Herman Veluwenkamp

Public administrations use predictive systems to decide whom to investigate for fraud or for other violations. In this paper, I show that many of these systems distribute the risk of exposure to the state’s coercive powers. These systems are often assessed in terms of familiar notions of algorithmic fairness, and are treated as analogous to systems that simply allocate scarce goods. I argue that this way of looking at these tools is morally misleading, because coercion requires a more demanding justification. The paper uses Scanlonian contractualism and argues that unequal coercive risk is permissible only if it is justified by reasons that can be addressed to the particular person who bears that risk. These reasons are often absent in practice. Population-level correlations or expected gains in accuracy count only insofar as they can be personalized into individualist reasons. However, in the administrative contexts that I discuss in the paper, they will typically be insufficient to justify imposing greater coercive risk on some persons than on others. If there are no stronger admissible reasons to justify differential selection, the default should therefore be an equal-chance lottery. To strengthen the argument, I add two supporting claims. It is not evident that profiling can sustain deterrence better than lotteries, and lotteries are in some respects epistemically superior.

Recent years have seen increasing attention to conceptual engineering, the enterprise of evaluating and improving our repertoire of representational devices (e.g., words, concepts, classificatory procedures). This paper contributes to the systematic investigation of the normative questions that arise within and about conceptual engineering. It does two things. First, it argues for a clear distinction between two normative domains that are often conflated: conceptual ethics and ethics of implementation. Conceptual ethics concerns what concepts we ought to have, and the ethics of implementation addresses what means are permissible in promoting conceptual uptake. It is important to keep these domains separate, as we would otherwise risk conflating or obscuring what kinds of normative concerns are relevant at what stage and why. Second, the paper investigates how these two normative domains interact. Specifically, it asks whether what it is permissible to do to increase the uptake of a representational device matters to the question what representational devices people ought to have, or vice versa. The paper argues that conceptual ethics can directly influence the ethics of implementation, but that the ethics of implementation has much less relevance to conceptual ethics.

Conceptual ethics asks which concepts we ought to use. Matthieu Queloz’s recent book, _The Ethics of Conceptualization_, offers the most developed account to date. He argues that reasons for concept choice should be grounded in our conceptual needs, which he understands as reflectively endorsed concerns, capacities, and circumstances. His case proceeds through a two-tier argument that rejects foundationalism, ironism, and holism, as well as approaches that rely exclusively on epistemic or theoretical virtues. We argue that Queloz’s dismissal of foundationalism is too quick. By relating his framework to the familiar distinction between applied, normative, and meta-ethics, we clarify the meta-ethical commitments implicit in his position. This, in turn, allows us to extract four desiderata that any plausible theory of conceptual ethics must meet: Non-Arbitrary Authority, Comprehensive Value Sensitivity, Context-Sensitivity, and Compatibility with the Modern World. We develop and defend an alternative foundationalist view that meets all four desiderata and avoids the objections Queloz levels against traditional foundationalism. This “proper foundationalism” grounds reasons for concept choice in objective values without assuming that all concepts must track metaphysical structure. The disagreement between Queloz and his foundationalist rivals, we argue, is best understood as a dispute within meta-ethics. Recognizing this clarifies the structure of conceptual ethics and expands the theoretical options available for guiding conceptual improvement.

What values and goals should guide conceptual engineering projects? In this paper, we propose that insights from responsible design, specifically Value Sensitive Design (VSD), can enrich current approaches to conceptual ethics. Philosophers of technology have long employed VSD as a structured way to create technologies that address real‐world problems while accommodating stakeholder values. Meanwhile, conceptual engineers have focused on how best to revise, introduce, or eliminate concepts in response to theoretical or practical needs. By bringing these two literatures together, this paper offers a systematic, empirically informed way to assess and design concepts. The approach uses VSD‐inspired methods to identify and weigh stakeholder values, goals, and concerns. To illustrate how this works in practice, the paper examines the technologically disrupted concept “colleague.” When advanced technologies, like robots or chatbots, begin performing roles similar to those of human workers, can they be considered colleagues?

A number of concerns have been recently raised regarding the possibility of human agents to effectively maintain control over intelligent and (partially) autonomous artificial systems. These issues have been deemed to raise “responsibility gaps.” To address these gaps, several scholars and other public and private stakeholders converged towards the idea that, in deploying intelligent technology, a meaningful form of human control (MHC) should be at all times exercised over autonomous intelligent technology. One of the main criticisms of the general idea of MHC is that it could be inherently problematic to have high degrees of control and high degrees of autonomy at the same time, as the two dimensions appear to be inversely related. Several ways to respond to this argument and deal with the dilemma between control and autonomy have been proposed in the literature. In this paper, we further contribute to the philosophical effort to overcome the trade-off between automation and human control, and to open up some space for moral responsibility. We will use the instrument of conceptual engineering to investigate whether and to what extent removing the element of direct causal intervention from the concept of control can preserve the main functions of that concept, specifically focusing on the extent it can act as a foundation of moral responsibility. We show that at least one philosophical account of MHC is indeed a conceptually viable theory to absolve the fundamental functions of control, even in the context of completely autonomous artificial systems.

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.

Responsibility gaps traditionally refer to scenarios in which no one is responsible for harm caused by artificial agents, such as autonomous machines or collective agents. By carefully examining the different ways this concept has been defined in the social ontology and ethics of technology literature, I argue that our current concept of responsibility gaps is defective. To address this conceptual flaw, I argue that the concept of responsibility gaps should be revised by distinguishing it into two more precise concepts: epistemic responsibility gaps (ERG) and control misalignments (CM). ERG functions as a refined version addressing specific aspects of responsibility attribution. It captures the challenge of identifying who is to blame. CM, on the other hand, introduces a new dimension by focusing on misalignments of control. These revisions offer better guidelines for the development and oversight of artificial agents, both in terms of assigning blame and ensuring the right distribution of control.

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.

In this special issue, we focus on the connection between conceptual engineering and the philosophy of technology. Conceptual engineering is the enterprise of introducing, eliminating, or revising words and concepts. The philosophy of technology examines the nature and significance of technology. We investigate how technologies such as AI and genetic engineering (so-called “socially disruptive technologies”) disrupt our practices and concepts, and how conceptual engineering can address these disruptions. We also consider how conceptual engineering can enhance the practice of ethical design. The issue features seven articles that discuss a range of topics, including trust in blockchain applications and the evolving concept of nature. These articles highlight that as technology changes the world and our concepts, conceptual engineering provides invaluable tools and frameworks to reflect on these changes and adapt accordingly.

Conceptual engineering is the design, evaluation and implementation of concepts. Despite its popularity, some have argued that the methodology is not worthwhile, because the implementation of new concepts is both inscrutable and beyond our control. In the recent literature we see different responses to this worry. Some have argued that it is for political reasons just as well that implementation is such a difficult task, while others have challenged the metasemantic and social assumptions that underlie this skepticism about implementation. In this paper, I argue that even if implementation is as difficult as critics maintain, there is at least one context in which conceptual engineering is extremely impactful and demonstrably so: the design of new technology. Different conceptions of control, freedom, trust, etc. lead to different designs and implementations of systems that are built to embed those concepts. This means that if we want to design for control, freedom, trust, etc., we have to decide which conception we ought to use. When we determine what the appropriate conception of a concept is in a technological context and use this conception to operationalize a norm or value, we generate requirements which have real-world effects. This not only shows that conceptual engineering can be extremely impactful, the fact that it leads to different design requirements means that we have a way to evaluate our conceptual choices and that we can use this feedback loop to improve upon our conceptual work By illustrating the direct impacts of different conceptions on technology design, this paper underscores the practical applicability and value of conceptual engineering, demonstrating how it serves as a crucial bridge between abstract philosophical analysis and concrete technological innovation.

Conceptual engineering is the enterprise of evaluating and improving our representational devices. But how should we conduct this enterprise? One increasingly popular answer to this question proposes that conceptual engineering should proceed in terms of the functions of our representational devices. In this paper, we argue that the best way of understanding this suggestion is in terms of normative functions, where normative functions of concepts are, roughly, things that they allow us to do that matter normatively (for example, things in virtue of which we have normative reasons to have these concepts). Not only does this introduce a novel view about functions to the literature. This proposal also fits more naturally than the alternatives with conceptual engineering as a normative enterprise, and it allows functions to play all of the explanatory roles assigned to them in the conceptual engineering literature. Our discussion of the explanatory advantages of normative functions also advances the understanding of the ways in which concepts can be authoritative, what this means for conceptual engineering, and highlights the importance of political philosophy for thinking about conceptual engineering in practice. Furthermore, the paper explicates what kind of role considerations about function can and should actually play in conceptual engineering.

Responsibility gaps concern the attribution of blame for harms caused by autonomous machines. The worry has been that, because they are artificial agents, it is impossible to attribute blame, even though doing so would be appropriate given the harms they cause. We argue that there are no responsibility gaps. The harms can be blameless. And if they are not, the blame that is appropriate is indirect and can be attributed to designers, engineers, software developers, manufacturers or regulators. The real problem lies elsewhere: autonomous machines should be built so as to exhibit a level of risk that is morally acceptable. If they fall short of this standard, they exhibit what we call ‘a control gap.’ The causal control that autonomous machines have will then fall short of the guidance control they should emulate.

Politicians and engineers are increasingly realizing that values are important in the development of technological artefacts. What is often overlooked is that different conceptualizations of these abstract values lead to different design-requirements. For example, designing social media platforms for deliberative democracy sets us up for technical work on completely different types of architectures and mechanisms than designing for so-called liquid or direct forms of democracy. Thinking about Democracy is not enough, we need to design for the proper conceptualization of these values. As we see it, we cannot responsibly engineer and innovate and shape technology in accordance with our moral values without engaging in systematic and continuous conceptual engineering: This is not only an academic, or theoretical issue, it is also not simply an issue for public policy or politics, or regulators, it has become a central problem for engineering and the world of technology. In this paper, we present a framework for doing the necessary conceptual work in the context of requirement engineering. We draw on the literature on conceptual engineering to lay out a methodology to (1) assess different conceptions and (2) to develop new conceptions. Moreover, we integrate this methodology with extant approaches in the philosophy of technology which aim at designing technological artefacts ethically. In the final section we apply this integrated framework to freedom in the context of social media networks.

”Meaningful human control” is a term invented in the political and legal debate on autonomous weapons system, but it is nowadays also used in many other contexts. It is supposed to specify conditions under which an artificial system is under the right kind of control to avoid responsibility gaps: that is, situations in which no moral agent is responsible. Santoni de Sio and Van den Hoven have recently suggested a framework that can be used by system designers to operationalize this kind of control. It is the purpose of this paper to facilitate further operationalization of ”meaningful human control”. This paper consists of two parts. In the first part I resolve an ambiguity that plagues current operationalizations of MHC. One of the design conditions says that the system should track the reasons of the relevant agents. This condition is ambiguous between the kind of reasons involved. On one interpretation it says that a system should track motivating reasons, while it is concerned with normative reasons on the other. Current participants in the debate interpret the framework as being concerned with (something in the vicinity of) motivating reasons. I argue against this interpretation by showing that meaningful human control requires that a system tracks normative reasons. Moreover, I maintain that an operationalization of meaningful human control that fails to track the right kind of reasons is morally problematic.When this is properly understood, it can be shown that the framework of MHC is committed to the agent-relativity of reasons. More precisely, I argue in the second part of this paper that if the tracking condition of MHC plays an important role in responsibility attribution (as the proponents of the view maintain), then the framework is incompatible with first-order normative theories that hold that normative reasons are agent-neutral (such as many versions of consequentialism). In the final section I present three ways forward for the proponent of MHC as reason-responsiveness.

New technologies are the source of uncertainties about the applicability of moral and morally connotated concepts. These uncertainties sometimes call for conceptual engineering, but it is not often recognized when this is the case. We take this to be a missed opportunity, as a recognition that different researchers are working on the same kind of project can help solve methodological questions that one is likely to encounter. In this paper, we present three case studies where philosophers of technology implicitly engage in conceptual engineering (without naming it as such). We subsequently reflect on the case studies to find out how these illustrate conceptual engineering as an appropriate method to deal with pressing concerns in the philosophy of technology. We have two main goals. We first want to contribute to the literature on conceptual engineering by presenting concrete examples of conceptual engineering in the philosophy of technology. This is especially relevant, because the technologies that are designed based on the conceptual work done by philosophers of technology potentially have crucial moral and social implications. Secondly, we want to make explicit what choices are made when doing this conceptual work. Making explicit that some of the implicit assumptions are, in fact, debated in the literature allows for reflection on these questions. Ultimately, our hope is that conscious reflection leads to an improvement of the conceptual work done.

A range of formal models of human reasoning have been proposed in a number of fields such as philosophy, logic, artificial intelligence, computer science, psychology, cognitive science, etc.: various logics, probabilistic systems, belief revision systems, neural networks, among others. Now, it seems reasonable to require that formal models of human reasoning be empirically adequate if they are to be viewed as models of the phenomena in question. How are formal models of human reasoning typically put to empirical test? One way to do so is to isolate a number of key principles of the system, and design experiments to gauge the extent to which participants do or do not follow them in reasoning tasks. Another way is to take relevant existing results and check whether a particular formal model predicts these results. The present investigation provides an illustration of the second kind of empirical testing by comparing two formal models for reasoning – namely the non-monotonic logic known as preferential logic; and a particular version of belief revision theories, screened belief revision – against the reasoning phenomenon known as belief bias in the psychology of reasoning literature: human reasoners typically seek to maintain the beliefs they already hold, and conversely to reject contradicting incoming information. The conclusion of our analysis will be that screened belief revision is more empirically adequate with respect to belief bias than preferential logic and non-monotonic logics in general, as what participants seem to be doing is above all a form of belief management on the basis of background knowledge. The upshot is thus that, while it may offer valuable insights into the nature of human reasoning, preferential logic is ultimately inadequate as a formal model of the phenomena in question.

Users of sociotechnical systems often have no way to independently verify whether the system output which they use to make decisions is correct; they are epistemically dependent on the system. We argue that this leads to problems when the system is wrong, namely to bad decisions and violations of the norm of practical reasoning. To prevent this from occurring we suggest the implementation of defeaters: information that a system is unreliable in a specific case (undercutting defeat) or independent information that the output is wrong (rebutting defeat). Practically, we suggest to design defeaters based on the different ways in which a system might produce erroneous outputs, and analyse this suggestion with a case study of the risk classification algorithm used by the Dutch tax agency.

Metasemantic inferentialismhas gained popularity in the last few decades. Traditionally, inferentialism is combined with a deflationary attitude towards semantic terms such as truth and reference, i.e., many inferentialists hold that when we use these semantic terms we do not purport to refer to substantive properties. This combination makes inferentialism attractive for philosophers who see themselves as antirealists. Although the attractions of combining inferentialism and deflationism are easy to see, deflationism is also a controversial position. For one, deflationists maintain that truth is an insubstantive property, but it is not altogether clear what an insubstantive property is. Secondly, as deflationists maintain that truth does not play an explanatory role, it is incompatible with the position that truth can explain the normativity of truth talk. Given that deflationism faces these objections, it would be preferable if the success of inferentialism did not depend on the deflationist’s ability to respond to these objections. I argue that someone attracted to inferentialism for its ability to accommodate antirealist intuitions about a domain (e.g. morality) is not committed to deflationism about truth. More specifically, I will show that inferentialism combined with a straightforward account of inferentialist truth-conditions is compatible with a version of truth pluralism. I call this position Inferentialist Truth Pluralism.

Thomas Scanlon and Derek Parfit have recently defended a meta-ethical view that is supposed to satisfy realistic intuitions about morality, without the metaphysical implications that many find hard to accept in other realist views. Both philosophers argue that truths in the normative domain do not have ontological implications, while truths in the scientific domain presuppose a metaphysical reality. What distinguishes Scanlon and Parfit’s approach from other realistic meta-ethical theories is that they maintain that normative entities exist in a way that is different from (some) non-normative entities. Moreover, they think that this view about the way normative entities exist helps to answer the metaphysical worries that are often thought to plague non-naturalism. To make sense of this idea, I develop their view as a version of alethic pluralism: the position that there is more than one truth property. I argue, however, that when their view is developed in this way it fails to satisfy realistic intuitions about morality. This shows that on a plausible and initially promising reading of what it takes for a normative entity to exist in a way that is different from non-normative entities, Parfit’s and Scanlon’s non-metaphysical moral realism fails to be more realistic than contemporary versions of anti-realism.