Many ethicists writing about automated systems attribute agency to these systems. Not only that; they seemingly attribute an autonomous or independent form of agency to these machines. This leads some ethicists to worry about responsibility-gaps and retribution-gaps in cases where automated systems harm or kill human beings. In this paper, I consider what sorts of agency it makes sense to attribute to most current forms of automated systems, in particular automated cars and military robots. I argue that whereas it indeed (...) makes sense to attribute different forms of fairly sophisticated agency to these machines, we ought not to regard them as acting on their own, independently of any human beings. Rather, the right way to understand the agency exercised by these machines is in terms of human–robot collaborations, where the humans involved initiate, supervise, and manage the agency of their robotic collaborators. This means, I argue, that there is much less room for justified worries about responsibility-gaps and retribution-gaps than many ethicists think. (shrink)

In a landmark essay, Andreas Matthias claimed that current developments in autonomous, artificially intelligent (AI) systems are creating a so-called responsibility gap, which is allegedly ever-widening and stands to undermine both the moral and legal frameworks of our society. But how severe is the threat posed by emerging technologies? In fact, a great number of authors have indicated that the fear is thoroughly instilled. The most pessimistic are calling for a drastic scaling-back or complete moratorium on AI systems, while the (...) optimists aim to show that the gap can be bridged nonetheless. Contrary to both camps, I argue against the prevailing assumption that there is a technology-based responsibility gap. I show how moral responsibility is a dynamic and flexible process, one that can effectively encompass emerging technological entities. (shrink)

Recent decades have witnessed tremendous progress in artificial intelligence and in the development of autonomous systems that rely on artificial intelligence. Critics, however, have pointed to the difficulty of allocating responsibility for the actions of an autonomous system, especially when the autonomous system causes harm or damage. The highly autonomous behavior of such systems, for which neither the programmer, the manufacturer, nor the operator seems to be responsible, has been suspected to generate responsibility gaps. This has been the cause of (...) much concern. In this article, I propose a more optimistic view on artificial intelligence, raising two challenges for responsibility gap pessimists. First, proponents of responsibility gaps must say more about when responsibility gaps occur. Once we accept a difficult-to-reject plausibility constraint on the emergence of such gaps, it becomes apparent that the situations in which responsibility gaps occur are unclear. Second, assuming that responsibility gaps occur, more must be said about why we should be concerned about such gaps in the first place. I proceed by defusing what I take to be the two most important concerns about responsibility gaps, one relating to the consequences of responsibility gaps and the other relating to violations of jus in bello. (shrink)

Self-driving cars hold out the promise of being safer than manually driven cars. Yet they cannot be a 100 % safe. Collisions are sometimes unavoidable. So self-driving cars need to be programmed for how they should respond to scenarios where collisions are highly likely or unavoidable. The accident-scenarios self-driving cars might face have recently been likened to the key examples and dilemmas associated with the trolley problem. In this article, we critically examine this tempting analogy. We identify three important ways (...) in which the ethics of accident-algorithms for self-driving cars and the philosophy of the trolley problem differ from each other. These concern: the basic decision-making situation faced by those who decide how self-driving cars should be programmed to deal with accidents; moral and legal responsibility; and decision-making in the face of risks and uncertainty. In discussing these three areas of disanalogy, we isolate and identify a number of basic issues and complexities that arise within the ethics of the programming of self-driving cars. (shrink)

Future weapons will make life-or-death decisions without a human in the loop. When such weapons inflict unwarranted harm, no one appears to be responsible. There seems to be a responsibility gap. I first reconstruct the argument for such responsibility gaps to then argue that this argument is not sound. The argument assumes that commanders have no control over whether autonomous weapons inflict harm. I argue against this assumption. Although this investigation concerns a specific case of autonomous weapons systems, I take (...) steps towards vindicating the more general idea that superiors can be morally responsible in virtue of being in command. (shrink)

Artificial agents create significant moral opportunities and challenges. Over the last two decades, discourse has largely focused on the concept of a ‘responsibility gap.’ We argue that this concept is incoherent, misguided, and diverts attention from the core issue of ‘control gaps.’ Control gaps arise when there is a discrepancy between the causal control an agent exercises and the moral control it should possess or emulate. Such gaps present moral risks, often leading to harm or ethical violations. We propose a (...) second-order ‘duty of moral control’ that mandates closing these gaps to reduce risks within acceptable moral limits. Our analysis encompasses both autonomous machines and collective agents, acknowledging their similarities and key differences in constitution and moral status. We suggest four methods to close control gaps: ensuring artificial agents attain moral agency, providing meaningful human control, implementing safety engineering, and employing social control. These methods aim to responsibly integrate artificial agents into society. We conclude that a realistic approach, which addresses the practical problems posed by control gaps, is essential. This approach provides solutions to manage the risks posed by artificial agents while maintaining acceptable moral standards, ensuring we responsibly harness their potential and address the ethical challenges they present. (shrink)

This paper proposes a methodological redirection of the philosophical debate on artificial moral agency in view of increasingly pressing practical needs due to technological development. This “normative approach” suggests abandoning theoretical discussions about what conditions may hold for moral agency and to what extent these may be met by artificial entities such as AI systems and robots. Instead, the debate should focus on how and to what extent such entities should be included in human practices normally assuming moral agency and (...) responsibility of participants. The proposal is backed up by an analysis of the AMA debate, which is found to be overly caught in the opposition between so-called standard and functionalist conceptions of moral agency, conceptually confused and practically inert. Additionally, we outline some main themes of research in need of attention in light of the suggested normative approach to AMA. (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)

Self‐driving cars hold out the promise of being much safer than regular cars. Yet they cannot be 100% safe. Accordingly, we need to think about who should be held responsible when self‐driving cars crash and people are injured or killed. We also need to examine what new ethical obligations might be created for car users by the safety potential of self‐driving cars. The article first considers what lessons might be learned from the growing legal literature on responsibility for crashes with (...) self‐driving cars. Next, worries about responsibility gaps and retribution gaps from the philosophical literature are introduced. This leads to a discussion of whether self‐driving cars are a form of agents that act independently of human agents. It is suggested that it is better to analyze their apparent agency in terms of human–robot collaborations, within which humans play the most important roles. The next topic is the idea that the safety potential of self‐driving cars might create a duty to either switch to self‐driving cars or seek means of making conventional cars safer. Lastly, there is a short discussion of ethical issues related to safe human–robot coordination within mixed traffic featuring both self‐driving cars and conventional cars. (shrink)

The development of highly humanoid sex robots is on the technological horizon. If sex robots are integrated into the legal community as “electronic persons”, the issue of sexual consent arises, which is essential for legally and morally permissible sexual relations between human persons. This paper explores whether it is conceivable, possible, and desirable that humanoid robots should be designed such that they are capable of consenting to sex. We consider reasons for giving both “no” and “yes” answers to these three (...) questions by examining the concept of consent in general, as well as critiques of its adequacy in the domain of sexual ethics; the relationship between consent and free will; and the relationship between consent and consciousness. Additionally we canvass the most influential existing literature on the ethics of sex with robots. (shrink)

The increasing demand for transparency in AI has recently come under scrutiny. The question is often posted in terms of “epistemic double standards”, and whether the standards for transparency in AI ought to be higher than, or equivalent to, our standards for ordinary human reasoners. I agree that the push for increased transparency in AI deserves closer examination, and that comparing these standards to our standards of transparency for other opaque systems is an appropriate starting point. I suggest that a (...) more fruitful exploration of this question will involve a different comparison class. We routinely treat judgments made by highly trained experts in specialized fields as fair or well-grounded even though—by the nature of expert/layperson division of epistemic labor—an expert will not be able to provide an explanation of the reasoning behind these judgments that makes sense to most other people. Regardless, laypeople are thought to be acting reasonably—and ethically—in deferring to the judgments of experts that concern their areas of specialization. I suggest that we reframe our question regarding the appropriate standards of transparency in AI as one that asks when, why, and to what degree it would be ethical to accept opacity in AI. I argue that our epistemic relation to certain opaque AI technology may be relevantly similar to the layperson’s epistemic relation to the expert in certain respects, such that the successful expert/layperson division of epistemic labor can serve as a blueprint for the ethical use of opaque AI. (shrink)

There have been two recent strands of argument arguing for the pro tanto impermissibility of fully autonomous weapon systems. On Sparrow's view, AWS are impermissible because they generate a morally problematic ‘responsibility gap’. According to Purves et al., AWS are impermissible because moral reasoning is not codifiable and because AWS are incapable of acting for the ‘right’ reasons. I contend that these arguments are flawed and that AWS are not morally problematic in principle. Specifically, I contend that these arguments presuppose (...) an incoherent conception of an AWS as somehow making genuine decisions but not being morally responsible for those very same decisions. Rather than conceiving of AWS in this way, I argue that an AWS is either a socially-constructed institution that has been physically instantiated or it is a genuine agent. If it is the former, then we should treat AWS as we do any other collective action problem. If it is the latter, then we should treat AWS as responsibility-bearers, but also as bearers of rights and/or interests. To reject this disjunction is not only conceptually incoherent but also potentially morally dangerous. (shrink)

Advances in artificial intelligence research allow us to build fairly sophisticated agents: robots and computer programs capable of acting and deciding on their own. These systems raise questions about who is responsible when something goes wrong—when such systems harm or kill humans. In a recent paper, Sven Nyholm has suggested that, because current AI will likely possess what we might call “supervised agency”, the theory of responsibility for individual agency is the wrong place to look for an answer to the (...) question of responsibility. Instead, or so argues Nyholm, because supervised agency is a form of collaborative agency—of acting together—the right place to look is the theory of collaborative responsibility—responsibility in cases of acting together. This paper concedes that current AI will possess supervised agency, but argues that it is nevertheless wrong to think of the relevant human-AI interactions as a form of collaborative agency and, hence, that responsibility in cases of collaborative agency is not the right place to look for the responsibility-grounding relation in human-AI interactions. It also suggests that the right place to look for this responsibility-grounding relation in human-AI interactions is the use of certain sorts of agents as instruments. (shrink)

Virtue ethics has many times been suggested as a promising recipe for the construction of artificial moral agents due to its emphasis on moral character and learning. However, given the complex nature of the theory, hardly any work has de facto attempted to implement the core tenets of virtue ethics in moral machines. The main goal of this paper is to demonstrate how virtue ethics can be taken all the way from theory to machine implementation. To achieve this goal, we (...) critically explore the possibilities and challenges for virtue ethics from a computational perspective. Drawing on previous conceptual and technical work, we outline a version of artificial virtue based on moral functionalism, connectionist bottom–up learning, and eudaimonic reward. We then describe how core features of the outlined theory can be interpreted in terms of functionality, which in turn informs the design of components necessary for virtuous cognition. Finally, we present a comprehensive framework for the technical development of artificial virtuous agents and discuss how they can be implemented in moral environments. (shrink)

Virtue ethics has many times been suggested as a promising recipe for the construction of artificial moral agents due to its emphasis on moral character and learning. However, given the complex nature of the theory, hardly any work has de facto attempted to implement the core tenets of virtue ethics in moral machines. The main goal of this paper is to demonstrate how virtue ethics can be taken all the way from theory to machine implementation. To achieve this goal, we (...) critically explore the possibilities and challenges for virtue ethics from a computational perspective. Drawing on previous conceptual and technical work, we outline a version of artificial virtue based on moral functionalism, connectionist bottom–up learning, and eudaimonic reward. We then describe how core features of the outlined theory can be interpreted in terms of functionality, which in turn informs the design of components necessary for virtuous cognition. Finally, we present a comprehensive framework for the technical development of artificial virtuous agents and discuss how they can be implemented in moral environments. (shrink)

To many, the idea of autonomous weapons systems (AWS) killing human beings is grotesque. Yet critics have had difficulty explaining why it should make a significant moral difference if a human combatant is killed by an AWS as opposed to being killed by a human combatant. The purpose of this paper is to explore the roots of various deontological concerns with AWS and to consider whether these concerns are distinct from any concerns that also apply to long-distance, human-guided weaponry. We (...) suggest that at least one major driver of the intuitive moral aversion to lethal AWS is that their use disrespects their human targets by violating the martial contract between human combatants. On our understanding of this doctrine, service personnel cede a right not to be directly targeted with lethal violence to other human agents alone. Artificial agents, of which AWS are one example, cannot understand the value of human life. A human combatant cannot transfer his privileges of targeting enemy combatants to a robot. Therefore, the human duty-holder who deploys AWS breaches the martial contract between human combatants and disrespects the targeted combatants. We consider whether this novel deontological objection to AWS forms the foundation of several other popular yet imperfect deontological objections to AWS. (shrink)

How should automated vehicles (AVs) react in emergency circumstances? Most research projects and scientific literature deal with this question from a moral perspective. In particular, it is customary to treat emergencies involving AVs as instances of moral dilemmas and to use the trolley problem as a framework to address such alleged dilemmas. Some critics have pointed out some shortcomings of this strategy and have urged to focus on mundane traffic situations instead of trolley cases involving AVs. Besides, these authors rightly (...) point out the political nature of the most interesting debates involving AVs. However, in our view, they do not offer an adequate account of the distinction between ethics and politics and still see their proposals as contributions to the ethics of AVs. We argue that many of the interesting questions about how AVs should behave, both in emergency and other situations, are of political, and not moral, nature. This view is based on a conception of politics and political normativity that we have developed elsewhere and that we call “political minimalism.” Additionally, we show that this proposed perspective has significant consequences for the design, management, and regulation of transport systems. (shrink)

Warfare is becoming increasingly automated, from automatic missile defense systems to micro-UAVs (WASPs) that can maneuver through urban environments with ease, and each advance brings with it ethical questions in need of resolving. Proponents of lethal autonomous weapons systems (LAWS) provide varied arguments in their favor; robots are capable of better identifying combatants and civilians, thus reducing "collateral damage"; robots need not protect themselves and so can incur more risks to protect innocents or gather more information before using deadly force; (...) robots can assess situations more quickly and do so without emotion, reducing the likelihood of fatal mistakes due to human error; and sending robots to war protects our own soldiers from harm. However, these arguments only point in favor of autonomous weapons systems, failing to demonstrate why such systems need be made *lethal*. In this paper I argue that if one grants all of the proponents' points in favor of LAWS, then, contrary to what might be expected, this leads to the conclusion that it would be both immoral and illegal to deploy *lethal* autonomous weapons, because the many features that speak in favor of them also undermine the need for them to be programmed to take lives. In particular, I argue that such systems, if lethal, would violate the moral and legal principle of necessity, which forbids the use of weapons that impose superfluous injury or unnecessary harm. I conclude by highlighting that the argument is not against autonomous weapons per se, but only against *lethal* autonomous weapons. (shrink)

In September 2015 a well-publicised Campaign Against Sex Robots (CASR) was launched. Modelled on the longer-standing Campaign to Stop Killer Robots, the CASR opposes the development of sex robots on the grounds that the technology is being developed with a particular model of female-male relations (the prostitute-john model) in mind, and that this will prove harmful in various ways. In this chapter, we consider carefully the merits of campaigning against such a technology. We make three main arguments. First, we argue (...) that the particular claims advanced by the CASR are unpersuasive, partly due to a lack of clarity about the campaign’s aims and partly due to substantive defects in the main ethical objections put forward by campaign’s founder(s). Second, broadening our inquiry beyond the arguments proferred by the campaign itself, we argue that it would be very difficult to endorse a general campaign against sex robots unless one embraced a highly conservative attitude towards the ethics of sex, which is likely to be unpalatable to those who are active in the campaign. In making this argument we draw upon lessons from the campaign against killer robots. Finally, we conclude by suggesting that although a generalised campaign against sex robots is unwarranted, there are legitimate concerns that one can raise about the development of sex robots. (shrink)

Recent advancements in artificial intelligence have fueled widespread academic discourse on the ethics of AI within and across a diverse set of disciplines. One notable subfield of AI ethics is machine ethics, which seeks to implement ethical considerations into AI systems. However, since different research efforts within machine ethics have discipline-specific concepts, practices, and goals, the resulting body of work is pestered with conflict and confusion as opposed to fruitful synergies. The aim of this paper is to explore ways to (...) alleviate these issues, both on a practical and theoretical level of analysis. First, we describe two approaches to machine ethics: the philosophical approach and the engineering approach and show how tensions between the two arise due to discipline specific practices and aims. Using the concept of disciplinary capture, we then discuss potential promises and pitfalls to cross-disciplinary collaboration. Drawing on recent work in philosophy of science, we finally describe how metacognitive scaffolds can be used to avoid epistemological obstacles and foster innovative collaboration in AI ethics in general and machine ethics in particular. (shrink)

Due to advances in military technology, there has been an outpouring of research on what are known as autonomous weapon systems (AWS). However, it is common in this literature for arguments to be made without first making clear exactly what definitions one is employing, with the detrimental effect that authors may speak past one another or even miss the targets of their arguments. In this article I examine the U.S. Department of Defense and International Committee of the Red Cross definitions (...) of AWS, showing that these definitions are far broader than some recognize, and that they therefore classify a much larger set of weapons as AWS. I then show that these broader views of AWS have implications for what moral and legal rules we might argue should be applied to such systems. I conclude by arguing there is a greater need for precision and clarity within AWS debates, in order to ensure that researchers are discussing the same weapon systems (autonomous or otherwise) when they argue for or against particular points. The purpose of this article is not to defend any specific view of AWS, nor to further any general endorsement or objection to such systems, but rather to show the importance of argumentative clarity in this debate. (shrink)

This article is a result of a unique project that brought together academics and military practitioners with a mind to addressing difficult moral questions in a way that is philosophically careful,...

Artificial intelligence and robotic technologies have become nearly ubiquitous. In some ways, the developments have likely helped us, but in other ways sophisticated technologies set back our interests. Among the latter sort is what has been dubbed the ‘severance problem’—the idea that technologies sever our connection to the world, a connection which is necessary for us to flourish and live meaningful lives. I grant that the severance problem is a threat we should mitigate and I ask: how can we stave (...) it off? In particular, the fact that some technologies exhibit behavior that is unclear to us seems to constitute a kind of severance. Building upon contemporary work on moral responsibility, I argue for a mechanism I refer to as ‘technological answerability’, namely the capacity to recognize human demands for answers and to respond accordingly. By designing select devices—such as robotic assistants and personal AI programs—for increased answerability, we see at least one way of satisfying our demands for answers and thereby retaining our connection to a world increasingly occupied by technology. (shrink)

Autonomous systems are machines that can alter their behavior without direct human oversight or control. How ought we to program them to behave? A plausible starting point is given by the Reduction to Acts Thesis, according to which we ought to program autonomous systems to do whatever a human agent ought to do in the same circumstances. Although the Reduction to Acts Thesis is initially appealing, we argue that it is false: it is sometimes permissible to program a machine to (...) do something that it would be wrong for a human to do. We advance two main arguments for this claim. First, the way an autonomous system will behave can be known in advance. This knowledge can indirectly affect the behavior of other agents, while the same may not be true at the time the system actually executes its programming. Second, a lack of knowledge of the identities of the victims and beneficiaries can provide a justification during the programming phase that would be unavailable to an agent at the time the autonomous system executes its programming. (shrink)

I here consider the reasonableness of punishing future autonomous military robots. I argue that it is an engineering desideratum that these devices be responsive to moral considerations as well as human criticism and blame. Additionally, I argue that someday it will be possible to build such machines. I use these claims to respond to the no subject of punishment objection to deploying autonomous military robots, the worry being that an “accountability gap” could result if the robot committed a war crime. (...) (Sparrow 2007). But these robots will be blameworthy and morally responsible for their wrongful conduct. I then show that we have reasons to punish them based both on good effects and the impact on the robot, itself. (shrink)

It is an increasingly popular view among philosophers that moral responsibility can, in principle, be attributed to unconscious autonomous agents. This trend is already remarkable in itself, but it is even more interesting that most proponents of this view provide more or less the same argument to support their position. I argue that as it stands, the Extension Argument, as I call it, is not sufficient to establish the thesis that unconscious autonomous agents can be morally responsible. I attempt to (...) show that the Extension Argument should overcome especially strong ethical considerations; moreover, its epistemological grounds are not too solid, partly because the justifications of its premises are in conflict. (shrink)

Who should decide how a machine will decide what to do when it is driving a car, performing a medical procedure, or, more generally, when it is facing any kind of morally laden decision? More and more, machines are making complex decisions with a considerable level of autonomy. We should be much more preoccupied by this problem than we currently are. After a series of preliminary remarks, this paper will go over four possible answers to the question raised above. First, (...) we may claim that it is the maker of a machine that gets to decide how it will behave in morally laden scenarios. Second, we may claim that the users of a machine should decide. Third, that decision may have to be made collectively or, fourth, by other machines built for this special purpose. The paper argues that each of these approaches suffers from its own shortcomings, and it concludes by showing, among other things, which approaches should be emphasized for different types of machines, situations, and/or morally laden decisions. (shrink)

Robert Sparrow argues that several initially plausible arguments in favor of the deployment of autonomous weapons systems (AWS) in warfare fail, and that their deployment faces a serious moral objection: deploying AWS fails to express the respect for the casualties of war that morality requires. We critically discuss Sparrow’s argument from respect and respond on behalf of some objections he considers. Sparrow’s argument against AWS relies on the claim that they are distinct from accepted weapons of war in that they (...) either fail to transmit an attitude of respect or they transmit an attitude of disrespect. We argue that this distinction between AWS and widely accepted weapons is illusory, and so cannot ground a moral difference between AWS and existing methods of waging war. We also suggest that, if deploying conventional soldiers in some situation would be permissible, and if we could expect deploying AWS to cause fewer civilian casualties, then it would be consistent with an intuitive understanding of respect to deploy AWS in this situation. (shrink)

In this paper I argue, contrary to recent literature, that it is unethical to create artificial agents possessing human-level intelligence that are programmed to be human beings’ obedient servants. In developing the argument, I concede that there are possible scenarios in which building such artificial servants is, on net, beneficial. I also concede that, on some conceptions of autonomy, it is possible to build human-level AI servants that will enjoy full-blown autonomy. Nonetheless, the main thrust of my argument is that, (...) in building such artificial agents, their creators cannot help but evince an objectionable attitude akin to the Aristotelian vice of manipulativeness. (shrink)

Can autonomous systems replace humans in the performance of their activities? How does the answer to this question inform the design of autonomous systems? The study of technical systems and their features should be preceded by the study of the activities in which they play roles. Each activity can be described by its overall goals, governing norms and the intermediate steps which are taken to achieve the goals and to follow the norms. This paper uses the activity realist approach to (...) conceptualize autonomous systems in the context of human activities. By doing so, it first argues for epistemic and logical conditions that illustrate the limitations of autonomous systems in tasks which they can and cannot perform, and then, it discusses the ramifications of the limitations of system autonomy on the design of autonomous systems. (shrink)

This paper introduces the concept of Operator Contestability in AI systems: the principle that those overseeing AI systems (operators) must have the necessary control to be accountable for the decisions made by these algorithms. We argue that designers have a duty to ensure operator contestability. We demonstrate how this duty can be fulfilled by applying the'Design for Defeaters' framework, which provides strategies to embed tools within AI systems that enable operators to challenge decisions. Defeaters are designed to contest either the (...) justification for the AI’s data inputs (undercutting defeaters) or the validity of the conclusions drawn from that data (rebutting defeaters). To illustrate the necessity and application of this framework, we examine case studies such as AI-driven recruitment processes, where operators need tools and authority to uncover and address potential biases, and autonomous driving systems, where real-time decision-making is crucial. The paper argues that operator contestability requires ensuring that operators have (1) epistemic access to the relevant normative reasons and (2) the authority and cognitive capacity to act on these defeaters. By addressing these challenges, the paper emphasizes the importance of designing AI systems in a way that enables operators to effectively contest AI decisions, thereby ensuring that the appropriate individuals can take responsibility for the outcomes of human-AI interactions. (shrink)

Autonomous vehicles, and the larger socio-technical systems that they are a part of are likely to have a deep and lasting impact on our societies. Trust is a key value that will play a role in the development of autonomous driving systems. This paper suggests that trust of autonomous driving systems will impact the ways that these systems are taken up, the norms and laws that guide them and the design of the systems themselves. Further to this, in order to (...) have autonomous driving systems that are worthy of our trust, we need a superstructure of oversight and a process that designs trust into these systems from the outset. Rather than banning or avoiding all autonomous vehicles should a tragedy occur, despite these systems having some level of risk, we want resilient systems that can survive tragedies, and indeed, improve from them. I will argue that trust plays a role in developing these resilient systems. (shrink)

Ethics settings allow for morally significant decisions made by humans to be programmed into autonomous machines, such as autonomous vehicles or autonomous weapons. Customizable ethics settings are a type of ethics setting in which the users of autonomous machines make such decisions. Here two arguments are provided in defence of customizable ethics settings. Firstly, by approaching ethics settings in the context of failure management, it is argued that customizable ethics settings are instrumentally and inherently valuable for building resilience into the (...) larger socio-technical systems in which autonomous machines operate. Secondly, after defining the preliminary condition of responsibility attribution and demonstrating how ethics settings enable humans to exert control over the outcomes of morally significant incidents, it is shown that ethics settings narrow the responsibility gap. (shrink)

As opaque algorithmic systems take up a larger and larger role in shaping our lives, calls for explainability in various algorithmic systems have increased. Many moral and political philosophers have sought to vindicate these calls for explainability by developing theories on which decision-subjects—that is, individuals affected by decisions—have a moral right to the explanation of the systems that affect them. Existing theories tend to suggest that the right to explanation arises solely in virtue of facts about how decision-subjects are affected (...) by opaque systems. But this ignores an important part of the normative landscape: sometimes, facts about a decision-maker’s rights and freedoms can make a difference to whether she owes anyone explanations of her decisions. I illustrate the point by calling attention to examples of artistic, romantic, and supererogatory altruistic decisions; in at least some cases of this sort, even highly influential decisions can legitimately be kept entirely private and opaque. (shrink)

Arguments from human dignity feature prominently in the Lethal Autonomous Weapons moral feasibility debate, even though their exists considerable controversy over their role and soundness and the notion of dignity remains under-defined. Drawing on the work of Dieter Birnbacher, I fix the sub-discourse as referring to the essential value of human persons in general, and to postulated moral rights of combatants not covered within the existing paradigm of the International Humanitarian Law in particular. I then review and critique dignity-based arguments (...) against LAWS: argument from faulty targeting process, argument from objectification, argument from underappreciation of the value of human life and the argument from the absence of mercy. I conclude that the argument from the absence of mercy is the only dignity-based argument that is both valid and irreducible to another class of arguments within the debate, and that it offers insufficient justification for a global ban on LAWS. (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)

Autonomous systems are machines that can alter their behavior without direct human oversight or control. How ought we to program them to behave? A plausible starting point is given by the Reduction to Acts Thesis, according to which we ought to program autonomous systems to do whatever a human agent ought to do in the same circumstances. Although the Reduction to Acts Thesis is initially appealing, we argue that it is false: it is sometimes permissible to program a machine to (...) do something that it would be wrong for a human to do. We advance two main arguments for this claim. First, the way an autonomous system will behave can be known in advance. This knowledge can indirectly affect the behavior of other agents, while the same may not be true at the time the system actually executes its programming. Second, a lack of knowledge of the identities of the victims and beneficiaries can provide a justification during the programming phase that would be unavailable to an agent at the time the autonomous system executes its programming. (shrink)

Many have recently argued that there are weighty reasons against making high-stakes decisions solely on the basis of recommendations from artificially intelligent (AI) systems. Even if deference to a given AI system were known to reliably result in the right action being taken, the argument goes, that deference would lack morally important characteristics: the resulting decisions would not, for instance, be based on an appreciation of right-making reasons. Nor would they be performed from moral virtue; nor would they have moral (...) worth. I argue that, even if these characteristics all have intrinsic value, that intrinsic value has no practical relevance to decisions about whether to defer to AI. I make that point by drawing on a lesson from the literature on moral testimony. Once it is granted that deference to a reliable source is the policy most likely to bring about right action, a refusal to defer carries with it a heightened risk of wronging and mistreating people. And that heightened risk of wrongdoing, I argue, cannot be justified by appeal to the intrinsic value of striving for a morally exemplary decision-making process. (shrink)

Raul Hakli & Pekka Mäkelä (2016, 2019) make a popular assumption in machine ethics explicit by arguing that artificial agents cannot be responsible because they are designed. Designed agents, they think, are analogous to manipulated humans and therefore not meaningfully in control of their actions. Contrary to this, I argue that under all mainstream theories of responsibility, designed agents can be responsible. To do so, I identify the closest parallel discussion in the literature on responsibility and free will, which concerns (...) ‘design cases’. Design cases are theoretical examples of agents that appear to lack responsibility because they were designed, philosophers use these cases to explore the relationship between design and responsibility. This paper presents several replies to design cases from the responsibility literature and uses those replies to situate the corresponding positions on the design and responsibility of artificial agents in machine ethics. I argue that each reply can support the design of responsible agents. However, each reply also entails different levels of severity in the constraints for the design of responsible agents. I offer a brief discussion of the nature of those constraints, highlighting the challenges respective to each reply. I conclude that designing responsible agents is possible, with the caveat that the difficulty of doing so will vary according to one’s favoured reply to design cases. (shrink)

With the rapid expansion of parameters in large language models (LLMs) and the application of Reinforcement Learning with Human Feedback (RLHF), there has been a noticeable growth in the moral competence of LLMs. However, several questions warrant further exploration: Is it really possible for LLMs to fully align with human values through RLHF? How can the current moral growth be philosophically contextualized? We identify similarities between LLMs’ moral growth and Deweyan ethics in terms of the discourse of human moral development. (...) We then attempt to use Dewey’s theory on an experimental basis to examine and further explain the extent to which the current alignment pathway enables the development of LLMs. A beating experiment serves as the foundational case for analyzing LLMs’ moral competence across various parameters and stages, including basic moral cognition, moral dilemma judgment, and moral behavior. The results demonstrate that the moral competence of the GPT series has seen a significant improvement, and Dewey’s Impulse-Habit-Character theory of moral development can be used to explain this: the moral competence of LLMs has been enhanced through experience-based learning, supported by human feedback. Nevertheless, LLMs’ moral development through RLHF remains constrained and does not reach the character stage described by Dewey, possibly due to their lack of self-consciousness. This fundamental difference between humans and LLMs underscores both the limitations of LLMs’ moral growth and the challenges of applying RLHF for AI alignment. It also emphasizes the need for external societal governance and legal regulation. (shrink)

This paper concerns top-down approaches in machine ethics. It is divided into three main parts. First, I briefly describe top-down design approaches, and in doing so I make clear what those approaches are committed to and what they involve when it comes to training an AI to behave ethically. In the second part, I formulate two underappreciated motivations for endorsing them, one relating to predictability of machine behavior and the other relating to scrutability of machine decision-making. Finally, I present three (...) major worries about such approaches, and I attempt to show that advocates of top-down approaches have some plausible avenues of response. I focus most of my attention on what I call the ‘technical manual objection’ to top-down approaches, inspired by the work of Annas (2004). In short, the idea is that top-down approaches treat ethical decision-making as being merely a matter of following some ethical instructions in the same way that one might follow some set of instructions contained in a technical manual (e.g., computer manual), and this invites sensible skepticism about the ethical wisdom of machines that have been trained on those approaches. I respond by claiming that the objection is successful only if it is understood as targeting machines that have certain kinds of goals, and it should not compel us to totally abandon top-down approaches. Such approaches could still be reasonably employed to design ethical AI that operate in contexts that include fairly noncontroversial answers to ethical questions. In fact, we should prefer top-down approaches when it comes to those types of context, or so I argue, due to the advantages I claim for them. (shrink)

The rapid technological advancement of AI in the civilian sector is accompanied by accelerating attempts to apply this technology in the military sector. This study focuses on the argument that AI-equipped lethal autonomous weapons systems (LAWS) pose a threat to human dignity. However, the precise meaning of why and how LAWS violate human dignity is not always clear because the concept of human dignity itself remains ambiguous. Drawing on philosophical research on this concept, this study distinguishes the multiple meanings of (...) dignity via five components—Agency, End-not-means, Uniqueness, Reciprocal respect, and Specialness—to organize existing dignity-based objections to LAWS. This study also extends the literature by providing empirical evidence of people’s attitudes toward LAWS, based on the results of a social survey conducted by the authors. This study concludes that attention must be directed toward the ontological distinction between humans and machines in advancing the ethical debate on LAWS. (shrink)

Autonomous Weapon Systems (AWS) are artificial intelligence systems that can make and act on decisions concerning the termination of enemy soldiers and installations without direct intervention from a human being. In this article, I provide the positive moral case for the development and use of supervised and fully autonomous weapons that can reliably adhere to the laws of war. Two strong, prima facie obligations make up the positive case. First, we have a strong moral reason to deploy AWS (in an (...) otherwise just war) because such systems decrease the psychological and moral risk of soldiers and would-be soldiers. Drones protect against lethal risk, AWS protect against psychological and moral risk in addition to lethal risk. Second, we have a prima facie obligation to develop such technologies because, once developed, we could employ forms of non-lethal warfare that would substantially reduce the risk of suffering and death for enemy combatants and civilians alike. These two arguments, covering both sides of a conflict, represent the normative hill that those in favor of a ban on autonomous weapons must overcome. Finally, I demonstrate that two recent objections to AWS fail because they misconstrue the way in which technology is used and conceptualized in modern warfare. (shrink)

Why should moral philosophers, moral psychologists, and machine ethicists care about computational complexity? Debates on whether artificial intelligence (AI) can or should be used to solve problems in ethical domains have mainly been driven by what AI can or cannot do in terms of human capacities. In this paper, we tackle the problem from the other end by exploring what kind of moral machines are possible based on what computational systems can or cannot do. To do so, we analyze normative (...) ethics through the lens of computational complexity. First, we introduce computational complexity for the uninitiated reader and discuss how the complexity of ethical problems can be framed within Marr’s three levels of analysis. We then study a range of ethical problems based on consequentialism, deontology, and virtue ethics, with the aim of elucidating the complexity associated with the problems themselves (e.g., due to combinatorics, uncertainty, strategic dynamics), the computational methods employed (e.g., probability, logic, learning), and the available resources (e.g., time, knowledge, learning). The results indicate that most problems the normative frameworks pose lead to tractability issues in every category analyzed. Our investigation also provides several insights about the computational nature of normative ethics, including the differences between rule- and outcome-based moral strategies, and the implementation-variance with regard to moral resources. We then discuss the consequences complexity results have for the prospect of moral machines in virtue of the trade-off between optimality and efficiency. Finally, we elucidate how computational complexity can be used to inform both philosophical and cognitive-psychological research on human morality by advancing the moral tractability thesis. (shrink)

Recent discussion on the ethics of killer robots has focused on the supposed lack of respect their deployment would show to combatants targeted, thereby causing their undignified deaths. I present two rebuttals of this argument. The weak rebuttal maintains that while deploying killer robots is an affront to the dignity of combatants, their use should nevertheless be thought of as a pro tanto wrong, making deployment permissible if the affront is outweighed by some right-making feature. This rebuttal is, however, vulnerable (...) to the charge that killer robots would cause unnecessary suffering. The strong rebuttal, in contrast, argues that the use of killer robots would not disrespect the dignity of combatants, for reasons discussed. I also argue that, irrespective of whether killer robots disrespect the dignity of combatants, they would not necessarily bring about their undignified deaths: for one can maintain dignity in the face of indignity. (shrink)

Can nonhuman animals and artificial intelligence (AI) entities be attributed moral agency? The general assumption in the philosophical literature is that moral agency applies exclusively to humans since they alone possess free will or capacities required for deliberate reflection. Consequently, only humans have been taken to be eligible for ascriptions of moral responsibility in terms of, for instance, blame or praise, moral criticism, or attributions of vice and virtue. Animals and machines may cause harm, but they cannot be appropriately ascribed (...) moral responsibility for their behavior. This thesis challenges the conventional paradigm by proposing an alternative approach where moral agency is conceived as the competence to participate in moral responsibility practices. By shifting focus from intra-individual to contextual and socially situated features, this practice-focused approach appears to make the attribution of moral agency to nonhuman animals and AI entities more plausible than commonly assumed. Moreover, considering the current and potential future prevalence of nonhuman animals and AI entities in everyday settings and social contexts, a potential extension of moral agency to such entities could very well transform our social, moral, and legal practices. Hence, this thesis proposes that the attribution or withholding of moral agency to different entities should be carefully evaluated, considering the potential normative implications. (shrink)

The ethics of emerging forms of artificial intelligence has become a prolific subject in both academic and public spheres. A great deal of these concerns flow from the need to ensure that these technologies do not cause harm—physical, emotional or otherwise—to the human agents with which they will interact. In the literature, this challenge has been met with the creation of artificial moral agents: embodied or virtual forms of artificial intelligence whose decision procedures are constrained by explicit normative principles, requiring (...) the implementation of what is commonly called artificial morality into these agents. To date, the types of reasoning structures and principles which inform artificial morality have been of two kinds: first, an ethically maximal vision of artificial morality which relies on the strict implementation of traditional moral theories such as Kantian deontology or Utilitarianism, and second, a more minimalist vision which applies stochastic AI techniques to large data sets of human moral preferences so as to illicit or intuit general principles and preferences for the design of artificial morality. Taken individually, each approach is unable to fully answer the challenge of producing inoffensive behavior in artificial moral agents, most especially since both forms are unable to strike a balance between the ideal set of constraints which morality imposes on one hand, and the types of constraints public acceptability imposes, on the other. We provide an alternative approach to the design of artificial morality, the Ethical Valence Theory, whose purpose is to accommodate this balance, and apply this approach to the case of autonomous vehicles. (shrink)