Michelle M. Dyke

When an AI algorithm like ChatGPT yields false information, do we blame the algorithm for its error? Presumably not. What can we learn from this? There are interesting ramifications for our understanding of blame. One related and ongoing debate concerns whether we engage in a practice of directing “epistemic blame” towards other people. To help make progress on this question, I focus here on two influential general theories of blame, including Scanlon’s (2008) relationship modification theory and Sher’s (2006) belief-desire account. I argue that our response to false information from AI satisfies the criteria for blame according to both theories. If we are confident that we do not genuinely blame AI for false information, then we must revise our understanding of what any kind of blame, whether moral or epistemic, generally involves. Only then can we settle the questions of whether there is a distinctive practice of epistemic blame, and of who, or what, might be an appropriate target of it.

This chapter provides an opinionated introduction to the view of moral relativism. The chapter first presents two of the philosophical motivations for endorsing relativism, which stem from a desire to respect the possibility of legitimate moral diversity, and from the separate goal of offering an account of moral normativity that is compatible with naturalism. The chapter also addresses three potential objections to relativism. The first is the concern that the moral relativist must be committed to the absurd view that everything is permitted. The second is the suggestion that relativism is a self-undermining philosophical position. The third is the charge that the relativist will be unable account for the genuine normative authority of moral requirements. The author argues that all three potential objections can be successfully overcome by the defender of a carefully formulated version of moral relativism.

In this paper, I draw attention to the phenomenon of warranted epistemic blame in order to pose a challenge for most forms of epistemic instrumentalism, which is the view that all of the demands of epistemic normativity are requirements of instrumental rationality. Because of the way in which the instrumentalist takes the force of one’s epistemic reasons to derive from one’s own individually held ends, the instrumentalist faces unique difficulties in explaining our standing to blame one another for violations of epistemic norms. In many cases, it is unclear why, according to the instrumentalist, we might be entitled to others’ adherence to epistemic norms at all. This is a serious problem. The upshot is that theorists of epistemic normativity should be prepared reject most forms of epistemic instrumentalism.

Could an entire society count as an agent in its own right? I argue here that it could. While previous defenders of group agency have focused primarily on groups such as states and corporations that exhibit a great deal of formalized internal structure, less attention has been devoted to more loosely structured social groups. I focus on defending the claims that societies can have ends or goals and that they engage in end-directed behavior. I defend this view by responding to three potential objections. The first is the allegation that the attribution of ends to societies would be redundant, given that the properties of any group will supervene upon the properties of the group’s individual human members. The second is the charge that societies are not genuine agents engaged in end-directed actions because their behavior is not under their own direct control. The third is the concern that while it may be useful to speak of societies ‘as if’ they were agents, this does not indicate that societies really act in pursuit of ends. I draw upon a functionalist approach from the philosophy of mind to argue that all three objections can be addressed by the defender of social agency.

The author argues that well-known forms of relativism are unable to accommodate, at once, a set of three highly intuitive theses about the distinctive character of moral reasons. Yet the author argues it is possible to formulate a novel form of normative relativism that has the power to accommodate these claims. The proposed view combines the relativist idea that the normative facts are attitude-dependent with the insight that there are non-human agents to which it makes sense to attribute the kinds of attitudes that give rise to normative reasons. Societies, too, can possess reasons to pursue their aims. What distinguishes moral reasons from reasons of practical rationality is that the former apply directly to societies in virtue of aims held by each society as a group, while the latter apply directly to persons in light of their own individual interests.

Are epistemic reasons merely a species of instrumental practical reasons, making epistemic rationality a specialized form of instrumental practical rationality? Or are epistemic reasons importantly different in kind? Despite the attractions of the former view, Kelly (2003) argues quite compellingly that epistemic rationality cannot be merely a matter of taking effective means to one’s epistemic ends. I argue here that Kelly’s objections can be sidestepped if we understand epistemic reasons as instrumental reasons that arise in light of the aims held by social collectives of which we are members, rather than being fixed by our own individual goals. This social version of epistemic instrumentalism would not be subject to counterexamples that point to the failure of individual people to possess desires or goals that would account for all of the epistemic reasons we find it natural to attribute to them. I conclude by comparing the proposed view to the alternative version of instrumentalism defended by Kornblith (1993). I argue that the social view I sketch here has one noteworthy advantage. It better accounts for the intuitive distinctness of our practical and epistemic reasons for belief in cases where flouting epistemic norms would better help us to achieve our own individual goals.

Street’s “Darwinian Dilemma” is a well-known epistemological objection to moral realism. In this paper, I argue that “third-factor” replies to this argument on behalf of the moral realist, as popularized by Enoch :413–438, 2010, Taking morality seriously: a defense of robust realism, Oxford University Press, Oxford, 2011), Skarsaune :229–243, 2011) and Wielenberg :441–464, 2010, Robust ethics: the metaphysics and epistemology of godless normative realism, Oxford University Press, Oxford, 2014), cannot succeed. This is because they are instances of the illegitimate form of reasoning known as “bootstrapping.” The phenomenon of bootstrapping has been discussed in detail, most notably by Vogel :602–623, 2000) and Cohen :309–329, 2002), in a different context as an objection to reliabilism and related theories of knowledge. I introduce four different characterizations of the error of bootstrapping from the epistemic literature in order to argue that the form of reasoning exemplified by third-factor replies would be deemed illegitimate by every one of them. I conclude that the moral realist should abandon third-factor replies, or else suggest a novel diagnosis for what goes wrong in bootstrapping cases that does not apply equally to the realist’s form of argument. However, I am not optimistic about the prospects for this latter strategy.