Mary Carman

For synergistic interactions between humans and artificial intelligence (AI) systems, AI outputs often need to be explainable to people. Explainable AI (XAI) systems are commonly tested in human user studies. However, whether XAI researchers consider potential cultural differences in human explanatory needs remains unexplored. We highlight psychological research that found significant differences in human explanations between many people from Western, commonly individualist countries and people from non-Western, often collectivist countries. We argue that XAI research currently overlooks these variations and that many popular XAI designs implicitly and problematically assume that Western explanatory needs are shared cross-culturally. Additionally, we systematically reviewed over 200 XAI user studies and found that most studies did not consider relevant cultural variations, sampled only Western populations, but drew conclusions about human-XAI interactions more generally. We also analyzed over 30 literature reviews of XAI studies. Most reviews did not mention cultural differences in explanatory needs or flag overly broad cross-cultural extrapolations of XAI user study results. Combined, our analyses provide evidence of a cultural bias toward Western populations in XAI research, highlighting an important knowledge gap regarding how culturally diverse users may respond to widely used XAI systems that future work can and should address.

Developing and implementing artificial intelligence (AI) systems in an ethical manner faces several challenges specific to the kind of technology at hand, including ensuring that decision-making systems making use of machine learning are just, fair, and intelligible, and are aligned with our human values. Given that values vary across cultures, an additional ethical challenge is to ensure that these AI systems are not developed according to some unquestioned but questionable assumption of universal norms but are in fact compatible with the societies in which they operate. This is particularly pertinent for AI research and implementation across Africa, a ground where AI systems are and will be used but also a place with a history of imposition of outside values. In this chapter, we thus critically examine one proposal for ensuring that decision-making systems are just, fair, and intelligible—that we adopt a principle of explicability to generate specific recommendations—to assess whether the principle should be adopted in an African research context. We argue that a principle of explicability not only can contribute to responsible and thoughtful development of AI that is sensitive to African interests and values, but can also advance tackling some of the computational challenges in machine learning research. In this way, the motivation for ensuring that a machine learning-based system is just, fair, and intelligible is not only to meet ethical requirements, but also to make effective progress in the field itself.

A dominant approach to conceptualizing a role for emotions in practical agency has been to focus on a relation between emotions and reasons, whereby emotions are claimed to track reason-giving considerations via their intentional content. Yet, if we reflect on the phenomenology of emotional consciousness and take seriously a growing consensus that emotions involve intentional feelings then, I argue, such a reason-tracking approach at best only provides part of the story and at worst is fundamentally misguided. This does not mean that emotion has no role in practical agency, however. I tentatively propose that the normative category of commitments offers a promising alternative for thinking about the role of emotions in practical agency, an alternative that has the potential to do justice to intentional feelings while avoiding the problems of a reason-tracking approach.

Many ethical frameworks require artificial intelligence (AI) systems to be explainable. Explainable AI (XAI) models are frequently tested for their adequacy in user studies. Since different people may have different explanatory needs, it is important that participant samples in user studies are large enough to represent the target population to enable generalizations. However, it is unclear to what extent XAI researchers reflect on and justify their sample sizes or avoid broad generalizations across people. We analyzed XAI user studies (N = 220) published between 2012 and 2022. Most studies did not offer rationales for their sample sizes. Moreover, most papers generalized their conclusions beyond their target population, and there was no evidence that broader conclusions in quantitative studies were correlated with larger samples. These methodological problems can impede evaluations of whether XAI systems implement the explainability called for in ethical frameworks. We outline principles for more inclusive XAI user studies.

In _A Relational Moral Theory: African ethics in and beyond the continent_ ( 2022 ), Thaddeus Metz proposes an African moral theory according to which we ought to respect and honour the capacity of individuals to be party to harmonious relationship. He aims to present a moral theory that should ‘be weighed up against at least contemporary Western moral theories’ (p. 2). As Metz intends his theory to be a serious contender with other moral theories, I assess how his moral theory can be applied and contribute to contemporary moral discussions _‘beyond the continent’_, as the subtitle of the book puts it. In particular, as Metz’s relational moral theory is grounded on the value of harmony, an important question is how well his theory handles _disharmony_ especially if his theory is to have application in our non-ideal world in which disharmony is rife. And here, I hold up Metz’s account to one area of disharmony, disruptive emotions and affective injustice. By first establishing how disruptive emotions and affective injustice pose a challenge to similar relational moral theories grounded in harmony, I argue that Metz’s theory has distinct advantages in the way in which it can indeed engage with such phenomena. I close the paper by considering the objection that Metz’s account is limited in making sense of a common form of disruptive emotion in response to structural injustices where affective injustice can arise, but ultimately conclude that such a limitation is superficial.

Digital Earth applications provide a common ground for visualizing, simulating, and modeling real-world situations. The potential of Digital Earth applications has increased significantly with the evolution of artificial intelligence systems and the capacity to collect and process complex amounts of geospatial data. Yet, the widespread techno-optimism at the root of Digital Earth must now confront concerns over high-risk artificial intelligence systems and power asymmetries of a datafied society. In this commentary, we claim that not only can current debates about data governance and ethical artificial intelligence inform development in the field of Digital Earth, but that the specificities of geospatial data, together with the expectations surrounding Digital Earth applications, offer a fruitful lens through which to examine current debates on data governance and artificial intelligence ethics. In particular, we argue that for the implementation of ethical artificial intelligence and inclusive approaches to data governance, Digital Earth initiatives need to involve stakeholders and communities at the local level and be sensitive to social, legal, cultural, and institutional contexts, including conflicts that might arise within those contexts.

Philosophical discussions of hatred have tended to examine whether and when hatred can be morally or rationally justified. If hatred is rational, for instance, it might be because it is a fitting response to the given circumstances (Ben-Ze’ev 2000; Bell 2011). At the same time, hatred typically motivates action, and action of quite characteristic types. As Fischer et al. (2018) note, the so-called ‘emotivational’ goal of hatred appears to be not merely to hurt the target of hatred, but to destroy them or let them suffer. Such types of action immediately raise questions about their moral status, but we should also ask questions about their rational status. In particular, when acting on the basis of a hatred that may itself be rational – at least in the sense of being fitting – are we also acting rationally? Working with a conception of rational agency according to which we act rationally when we guide our actions by reasons seen as reasons (see Jones 2003; Carman 2018), and granting that hatred can be fitting and thereby be a response to genuine reasons, I explore the ways in which acting on the basis of hatred could or could not be rational.

Anger has potential epistemic value in the way that it can facilitate a process of our coming to have knowledge and understanding regarding the issue about which we are angry. The nature of anger, however, may nevertheless be such that it ultimately undermines this very process. Common non-philosophical complaints about anger, for instance, often target the angry person as being somehow irrational, where an unformulated assumption is that her anger undermines her capacity to rationally engage with the issue about which she is angry. Call this assumption the charge of emotional irrationality regarding anger. Such a charge is pernicious when levelled at the anger of those in positions of marginalisation or oppression, where it can threaten to silence voices on the very issue of the injustices that they face. In this paper I thus provide a much-needed interrogation of this charge. Firstly, and drawing on empirical literature on the effects of anger on decision-making, I flesh out the charge and why it poses a threat to how the epistemic value of anger has been defended. Secondly, I argue that the charge of emotional irrationality regarding anger can nevertheless be unwarranted, at least within a common context of political anger.

Bernard Matolino has recently argued that African communitarianism is an ethics grounded in emotion aligned with reason. If he is correct, questions arise about what emotions have value within African communitarianism, especially as emotions like anger or resentment could stand in tension with important communitarian values, such as social harmony. While little critical attention has so far been paid to such emotions within an African communitarian framework, a wider philosophical literature examining the moral value of disruptive emotions could be drawn on to develop analyses of emotion within African communitarianism. In this paper, I explore how such an analysis could proceed. I argue that drawing on the wider emotion literature, and especially the concept of reactive attitudes introduced by P. F. Strawson, provides an initial case for even disruptive emotions to have value. Even so, we must question whether an analysis of emotion plausibly based on individualistic commitments is compatible with relational communitarian commitments. I nevertheless defend the compatibility and argue that, not only can disruptive emotions have instrumental value through their epistemic and motivational roles in the promotion of community but, importantly, they are partially constitutive of the interpersonal relationships within which we are embedded and that form community.

Artificial Intelligence (AI) is increasingly being used in disaster risk management applications to predict the effect of upcoming disasters, plan for mitigation strategies, and determine who needs how much aid after a disaster strikes. The media is filled with unintended ethical concerns of AI algorithms, such as image recognition algorithms not recognizing persons of color or racist algorithmic predictions of whether offenders will recidivate. We know such unintended ethical consequences must play a role in DRM as well, yet there is surprisingly little research on exactly what the unintended consequences are and what we can do to mitigate them. The aim of this perspective is to call researchers working on fairness, accountability, and transparency to work with DRM and local experts—so we can ensure that disaster mitigation and relief is accountable, considers local values, and is not unintentionally biased.

Assuming that moral enhancement is morally permissible, I contend that a more careful theoretical treatment of emotion and the affective landscape is needed to advance both our understanding and the prospects of interventions aimed at moral enhancement. Using Douglas’ proposal for the direct modulation of counter‐moral emotions as a foil for discussion, I argue that the direct modulation of emotion fails to address underlying aspects of an agent’s psychology that will give rise to a range of counter‐moral motives beyond the emotion that is targeted. This is because the direct modulation of emotion does not accommodate the fact that an emotional episode or emotional disposition is part of a network of other dispositions and responses, against a backdrop of our having cares and goals, as well as other cognitive content that can mediate how we apprehend a situation. By focusing on emotion rather than on moral capacity as other critiques have done, I thus argue that direct modulation is likely to be ineffective, and, more positively, by drawing out patterns of connections across the affective landscape, we can better tease out what it is that we should really be targeting if we are to achieve robust and meaningful moral change.

Developing and implementing artificial intelligence (AI) systems in an ethical manner faces several challenges specific to the kind of technology at hand, including ensuring that decision-making systems making use of machine learning are just, fair, and intelligible, and are aligned with our human values. Given that values vary across cultures, an additional ethical challenge is to ensure that these AI systems are not developed according to some unquestioned but questionable assumption of universal norms but are in fact compatible with the societies in which they operate. This is particularly pertinent for AI research and implementation across Africa, a ground where AI systems are and will be used but also a place with a history of imposition of outside values. In this paper, we thus critically examine one proposal for ensuring that decision-making systems are just, fair, and intelligible—that we adopt a principle of explicability to generate specific recommendations—to assess whether the principle should be adopted in an African research context. We argue that a principle of explicability not only can contribute to responsible and thoughtful development of AI that is sensitive to African interests and values, but can also advance tackling some of the computational challenges in machine learning research. In this way, the motivation for ensuring that a machine learning-based system is just, fair, and intelligible is not only to meet ethical requirements, but also to make effective progress in the field itself.

Frustration is widely recognised to be central to many cases of moral anger in a political context, yet little philosophical attention has been paid to it. In this paper, I offer a much-needed philosophical analysis of frustration, working primarily with the example of the recent South African student protests. By developing a deeper philosophical understanding of frustration and its connections to moral anger, I argue that the movement between the two has a couple of important aspects. First, the movement involves a transformation from simply responding to some opposition to viewing that opposition as wrongfully imposed by an agent; second, in the cases that are my focus, the movement is a response to a persistent opposition that is compounded by an erosion of trust, confidence and hope within a moral community, which ultimately changes the very nature of the opposition into a form of wrongdoing. These aspects allow us to assess the appropriateness of the movement, thereby developing a rich analysis of moral anger arising out of frustration in a political context.

In a recent article in this journal, Steve Clarke (2017) identifies two different bases for conscience-based refusals in healthcare: (1) all-things-considered moral judgments, and (2) the dictates of conscience. He argues that these two bases have distinct roles in justifying conscientious objection. However, accepting that there are these two bases, I argue that both are not able to justify conscientious objection. In particular, I argue that the second basis of the dictates of conscience cannot justify conscience-based refusal in a healthcare context. Even if someone objects in a healthcare context on the basis of the dictates of her conscience, and even if we can explain why she objects with reference to the dictates of her conscience, her objection will only be justified if she makes a judgment.

If emotions provide reasons for action through their intentional content, as is often argued, where does this leave the role of the affective element of an emotion? Can it be more than a motivator and have significant bearing of its own on our emotional actions, as actions done for reasons? One way it can is through reinforcing other reasons that we might have, as Greenspan argues. Central to Greenspan’s account is the claim that the affective discomfort of an emotion, as a fact about the agent’s state of being, provides an additional normative reason to act to alleviate the state. This, I argue, is not correct, nor is it the best way to understand emotions as reason-reinforcers. In this paper, I thus do two things: I provide an examination of how and why the affect of emotion could provide reasons to act to alleviate it and I propose that the real way emotions reinforce reasons is through the way they orient our attention onto things that matter, registering them as salient.

If emotions have a rational role in action, then one challenge for accounting for how we can act rationally when acting emotionally is to show how we can guide our actions by our emotional considerations, seen as reasons. In this paper, I put forward a novel proposal for how this can be so. Drawing on the interconnection between emotions, cares and caring, I argue that, as the emotional agent is a caring agent, she can be aware of the emotional consideration as a pro tanto reason favouring an action choice and, even, as the reason for which she should act.

Calls to decolonise the university and revise what we research and teach is a challenge that ought to be taken up by those working in African philosophy and philosophy in Africa, more generally. Often, the thought is that such decolonisation will involve a complete subversion, destruction or deconstruction of colonial attitudes, processes and concepts. A more moderate proposal for decolonisation of philosophy can be found, however, which is Kwasi Wiredu’s project of conceptual decolonisation. In this paper, I defend the project from two categories of objection, what I call the decolonisation and methodology objections. I argue that the objections are misdirected and unfair because they fail to recognise the theoretical backing of Wiredu’s wider work. The critical reflection central to Wiredu’s project of conceptual decolonisation can be a genuinely helpful and viable route to take when tackling the challenges of decolonising philosophy.