AI ethics and systemic risks in finance

The paper suggests that AI ethics should pay attention to morally relevant systemic effects of AI use. It draws the attention of ethicists and practitioners to systemic risks that have been neglected so far in professional AI-related codes of conduct, industrial standards and ethical discussions more generally. The paper uses the financial industry as an example to ask: how can AI-enhanced systemic risks be ethically accounted for? Which specific issues does AI use raise for ethics that takes systemic effects into account? The paper (1) relates the literature about AI ethics to the ethics of systemic risks to clarify the moral relevance of AI use with respect to the imposition of systemic risks, (2) proposes a theoretical framework based on the ethics of complexity and (3) applies this framework to discuss implications for AI ethics concerned with AI-enhanced systemic risks.

Intention and Wrongdoing

According to the principle of double effect, there is a strict moral constraint against bringing about serious harm to the innocent intentionally, but it is permissible in a wider range of circumstances to act in a way that brings about harm as a foreseen but non-intended side effect. This idea plays an important role in just war theory and international law, and in the twentieth century Elizabeth Anscombe and Philippa Foot invoked it as a way of resisting consequentialism. However, many moral philosophers now regard the principle with hostility or suspicion. Challenging the philosophical orthodoxy, Joshua Stuchlik defends the principle of double effect, situating it within a moral framework of human solidarity and responding to philosophical objections to it. His study uncovers links between ethics, philosophy of action, and moral psychology, and will be of interest to anyone seeking to understand the moral relevance of intention.

Epistemic values and the Big Five: Personality characteristics of those who ascribe personal and moral value to epistemic rationality

People differ in how much personal importance, and moral relevance, they ascribe to epistemic rationality. These stable individual differences can be assessed using the Importance of Rationality Scale (IRS), and Moralized Rationality Scale (MRS). Furthermore, these individual differences are conceptually distinct, and associated with different cognitive, affective, and behavioral outcomes. However, little is known about what signifies and differentiates people who score high (vs. low) on the IRS and MRS respectively, and where these individual differences stem from. In the present research we begin to address these questions by examining how these epistemic values relate to the Big Five personality traits. Two studies consistently show that both the IRS and MRS are positively related to Openness to experience. However, only the MRS is negatively associated with Agreeableness, and only the IRS is positively associated with Conscientiousness.

An Interest-Based Model of Moral Status

This chapter briefly defends each of the following theses, which together comprise an interest-based model of moral status: (1) Being human is neither necessary nor sufficient for moral status; (2) The capacity for consciousness is necessary but not sufficient; (3) Sentience is necessary and sufficient; (4) Social relations are not a basis for moral status but may ground special obligations to those with moral status; (5) The concept of personhood is unhelpful in modeling moral status, unless a non-vague conception is identified and its moral relevance clarified; (6) Sentient beings are entitled to equal consequentialist consideration; and (7) Sentient beings with substantial temporal self-awareness have special interests that justify the added protection of rights. This model will be engaged in illuminating the moral status of ordinary, self-aware human beings, non-paradigm humans, animals, robots and AI systems, brain organoids, and post-humans with superior self-awareness.

Does kindness towards robots lead to virtue? A reply to Sparrow’s asymmetry argument

Does cruel behavior towards robots lead to vice, whereas kind behavior does not lead to virtue? This paper presents a critical response to Sparrow’s argument that there is an asymmetry in the way we (should) think about virtue and robots. It discusses how much we should praise virtue as opposed to vice, how virtue relates to practical knowledge and wisdom, how much illusion is needed for it to be a barrier to virtue, the relation between virtue and consequences, the moral relevance of the reality requirement and the different ways one can deal with it, the risk of anthropocentric bias in this discussion, and the underlying epistemological assumptions and political questions. This response is not only relevant to Sparrow’s argument or to robot ethics but also touches upon central issues in virtue ethics.

Ties That Remind: Known Family Connections to Past Events as Salience Cues and Collective Memory of Stalin’s Repressions of the 1930s in Contemporary Russia

Building on ideas of Halbwachs and others regarding how families shape collective memory, we argue that known family connections to past events serve as salience cues. Due to kin preference (humans’ tendency to empathize with family members more than strangers), awareness that a relative participated in a specific past event increases its visibility, moral relevance, and emotional resonance, compared to that of the vast number of other historical occurrences, with intuitive consequences for whether and how the event is remembered in the present. We illustrate this effect of known family connections to the past by analyzing whether and how contemporary Russians recall a controversial episode from the Soviet period: Stalin’s repressions of the 1930s. We use qualitative data from focus groups and unusually detailed survey data, collected in 2010, to illustrate this property of recognized family connections to a past mass trauma. We also propose four distinct components of perceptions of past events: awareness, knowledge, importance, and moral valence. Our findings confirm the strong influence of known family ties to victims, which exhibit more consistent connections to memories of the repressions than do other factors, although family socialization through childhood discussions, cohort differences, education, and exposure to official narratives also matter.

Letting Climate Change

Recent work by Ingmar Persson and Jason Hanna has posed an interesting new challenge for deontologists: How can they account for so-called cases of letting oneself do harm? In this article, I argue that cases of letting oneself do harm are structurally similar to real-world cases such as climate change, and that deontologists need an account of the moral status of these cases to provide moral guidance in real-world cases. I then explore different ways in which deontologists can solve this challenge and argue that the most promising way to conceive of cases of letting oneself do harm is as nonstandard cases of allowing harm, supplemented with an additional argument for the moral relevance of one's own agency. The upshot is that cases of letting oneself do harm are both more theoretically challenging and practically important than has been acknowledged.

Among equity and dignity: an argument-based review of European ethical guidelines under COVID-19

Background Under COVID-19 pandemic, many organizations developed guidelines to deal with the ethical aspects of resources allocation. This study describes the results of an argument-based review of ethical guidelines developed at the European level. It aims to increase knowledge and awareness about the moral relevance of the outbreak, especially as regards the balance of equity and dignity in clinical practice and patient’s care. Method According to the argument-based review framework, we started our research from the following two questions: what are the ethical principles adopted by the ethical guidelines produced at the beginning of the COVID-19 outbreak related to resource allocation? And what are the practical consequences in terms of 'priority' of access, access criteria, management of the decision-making process and patient care? Results Twenty-two ethical guidelines met our inclusion criteria and the results of our analysis are organized into 4 ethical concepts and related arguments: the equity principle and emerging ethical theories; triage criteria; respecting patient’s dignity, and decision making and quality of care. Conclusion Further studies can investigate the practical consequences of the application of the guidelines described, in terms of quality of care and health care professionals’ moral distress.