ABSTRACTFinance may suffer from institutional deformations that subordinate its distinctive goods to the pursuit of external goods, but this should encourage attempts to reform the institutionalization of finance rather than to reject its potential for virtuous business activity. This article argues that finance should be regarded as a domain-relative practice. Alongside management, its moral status thereby varies with the purposes it serves. Hence, when practitioners working in finance facilitate projects that create common goods, it allows them to develop virtues. This (...) argument applies MacIntyre’s widely acknowledged account of the relationship between practices and the development of virtues while questioning some of his claims about finance. It also takes issue with extant accounts of particular financial functions that have failed to identify the distinctive goods of financial practice. (shrink)

In an important and widely cited paper, Zerilli, Knott, Maclaurin, and Gavaghan (2019) argue that opaque AI decision makers are at least as transparent as human decision makers and therefore the concern that opaque AI is not sufficiently transparent is mistaken. I argue that the concern about opaque AI should not be understood as the concern that such AI fails to be transparent in a way that humans are transparent. Rather, the concern is that the way in which opaque AI (...) is opaque is very different than the way in which humans are opaque. What matters is the degree to which the opaque processes of a class of decision makers are stable, uniform, and safe. The degree to which such processes have these features in humans is higher than the degree to which such processes have these features in opaque AI. And therefore we should require AI to be more transparent than humans. (shrink)

Artificial intelligent (AI) systems that perform image classification tasks are being used to great success in many application contexts. However, many of these systems are opaque, even to experts. This lack of understanding can be problematic for ethical, legal, or practical reasons. The research field Explainable AI (XAI) has therefore developed several approaches to explain image classifiers. The hope is to bring about understanding, e.g., regarding why certain images are classified as belonging to a particular target class. Most of these (...) approaches use visual explanations. Drawing on Elgin’s work (True enough. MIT Press, Cambridge, 2017), I argue that analyzing what those explanations exemplify can help to assess their suitability for producing understanding. More specifically, I suggest to distinguish between two forms of examples according to their suitability for producing understanding. I call these forms samples and exemplars, respectively. Samples are prone to misinterpretation and thus carry the risk of leading to misunderstanding. Exemplars, by contrast, are intentionally designed or chosen to meet contextual requirements and to mitigate the risk of misinterpretation. They are thus preferable for bringing about understanding. By reviewing several XAI approaches directed at image classifiers, I show that most of them explain with samples. If my analysis is correct, it will be beneficial if such explainability methods use explanations that qualify as exemplars. (shrink)

In the ethics of algorithms, a specifically epistemological analysis is rarely undertaken in order to gain a critique (or a defense) of the handling of or trust in medical black box algorithms (BBAs). This article aims to begin to fill this research gap. Specifically, the thesis is examined according to which such algorithms are regarded as epistemic authorities (EAs) and that the results of a medical algorithm must completely replace other convictions that patients have (preemptionism). If this were true, it (...) would be a reason to distrust medical BBAs. First, the author describes what EAs are and why BBAs can be considered EAs. Then, preemptionism will be outlined and criticized as an answer to the question of how to deal with an EA. The discussion leads to some requirements for dealing with a BBA as an EA. (shrink)

Odgovornost pripisujemo sa namerom da postignemo neki cilj. Jedno od opših mesta u filozofskoj literaturi je da osobi možemo pripisati moralnu odgovornost ako su zadovoljena bar dva uslova: da subjekt delanja ima kontrolu nad svojim postupcima i da je u stanju da navede razloge u prilog svog postupka. Međutim, četvrtu industrijsku revoluciju karakterišu sociotehnološke pojave koje nas potencijalno suočavaju sa tzv. problemom jaza u odgovornosti. Rasprave o odgovornosti u kontekstu veštačke inteligencije karakteriše nejasna i neodređena upotreba ovog pojma. Da bismo (...) razumeli problem jaza u odgovornosti, njegove domete, kao i eventualna rešenja, u radu razmatramo različite vrste pojma odgovornosti. Identifikujući uslove primene ovih pojmova uviđamo da problem jaza u moralnoj odgovornosti dobija na značaju utoliko što nije očigledno na koji način bi mogao biti premošćen. (shrink)

Harms and risks to groups and third-parties can be significant in the context of research, particularly in data-centric studies involving genomic, artificial intelligence, and/or machine learning technologies. This article explores whether and how United States federal regulations should be adapted to better align with current ethical thinking and protect group interests. Three aspects of the Common Rule deserve attention and reconsideration with respect to group interests: institutional review board (IRB) assessment of the risks/benefits of research; disclosure requirements in the informed (...) consent process; and criteria for waivers of informed consent. In accordance with respect for persons and communities, investigators and IRBs should systematically consider potential group harm when designing and reviewing protocols, respectively. Research participants should be informed about any potential group harm in the consent process. We call for additional public discussion, empirical research, and normative analysis on these issues to determine the right regulatory and policy path forward. (shrink)

Artificial Intelligence (AI) technologies offer new ways of conducting decision-making tasks that influence the daily lives of citizens, such as coordinating traffic, energy distributions, and crowd flows. They can sort, rank, and prioritize the distribution of fines or public funds and resources. Many of the changes that AI technologies promise to bring to such tasks pertain to decisions that are collectively binding. When these technologies become part of critical infrastructures, such as energy networks, citizens are affected by these decisions whether (...) they like it or not, and they usually do not have much say in them. The democratic challenge for those working on AI technologies with collectively binding effects is both to _develop_ and _deploy_ technologies in such a way that the democratic legitimacy of the relevant decisions is safeguarded. In this paper, we develop a conceptual framework to help policymakers, project managers, innovators, and technologists to assess and develop approaches to democratize AI. This framework embraces a broad sociotechnical perspective that highlights the interactions between technology and the complexities and contingencies of the context in which these technologies are embedded. We start from the problem-based and practice-oriented approach to democracy theory as developed by political theorist Mark Warren. We build on this approach to describe practices that can enhance or challenge democracy in political systems and extend it to integrate a sociotechnical perspective and make the role of technology explicit. We then examine how AI technologies can play a role in these practices to improve or inhibit the democratic nature of political systems. We focus in particular on AI-supported political systems in the energy domain. (shrink)

Artificial Intelligence (AI) is increasingly adopted in society, creating numerous opportunities but at the same time posing ethical challenges. Many of these are familiar, such as issues of fairness, responsibility and privacy, but are presented in a new and challenging guise due to our limited ability to steer and predict the outputs of AI systems. This chapter first introduces these ethical challenges, stressing that overviews of values are a good starting point but frequently fail to suffice due to the context (...) sensitivity of ethical challenges. Frequently, additional (ethical) values emerge for specific applications, as e.g. challenges with fraud detection are very different from those around language technologies. Second, methods to tackle these challenges are discussed. Main ethical theories (virtue ethics, consequentialism, and deontology) are shown to provide a starting point, but often lack the details needed for actionable AI Ethics. Instead, mid-level philosophical theories coupled to design-approaches such as Design for Values, together with interdisciplinary working methods, offer the best way forwards. The chapter aims to show how these approaches can lead to an ethics of AI that is actionable and that can be proactively integrated in the design of AI systems. (shrink)

The ongoing explosion of interest in artificial intelligence is fueled in part by recently developed techniques in machine learning. Those techniques allow automated systems to process huge amounts of data, utilizing mathematical methods that depart from traditional statistical approaches, and resulting in impressive advancements in our ability to make predictions and uncover correlations across a host of interesting domains. But as is now widely discussed, the way that those systems arrive at their outputs is often opaque, even to the experts (...) who design and deploy them. Is it morally problematic to make use of opaque automated methods when making high-stakes decisions, like whether to issue a loan to an applicant, or whether to approve a parole request? Many scholars answer in the affirmative. However, there is no widely accepted explanation for why transparent systems are morally preferable to opaque systems. We argue that the use of automated decision-making systems sometimes violates duties of consideration that are owed by decision-makers to decision-subjects, duties that are both epistemic and practical in character. Violations of that kind generate a weighty consideration against the use of opaque decision systems. In the course of defending our approach, we show that it is able to address three major challenges sometimes leveled against attempts to defend the moral import of transparency in automated decision-making. (shrink)

Unsupervised learning algorithms are widely used for many important statistical tasks with numerous applications in science and industry. Yet despite their prevalence, they have attracted remarkably little philosophical scrutiny to date. This stands in stark contrast to supervised and reinforcement learning algorithms, which have been widely studied and critically evaluated, often with an emphasis on ethical concerns. In this article, I analyze three canonical unsupervised learning problems: clustering, abstraction, and generative modeling. I argue that these methods raise unique epistemological and (...) ontological questions, providing data-driven tools for discovering natural kinds and distinguishing essence from contingency. This analysis goes some way toward filling the lacuna in contemporary philosophical discourse on unsupervised learning, as well as bringing conceptual unity to a heterogeneous field more often described by what it isnot(i.e., supervised or reinforcement learning) than by what itis. I submit that unsupervised learning is not just a legitimate subject of philosophical inquiry but perhaps the most fundamental branch of all AI. However, an uncritical overreliance on unsupervised methods poses major epistemic and ethical risks. I conclude by advocating for a pragmatic, error-statistical approach that embraces the opportunities and mitigates the challenges posed by this powerful class of algorithms. (shrink)

Algorithm engineering is sometimes portrayed as a new 21st century return of manipulative social engineering. Yet algorithms are necessary tools for individuals to navigate online platforms. Algorithms are like a sensory apparatus through which we perceive online platforms: this is also why individuals can be subtly but pervasively manipulated by biased algorithms. How can we better understand the nature of algorithm engineering and its proper function? In this chapter I argue that algorithm engineering can be best conceptualized as a type (...) of environmental engineering aimed at making the online environment more hospitable to human use, in particular by safeguarding the conditions that allow for trust. (shrink)

Danaher (2016) has argued that increasing robotization can lead to retribution gaps: Situation in which the normative fact that nobody can be justly held responsible for a harmful outcome stands in conflict with our retributivist moral dispositions. In this paper, we report a cross-cultural empirical study based on Sparrow’s (2007) famous example of an autonomous weapon system committing a war crime, which was conducted with participants from the US, Japan and Germany. We find that (i) people manifest a considerable willingness (...) to hold autonomous systems morally responsible, (ii) partially exculpate human agents when interacting with such systems, and that more generally (iii) the possibility of normative responsibility gaps is indeed at odds with people’s pronounced retributivist inclinations. We discuss what these results mean for potential implications of the retribution gap and other positions in the responsibility gap literature. (shrink)

This article explores the trend of increasing automation in law enforcement and criminal justice settings through three use cases: predictive policing, machine evidence and recidivism algorithms. The focus lies on artificial-intelligence-driven tools and technologies employed, whether at pre-investigation stages or within criminal proceedings, in order to decode human behaviour and facilitate decision-making as to whom to investigate, arrest, prosecute, and eventually punish. In this context, this article first underlines the existence of a persistent dilemma between the goal of increasing the (...) operational efficiency of police and judicial authorities and that of safeguarding fundamental rights of the affected individuals. Subsequently, it shifts the focus onto key principles of criminal procedure and the presumption of innocence in particular. Using Article 6 ECHR and the Directive (EU) 2016/343 as a starting point, it discusses challenges relating to the protective scope of presumption of innocence, the burden of proof rule and the _in dubio pro reo_ principle as core elements of it. Given the transformations law enforcement and criminal proceedings go through in the era of algorithms, big data and artificial intelligence, this article advocates the adoption of specific procedural safeguards that will uphold rule of law requirements, and particularly transparency, fairness and explainability. In doing so, it also takes into account EU legislative initiatives, including the reform of the EU data protection _acquis_, the E-evidence Proposal, and the Proposal for an EU AI Act. Additionally, it argues in favour of revisiting the protective scope of key fundamental rights, considering, _inter alia_, the new dimensions suspicion has acquired. (shrink)

This paper presents an ethical framework designed to support the development of critical data literacy for research methods courses and data training programmes in higher education. The framework we present draws upon our reviews of literature, course syllabi and existing frameworks on data ethics. For this research we reviewed 250 research methods syllabi from across the disciplines, as well as 80 syllabi from data science programmes to understand how or if data ethics was taught. We also reviewed 12 data ethics (...) frameworks drawn from different sectors. Finally, we reviewed an extensive and diverse body of literature about data practices, research ethics, data ethics and critical data literacy, in order to develop a transversal model that can be adopted across higher education. To promote and support ethical approaches to the collection and use of data, ethics training must go beyond securing informed consent to enable a critical understanding of the techno-centric environment and the intersecting hierarchies of power embedded in technology and data. By fostering ethics as a method, educators can enable research that protects vulnerable groups and empower communities. (shrink)

Many popular artificial intelligence (AI) ethics frameworks center the principle of autonomy as necessary in order to mitigate the harms that might result from the use of AI within society. These harms often disproportionately affect the most marginalized within society. In this paper, we argue that the principle of autonomy, as currently formalized in AI ethics, is itself flawed, as it expresses only a mainstream mainly liberal notion of autonomy as rational self-determination, derived from Western traditional philosophy. In particular, we (...) claim that the adherence to such principle, as currently formalized, does not only fail to address many ways in which people’s autonomy can be violated, but also to grasp a broader range of AI-empowered harms profoundly tied to the legacy of colonization, and which particularly affect the already marginalized and most vulnerable on a global scale. To counter such a phenomenon, we advocate for the need of a relational turn in AI ethics, starting from a relational rethinking of the AI ethics principle of autonomy that we propose by drawing on theories on relational autonomy developed both in moral philosophy and Ubuntu ethics. (shrink)

Artificial Social Agents (ASAs), which are AI software driven entities programmed with rules and preferences to act autonomously and socially with humans, are increasingly playing roles in society. As their sophistication grows, humans will share greater amounts of personal information, thoughts, and feelings with ASAs, which has significant ethical implications. We conducted a study to investigate what ethical principles are of relative importance when people engage with ASAs and whether there is a relationship between people’s values and the ethical principles (...) they prioritise. The study uses scenarios, embedded with the five AI4People ethical principles (Beneficence, Non-maleficence, Autonomy, Justice, and Explicability), involving ASAs taking on roles traditionally played by humans to understand whether these roles and behaviours (including dialogues) are seen as acceptable or unacceptable. Results from 268 participants reveal the greatest sensitivity to ASA behaviours that relate to Autonomy, Justice, Explicability, and the privacy of their personal data. Models were created using Schwartz’s Refined Values as a possible indicator of how stakeholders discern and prioritise the different AI4People ethical principles when interacting with ASAs. Our findings raise issues around the ethical acceptability of ASAs for nudging and changing behaviour due to participants’ desire for autonomy and their concerns over deceptive ASA behaviours such as pretending to have memories and emotions. (shrink)

Many artificial intelligence (AI) systems currently used for decision-making are opaque, i.e., the internal factors that determine their decisions are not fully known to people due to the systems’ computational complexity. In response to this problem, several researchers have argued that human decision-making is equally opaque and since simplifying, reason-giving explanations (rather than exhaustive causal accounts) of a decision are typically viewed as sufficient in the human case, the same should hold for algorithmic decision-making. Here, I contend that this argument (...) overlooks that human decision-making is sometimes significantly more transparent and trustworthy than algorithmic decision-making. This is because when people explain their decisions by giving reasons for them, this frequently prompts those giving the reasons to govern or regulate themselves so as to think and act in ways that confirm their reason reports. AI explanation systems lack this self-regulative feature. Overlooking it when comparing algorithmic and human decision-making can result in underestimations of the transparency of human decision-making and in the development of explainable AI that may mislead people by activating generally warranted beliefs about the regulative dimension of reason-giving. (shrink)

While the consequences of mathematically-based software, algorithms and strategies have become ever wider and better appreciated, ethical reflection on mathematics has remained primitive. We review the somewhat disconnected suggestions of commentators in recent decades with a view to piecing together a coherent approach to ethics in mathematics. Calls for a Hippocratic Oath for mathematicians are examined and it is concluded that while lessons can be learned from the medical profession, the relation of mathematicians to those affected by their work is (...) significantly different. There is something to be learned also from the codes of conduct of cognate but professionalised quantitative disciplines such as engineering and accountancy, as well as from legal principles bearing on professional work. We conclude with recommendations that professional societies in mathematics should sponsor an (international) code of ethics, institutional mission statements for mathematicians and syllabuses of ethics courses for incorporation into mathematics degrees. (shrink)

Embedding ethics modules within computer science courses has become a popular response to the growing recognition that CS programs need to better equip their students to navigate the ethical dimensions of computing technologies like AI, machine learning, and big data analytics. However, the popularity of this approach has outpaced the evidence of its positive outcomes. To help close that gap, this empirical study reports positive results from Northeastern’s program that embeds values analysis modules into CS courses. The resulting data suggest (...) that such modules have a positive effect on students’ moral attitudes and that students leave the modules believing they are more prepared to navigate the ethical dimensions they’ll likely face in their eventual careers. Importantly, these gains were accomplished at an institution without a philosophy doctoral program, suggesting this strategy can be effectively employed by a wider range of institutions than many have thought. (shrink)

The paper presents a framework to realise “meaningful human control” over Automated Driving Systems. The framework is based on an original synthesis of the results of the multidisciplinary research project “Meaningful Human Control over Automated Driving Systems” lead by a team of engineers, philosophers, and psychologists at Delft University of the Technology from 2017 to 2021. Meaningful human control aims at protecting safety and reducing responsibility gaps. The framework is based on the core assumption that human persons and institutions, not (...) hardware and software and their algorithms, should remain ultimately—though not necessarily directly—in control of, and thus morally responsible for, the potentially dangerous operation of driving in mixed traffic. We propose an Automated Driving System to be under meaningful human control if it behaves according to the relevant reasons of the relevant human actors (tracking), and that any potentially dangerous event can be related to a human actor (tracing). We operationalise the requirements for meaningful human control through multidisciplinary work in philosophy, behavioural psychology and traffic engineering. The tracking condition is operationalised via a proximal scale of reasons and the tracing condition via an evaluation cascade table. We review the implications and requirements for the behaviour and skills of human actors, in particular related to supervisory control and driver education. We show how the evaluation cascade table can be applied in concrete engineering use cases in combination with the definition of core components to expose deficiencies in traceability, thereby avoiding so-called responsibility gaps. Future research directions are proposed to expand the philosophical framework and use cases, supervisory control and driver education, real-world pilots and institutional embedding. (shrink)

Healthcare provision, like many other sectors of society, is undergoing major changes due to the increased use of data-driven methods and technologies. This increased reliance on big data in medicine can lead to shifts in the norms that guide healthcare providers and patients. Continuous critical normative reflection is called for to track such potential changes. This article presents the results of an interview-based study with 20 German and Swiss experts from the fields of medicine, life science research, informatics and humanities (...) of digitalisation. The aim of the study was to explore expert opinions regarding current challenges and opportunities related to data-driven medicine and medical research and to provide a methodological framework for empirically grounded, continuous normative reflection. To this end, we developed a heuristic tool to map and structure empirical findings for normative analysis. Using this tool, our interview material points to a polarisation between individualistic and collectivistic orientated argumentations among experts. The study shows that a multilevel analysis is required to deal with complex normative implications of data-driven approaches in medical research and healthcare. (shrink)

AI systems have often been found to contain gender biases. As a result of these gender biases, AI routinely fails to adequately recognize the needs, rights, and accomplishments of women. In this article, we use Axel Honneth’s theory of recognition to argue that AI’s gender biases are not only an ethical problem because they can lead to discrimination, but also because they resemble forms of misrecognition that can hurt women’s self-development and self-worth. Furthermore, we argue that Honneth’s theory of recognition (...) offers a fruitful framework for improving our understanding of the psychological and normative implications of gender bias in modern technologies. Moreover, our Honnethian analysis of gender bias in AI shows that the goal of responsible AI requires us to address these issues not only through technical interventions, but also through a change in how we grant and deny recognition to each other. (shrink)

This chapter examines the possibility of using AI technologies to improve human moral reasoning and decision-making, especially in the context of purchasing and consumer decisions. We characterize such AI technologies as artificial ethics assistants (AEAs). We focus on just one part of the AI-aided moral improvement question: the case of the individual who wants to improve their morality, where what constitutes an improvement is evaluated by the individual’s own values. We distinguish three broad areas in which an individual might think (...) their own moral reasoning and decision-making could be improved: one’s actions, character, or other evaluable attributes fall short of one’s values and moral beliefs; one sometimes misjudges or is uncertain about what the right thing to do is in particular situations, given one’s values; one is uncertain about some fundamental moral questions or recognizes a possibility that some of one’s core moral beliefs and values are mistaken. We sketch why one might think that AI tools could be used to support moral improvement in those areas, and describe two types of assistance: preparatory assistance, including advice and training supplied in advance of moral deliberation; and on-the-spot assistance, including on-the-spot advice and facilitation of moral functioning over the course of moral deliberation. Then, we turn to some of the ethical issues that AEAs might raise, looking in particular at three under-appreciated problems posed by the use of AI for moral self-improvement: namely, reliance on sensitive moral data; the inescapability of outside influences on AEAs; and AEA usage prompting the user to adopt beliefs and make decisions without adequate reasons. (shrink)

Algorithmic systems are increasingly used by state agencies to inform decisions about humans. They produce scores on risks of recidivism in criminal justice, indicate the probability for a job seeker to find a job in the labor market, or calculate whether an applicant should get access to a certain university program. In this contribution, we take an interdisciplinary perspective, provide a bird’s eye view of the different key decisions that are to be taken when state actors decide to use an (...) algorithmic system, and illustrate these decisions with empirical examples from case studies. Building on these insights, we discuss the main pitfalls and promises of the use of algorithmic system by the state and focus on four levels: The most basic question whether an algorithmic system should be used at all, the regulation and governance of the system, issues of algorithm design, and, finally, questions related to the implementation of the system on the ground and the human–machine-interaction that comes with it. Based on our assessment of the advantages and challenges that arise at each of these levels, we propose a set of crucial questions to be asked when such intricate matters are addressed. (shrink)

Technologies that deploy data science methods are liable to result in epistemic harms involving the diminution of individuals with respect to their standing as knowers or their credibility as sources of testimony. Not all harms of this kind are unjust but when they are we ought to try to prevent or correct them. Epistemically unjust harms will typically intersect with other more familiar and well-studied kinds of harm that result from the design, development, and use of data science technologies. However, (...) we argue that epistemic injustices can be distinguished conceptually from more familiar kinds of harm. We argue that epistemic harms are morally relevant even in cases where those who suffer them are unharmed in other ways. Via a series of examples from the criminal justice system, workplace hierarchies, and educational contexts we explain the kinds of epistemic injustice that can result from common uses of data science technologies. (shrink)

This chapter will explore the dynamics of power underpinning ethical issues within smart cities via a new paradigm derived from Systems Theory. The smart city is an expression of technology as a socio-technical system. The vision of the smart city contains a deep fusion of many different technical systems into a single integrated “ambient intelligence”. ETICA Project, 2010, p. 102). Citizens of the smart city will not experience a succession of different technologies, but a single intelligent and responsive environment through (...) which they move. Analysis of such an environment requires a framework which transcends traditional ontologically-based models in order to accommodate this deep fusion. This chapter will outline a framework based on Latour’s Actor-Network Theory and Luhmann’s treatment of society as an autopoetic system. We shall use this framework to map the influence of relevant factors on ethical issues, irrespective of their composition or type. For example, under this treatment, both human praxis and technical design can be viewed as comparable tools of domination. This chapter will provide a framework for the analysis of relations between any elements of the smart city, ranging from top-level urban management processes down to individual device operations. While we will illustrate the use of this schema through examination of ethical issues arising from power dynamics within the smart city, it is intended that this example will demonstrate the wider utility of the model in general. (shrink)

The modern abundance and prominence of data has led to the development of “data science” as a new field of enquiry, along with a body of epistemological reflections upon its foundations, methods, and consequences. This article provides a systematic analysis and critical review of significant open problems and debates in the epistemology of data science. We propose a partition of the epistemology of data science into the following five domains: (i) the constitution of data science; (ii) the kind of enquiry (...) that it identifies; (iii) the kinds of knowledge that data science generates; (iv) the nature and epistemological significance of “black box” problems; and (v) the relationship between data science and the philosophy of science more generally. (shrink)

The problem that I present in this paper concerns the issue of ethical evaluation of algorithms, especially those used in social media and which create profiles of users of these media and new technologies that have recently emerged and are intended to change the functioning of technologies used in data management. Systems such as Overton, SambaNova or Snorkel were created to help engineers create data management models, but they are based on different assumptions than the previous approach in machine learning (...) and deep learning. There is a need to analyze both deep learning algorithms and new technologies in database management in terms of their actions towards a person who leaves their digital footprints, on which these technologies work. Then, the possibilities of applying the existing deep learning technology and new Big Data systems in the economy will be shown. The opportunities offered by the systems mentioned above seem to be promising for many companies and—if implemented on a larger scale—they will affect the functioning of the free market. (shrink)

Although existing work draws attention to a range of obstacles in realizing fair AI, the field lacks an account that emphasizes how these worries hang together in a systematic way. Furthermore, a review of the fair AI and philosophical literature demonstrates the unsuitability of ‘treat like cases alike’ and other intuitive notions as conceptions of fairness. That review then generates three desiderata for a replacement conception of fairness valuable to AI research: (1) It must provide a meta-theory for understanding tradeoffs, (...) entailing that it must be flexible enough to capture diverse species of objection to decisions. (2) It must not appeal to an impartial perspective (neutral data, objective data, or final arbiter.) (3) It must foreground the way in which judgments of fairness are sensitive to context, i.e., to historical and institutional states of affairs. We argue that a conception of fairness as appropriate concession in the historical iteration of institutional decisions meets these three desiderata. On the basis of this definition, we organize the insights of commentators into a process-structure map of the ethical territory that we hope will bring clarity to computer scientists and ethicists analyzing Fair AI while clearing some ground for further technical and philosophical work. (shrink)

We argue that explainable artificial intelligence (XAI), specifically reason-giving XAI, often constitutes the most suitable way of ensuring that someone can properly be held responsible for decisions that are based on the outputs of artificial intelligent (AI) systems. We first show that, to close moral responsibility gaps (Matthias 2004), often a human in the loop is needed who is directly responsible for particular AI-supported decisions. Second, we appeal to the epistemic condition on moral responsibility to argue that, in order to (...) be responsible for her decision, the human in the loop has to have an explanation available of the system’s recommendation. Reason explanations are especially well-suited to this end and we examine whether – and how – it might be possible to make such explanations fit with AI systems. We support our claims by focusing on a case of disagreement between human in the loop and AI system. (shrink)

There is a long history of the science of intelligent machines and its potential to provide scientific insights have been debated since the dawn of AI. In particular, there is renewed interest in the role of AI in research and research policy as an enabler of new methods, processes, management and evaluation which is still relatively under-explored. This empirical paper explores interviews with leading scholars on the potential impact of AI on research practice and culture through deductive, thematic analysis to (...) show the issues affecting academics and universities today. Our interviewees identify positive and negative consequences for research and researchers with respect to _collective_ and _individual use_. AI is perceived as helpful with respect to information gathering and other narrow tasks, and in support of impact and interdisciplinarity. However, using AI as a way of ‘speeding up—to keep up’ with bureaucratic and metricised processes, may proliferate negative aspects of academic culture in that the expansion of AI in research should assist and not replace human creativity. Research into the future role of AI in the research process needs to go further to address these challenges, and ask fundamental questions about how AI might assist in providing new tools able to question the values and principles driving institutions and research processes. We argue that to do this an explicit movement of meta-research on the role of AI in research should consider the effects for research and researcher creativity. Anticipatory approaches and engagement of diverse and critical voices at policy level and across disciplines should also be considered. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealisedexplanation gamein which players collaborate to find the best explanation(s) for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal patterns of (...) variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a (conditionally) optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

Traditional medical practices and relationships are changing given the widespread adoption of AI-driven technologies across the various domains of health and healthcare. In many cases, these new technologies are not specific to the field of healthcare. Still, they are existent, ubiquitous, and commercially available systems upskilled to integrate these novel care practices. Given the widespread adoption, coupled with the dramatic changes in practices, new ethical and social issues emerge due to how these systems nudge users into making decisions and changing (...) behaviours. This article discusses how these AI-driven systems pose particular ethical challenges with regards to nudging. To confront these issues, the value sensitive design approach is adopted as a principled methodology that designers can adopt to design these systems to avoid harming and contribute to the social good. The AI for Social Good factors are adopted as the norms constraining maleficence. In contrast, higher-order values specific to AI, such as those from the EU High-Level Expert Group on AI and the United Nations Sustainable Development Goals, are adopted as the values to be promoted as much as possible in design. The use case of Amazon Alexa's Healthcare Skills is used to illustrate this design approach. It provides an exemplar of how designers and engineers can begin to orientate their design programs of these technologies towards the social good. (shrink)

According to a recent survey by the HR Research Institute, as the presence of artificial intelligence (AI) becomes increasingly common in the workplace, HR professionals are worried that the use of recruitment algorithms will lead to a “dehumanization” of the hiring process. Our main goals in this paper are threefold: i) to bring attention to this neglected issue, ii) to clarify what exactly this concern about dehumanization might amount to, and iii) to sketch an argument for why dehumanizing the hiring (...) process is ethically suspect. After distinguishing the use of the term “dehumanization” in this context (i.e., removing the human presence) from its more common meaning in the interdisciplinary field of dehumanization studies (i.e., conceiving of other humans as subhuman), we argue that the use of hiring algorithms may negatively impact the employee-employer relationship. We argue that there are good independent reasons to accept a substantive employee-employer relationship, as well as an applicant-employer relationship, both of which are consistent with a stakeholder theory of corporate obligations. We further argue that dehumanizing the hiring process may negatively impact these relationships because of the difference between the values of human recruiters and the values embedded in recruitment algorithms. Drawing on Nguyen’s (2021) critique of how Twitter “gamifies communication”, we argue that replacing human recruiters with algorithms imports artificial values into the hiring process. We close by briefly considering some ways to potentially mitigate the problems posed by recruitment algorithms, along with the possibility that some difficult trade-offs will need to be made. (shrink)

The primary goal of this essay is to provide a comprehensive overview and analysis of the various relations between material artifacts and the embodied mind. A secondary goal of this essay is to identify some of the trends in the design and use of artifacts. First, based on their functional properties, I identify four categories of artifacts co-opted by the embodied mind, namely (1) embodied artifacts, (2) perceptual artifacts, (3) cognitive artifacts, and (4) affective artifacts. These categories can overlap and (...) so some artifacts are members of more than one category. I also identify some of the techniques (or skills) we use when interacting with artifacts. Identifying these categories of artifacts and techniques allows us to map the landscape of relations between embodied minds and the artifactual world. Second, having identified categories of artifacts and techniques, this essay then outlines some of the trends in the design and use of artifacts, focussing on neuroprosthetics, brain-computer interfaces, and personalisation algorithms nudging their users towards particular epistemic paths of information consumption. (shrink)

The notion of “responsibility gap” with artificial intelligence (AI) was originally introduced in the philosophical debate to indicate the concern that “learning automata” may make more difficult or impossible to attribute moral culpability to persons for untoward events. Building on literature in moral and legal philosophy, and ethics of technology, the paper proposes a broader and more comprehensive analysis of the responsibility gap. The responsibility gap, it is argued, is not one problem but a set of at least four interconnected (...) problems – gaps in culpability, moral and public accountability, active responsibility—caused by different sources, some technical, other organisational, legal, ethical, and societal. Responsibility gaps may also happen with non-learning systems. The paper clarifies which aspect of AI may cause which gap in which form of responsibility, and why each of these gaps matter. It proposes a critical review of partial and non-satisfactory attempts to address the responsibility gap: those which present it as a new and intractable problem (“fatalism”), those which dismiss it as a false problem (“deflationism”), and those which reduce it to only one of its dimensions or sources and/or present it as a problem that can be solved by simply introducing new technical and/or legal tools (“solutionism”). The paper also outlines a more comprehensive approach to address the responsibility gaps with AI in their entirety, based on the idea of designing socio-technical systems for “meaningful human control", that is systems aligned with the relevant human reasons and capacities. (shrink)

Computers are used to make decisions in an increasing number of domains. There is widespread agreement that some of these uses are ethically problematic. Far less clear is where ethical problems arise, and what might be done about them. This paper expands and defends the Ethical Gravity Thesis: ethical problems that arise at higher levels of analysis of an automated decision-making system are inherited by lower levels of analysis. Particular instantiations of systems can add new problems, but not ameliorate more (...) general ones. We defend this thesis by adapting Marr’s famous 1982 framework for understanding information-processing systems. We show how this framework allows one to situate ethical problems at the appropriate level of abstraction, which in turn can be used to target appropriate interventions. (shrink)

Research on the ethics of algorithms has grown substantially over the past decade. Alongside the exponential development and application of machine learning algorithms, new ethical problems and solutions relating to their ubiquitous use in society have been proposed. This article builds on a review of the ethics of algorithms published in 2016, 2016). The goals are to contribute to the debate on the identification and analysis of the ethical implications of algorithms, to provide an updated analysis of epistemic and normative (...) concerns, and to offer actionable guidance for the governance of the design, development and deployment of algorithms. (shrink)

Previous research in Explainable Artificial Intelligence (XAI) suggests that a main aim of explainability approaches is to satisfy specific interests, goals, expectations, needs, and demands regarding artificial systems (we call these “stakeholders' desiderata”) in a variety of contexts. However, the literature on XAI is vast, spreads out across multiple largely disconnected disciplines, and it often remains unclear how explainability approaches are supposed to achieve the goal of satisfying stakeholders' desiderata. This paper discusses the main classes of stakeholders calling for explainability (...) of artificial systems and reviews their desiderata. We provide a model that explicitly spells out the main concepts and relations necessary to consider and investigate when evaluating, adjusting, choosing, and developing explainability approaches that aim to satisfy stakeholders' desiderata. This model can serve researchers from the variety of different disciplines involved in XAI as a common ground. It emphasizes where there is interdisciplinary potential in the evaluation and the development of explainability approaches. (shrink)

In recent years, the ethical impact of AI has been increasingly scrutinized, with public scandals emerging over biased outcomes, lack of transparency, and the misuse of data. This has led to a growing mistrust of AI and increased calls for mandated ethical audits of algorithms. Current proposals for ethical assessment of algorithms are either too high level to be put into practice without further guidance, or they focus on very specific and technical notions of fairness or transparency that do not (...) consider multiple stakeholders or the broader social context. In this article, we present an auditing framework to guide the ethical assessment of an algorithm. The audit instrument itself is comprised of three elements: a list of possible interests of stakeholders affected by the algorithm, an assessment of metrics that describe key ethically salient features of the algorithm, and a relevancy matrix that connects the assessed metrics to stakeholder interests. The proposed audit instrument yields an ethical evaluation of an algorithm that could be used by regulators and others interested in doing due diligence, while paying careful attention to the complex societal context within which the algorithm is deployed. (shrink)

Transparency is now a fundamental principle for data processing under the General Data Protection Regulation. We explore what this requirement entails for artificial intelligence and automated decision-making systems. We address the topic of transparency in artificial intelligence by integrating legal, social, and ethical aspects. We first investigate the ratio legis of the transparency requirement in the General Data Protection Regulation and its ethical underpinnings, showing its focus on the provision of information and explanation. We then discuss the pitfalls with respect (...) to this requirement by focusing on the significance of contextual and performative factors in the implementation of transparency. We show that human–computer interaction and human-robot interaction literature do not provide clear results with respect to the benefits of transparency for users of artificial intelligence technologies due to the impact of a wide range of contextual factors, including performative aspects. We conclude by integrating the information- and explanation-based approach to transparency with the critical contextual approach, proposing that transparency as required by the General Data Protection Regulation in itself may be insufficient to achieve the positive goals associated with transparency. Instead, we propose to understand transparency relationally, where information provision is conceptualized as communication between technology providers and users, and where assessments of trustworthiness based on contextual factors mediate the value of transparency communications. This relational concept of transparency points to future research directions for the study of transparency in artificial intelligence systems and should be taken into account in policymaking. (shrink)

In this article, we develop the concept of Transparency by Design that serves as practical guidance in helping promote the beneficial functions of transparency while mitigating its challenges in automated-decision making environments. With the rise of artificial intelligence and the ability of AI systems to make automated and self-learned decisions, a call for transparency of how such systems reach decisions has echoed within academic and policy circles. The term transparency, however, relates to multiple concepts, fulfills many functions, and holds different (...) promises that struggle to be realized in concrete applications. Indeed, the complexity of transparency for ADM shows tension between transparency as a normative ideal and its translation to practical application. To address this tension, we first conduct a review of transparency, analyzing its challenges and limitations concerning automated decision-making practices. We then look at the lessons learned from the development of Privacy by Design, as a basis for developing the Transparency by Design principles. Finally, we propose a set of nine principles to cover relevant contextual, technical, informational, and stakeholder-sensitive considerations. Transparency by Design is a model that helps organizations design transparent AI systems, by integrating these principles in a step-by-step manner and as an ex-ante value, not as an afterthought. (shrink)

The use of artificial intelligence in healthcare has led to debates about the role of human clinicians in the increasingly technological contexts of medicine. Some researchers have argued that AI will augment the capacities of physicians and increase their availability to provide empathy and other uniquely human forms of care to their patients. The human vulnerabilities experienced in the healthcare context raise the stakes of new technologies such as AI, and the human dimensions of AI in healthcare have particular significance (...) for research in the humanities. This article explains four key areas of concern relating to AI and the role that medical/health humanities research can play in addressing them: definition and regulation of “medical” versus “health” data and apps; social determinants of health; narrative medicine; and technological mediation of care. Issues include data privacy and trust, flawed datasets and algorithmic bias, racial discrimination, and the rhetoric of humanism and disability. Through a discussion of potential humanities contributions to these emerging intersections with AI, this article will suggest future scholarly directions for the field. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal (...) patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

Advances in AI are powering increasingly precise and widespread computational propaganda, posing serious threats to national security. The military and intelligence communities are starting to discuss ways to engage in this space, but the path forward is still unclear. These developments raise pressing ethical questions, about which existing ethics frameworks are silent. Understanding these challenges through the lens of “cognitive security,” we argue, offers a promising approach.

Advanced AI systems are rapidly making their way into medical research and practice, and, arguably, it is only a matter of time before they will surpass human practitioners in terms of accuracy, reliability, and knowledge. If this is true, practitioners will have a prima facie epistemic and professional obligation to align their medical verdicts with those of advanced AI systems. However, in light of their complexity, these AI systems will often function as black boxes: the details of their contents, calculations, (...) and procedures cannot be meaningfully understood by human practitioners. When AI systems reach this level of complexity, we can also speak of black-box medicine. In this paper, we want to argue that black-box medicine conflicts with core ideals of patient-centered medicine. In particular, we claim, black-box medicine is not conducive for supporting informed decision-making based on shared information, shared deliberation, and shared mind between practitioner and patient. (shrink)

The debate about the ethical implications of Artificial Intelligence dates from the 1960s :741–742, 1960; Wiener in Cybernetics: or control and communication in the animal and the machine, MIT Press, New York, 1961). However, in recent years symbolic AI has been complemented and sometimes replaced by Neural Networks and Machine Learning techniques. This has vastly increased its potential utility and impact on society, with the consequence that the ethical debate has gone mainstream. Such a debate has primarily focused on principles—the (...) ‘what’ of AI ethics —rather than on practices, the ‘how.’ Awareness of the potential issues is increasing at a fast rate, but the AI community’s ability to take action to mitigate the associated risks is still at its infancy. Our intention in presenting this research is to contribute to closing the gap between principles and practices by constructing a typology that may help practically-minded developers apply ethics at each stage of the Machine Learning development pipeline, and to signal to researchers where further work is needed. The focus is exclusively on Machine Learning, but it is hoped that the results of this research may be easily applicable to other branches of AI. The article outlines the research method for creating this typology, the initial findings, and provides a summary of future research needs. (shrink)