal finite superiority, significantly benefiting many people matters more than slightly benefiting one person. I consider both axiological and deontic interpretations of these conditions. However, I explain why none of the simple (...) classic moral principles—the simple total and the maximin principles—satisfy both conditions. Furthermore, in chapter 3, I strengthen several proved impossibility results according to which no moral theory satisfies weak interpretations of these central trade-off conditions and several other seemingly plausible minimal conditions. I show that giving up structural axiological and deontic conditions is not a satisfactory solution to these paradoxes. In chapter 4, I discuss the modification of a background assumption of these impossibility results on the measurement of welfare. I show that, given a modification that allows for lexicographically ordered welfare components, a total principle can satisfy all the conditions of the impossibility results. However, I argue that such a modification is not entirely satisfactory because it does not apply in certain instances of the paradoxes. In chapter 5, I discuss a further weakening of minimal infinite superiority. However, I show that a suggested possibility result based on this modification is not valid and that further moral conditions, such as minimal finite superiority, need to be modified. Moreover, I argue that these modified conditions and the principle suggested in the possibility proof—a minimax complaint principle—do not capture the basic idea of the two central limited trade-off conditions sufficiently well. In chapter 6, I argue that other principles suggested for the task—the total claim principles—share the same fate as the simple total principles or the minimax complaint principle. In chapter 7, I propose new principles that take their structural roots from voting theory. I show that, in contrast to the other principles discussed, these principles give plausible verdicts where the other principles failed. Finally, in chapter 8, I consider possible objections levelled against this proposal, and I suggest solutions and avenues for future research. (shrink)