The aim of this paper is twofold: first, I provide a cluster of theories of truth in classical logic that is (internally) consistent with global reflection principles: the theories of positive truth (and falsity). After that, I analyse the _epistemic value_ of such theories. I do so employing the framework of cognitive projects introduced by Wright (Proc Aristot Soc 78:167–245, 2004), and employed—in the context of theories of truth—by Fischer et al. (Noûs 2019. https://doi.org/10.1111/nous.12292 ). In particular, I will (...) argue that theories of positive truth are _trustworthy_, analogously to the theories of full disquotational truth. Moreover, I argue that, for a given cognitive project, _if_ the acceptance of trustworthy theories is taken to be an _epistemic norm_ of cognitive project, then one has good reasons to accept theories of positive truth over other rival theories of truth in classical logic. On the other hand, the latter theories are deemed epistemically unacceptable. (shrink)

Data preparation, feature engineering, model training, and performance evaluation are all part of the study methodology. To ensure reliable and broadly applicable models, we utilize optimization techniques like Grid Search and Genetic Algorithms to precisely adjust model parameters. Features including age, blood pressure, cholesterol levels, and lifestyle choices are employed as inputs for the machine learning models in the dataset, which consists of patient medical information. The predictive capacity of the model is evaluated using evaluation measures, such as accuracy, precision, (...) recall, F1-score, and the area under the ROC curve (AUC-ROC). Our findings show that improved machine learning models perform better than conventional methods, offering trustworthy forecasts that can help medical practitioners with early diagnosis and individualized treatment planning. In order to achieve even higher predicted accuracy, the study's conclusion discusses the significance of its findings for clinical practice as well as future improvements that might be made, like adding wearable device data in real-time or investigating deep learning techniques. (shrink)

The increasing use of artificial intelligence (AI) systems in our daily life through various applications, services, and products explains the significance of trust/distrust in AI from a user perspective. AI-driven systems have significantly diffused into various fields of our lives, serving as beneficial tools used by human agents. These systems are also evolving to act as co-assistants or semi-agents in specific domains, potentially influencing human thought, decision-making, and agency. Trust/distrust in AI plays the role of a regulator and could significantly (...) control the level of this diffusion, as trust can increase, and distrust may reduce the rate of adoption of AI. Recently, varieties of studies have paid attention to the variant dimension of trust/distrust in AI, and its relevant considerations. In this systematic literature review, after conceptualization of trust in the current AI literature review, we will investigate trust in different types of human-Machine interaction, and its impact on technology acceptance in different domains. In addition to that, we propose a taxonomy of technical (i.e., safety, accuracy, robustness) and non-technical axiological (i.e., ethical, legal, and mixed) trustworthiness metrics, and some trustworthy measurements. Moreover, we examine some major trust-breakers in AI (e.g., autonomy and dignity threat), and trust makers; and propose some future directions and probable solutions for the transition to a trustworthy AI. (shrink)

Researchers in the biological and biomedical sciences, particularly those working in laboratories, use a variety of artifacts to help them perform their cognitive tasks. This paper analyses the relationship between researchers and cognitive artifacts in terms of integration. It first distinguishes different categories of cognitive artifacts used in biological practice on the basis of their informational properties. This results in a novel classification of scientific instruments, conducive to an analysis of the cognitive interactions between researchers and artifacts. It then uses (...) a multidimensional framework in line with complementarity-based extended and distributed cognition theory to conceptualize how deeply instruments in different informational categories are integrated into the cognitive systems of their users. The paper concludes that the degree of integration depends on various factors, including the amount of informational malleability, the intensity and kind of information flow between agent and artifact, the trustworthiness of the information, the procedural and informational transparency, and the degree of individualisation. (shrink)

Recent breakthroughs in artificial intelligence have primarily resulted from training deep neural networks (DNNs) with vast numbers of adjustable parameters on enormous datasets. Due to their complex internal structure, DNNs are frequently characterized as inscrutable ``black boxes,'' making it challenging to interpret the mechanisms underlying their impressive performance. This opacity creates difficulties for explanation, safety assurance, trustworthiness, and comparisons to human cognition, leading to divergent perspectives on these systems. This chapter examines recent developments in interpretability methods for DNNs, with a (...) focus on interventionist approaches inspired by causal explanation in philosophy of science. We argue that these methods offer a promising avenue for understanding how DNNs process information compared to merely behavioral benchmarking and correlational probing. We review key interventionist methods and illustrate their application through practical case studies. These methods allow researchers to identify and manipulate specific computational components within DNNs, providing insights into their causal structure and internal representations. We situate these approaches within the broader framework of causal abstraction, which aims to align low-level neural computations with high-level interpretable models. While acknowledging current limitations, we contend that interventionist methods offer a path towards more rigorous and theoretically grounded interpretability research, potentially informing both AI development and computational cognitive neuroscience. (shrink)

Why comply with epistemic norms? In this paper, I argue that complying with epistemic norms, engaging in epistemically responsible conduct, and being epistemically trustworthy are constitutive elements of maintaining good epistemic relations with oneself and others. Good epistemic relations are in turn both instrumentally and finally valuable: they enable the kind of coordination and knowledge acquisition underpinning much of what we tend to associate with a flourishing human life; and just as good interpersonal relations with others can be good (...) for their own sake, standing in good epistemic relations is good for its own sake. On my account, we have reason to comply with epistemic norms because it is a way of respecting the final value of something that also tends to be an instrumentally valuable thing: good epistemic relations. Situating the account within the recent social turn in debates about epistemic instrumentalism, I argue that the dual‐value aspect of good epistemic relations can explain important anti‐instrumentalist intuitions, in a well‐motivated way, within a broadly instrumentalist framework. (shrink)

We ordinarily assume that we have reliable knowledge of our immediate surroundings, so much so that almost all the time we entrust our lives to the truth of what we take ourselves to know, without a moment’s thought. But if, as Karl Popper and others have maintained, all our knowledge is conjectural, then this habitual assumption that our common sense knowledge of our environment is secure and trustworthy would seem to be an illusion. Popper’s philosophy of science, in particular, (...) fails to do justice to the distinction we ordinarily draw between secure knowledge and mere conjecture. But Popper’s philosophy of science, in particular his attempted solution to the problem of induction, is defective. It fails to take into account that physics, in only accepting unified theories, even though endlessly empirically more successful disunified rivals are always available, makes the persistent metaphysical assumption that all disunified theories are false. Once this point is acknowledged, it becomes clear that a new conception of scientific method is required which sees science as making a hierarchy of metaphysical assumptions concerning the comprehensibility and knowability of the universe. This provides a framework of relatively unproblematic assumptions and associated methods of science within which much more problematic assumptions and associated methods can be critically assessed and improved. This hierarchical view seems at first to intensify the problem of distinguishing certainty from conjecture, in that it emphasizes that scientific knowledge, and even humble common sense knowledge, contain usually unacknowledged cosmological conjectures. But actually it explicates the basis we have for drawing the distinction between trustworthy knowledge and mere conjecture, and even goes some way towards providing a rationale for this distinction, in so far as one exists. (shrink)

The purpose of this paper is to show that understanding AI hallucination requires an interdisciplinary approach that combines insights from epistemology and cognitive science to address the nature of AI-generated knowledge, with a terminological worry that concepts we often use might carry unnecessary presuppositions. Along with terminological issues, it is demonstrated that AI systems, comparable to human cognition, are susceptible to errors in judgement and reasoning, and proposes that epistemological frameworks, such as reliabilism, can be similarly applied to enhance the (...) trustworthiness of AI outputs. This exploration seeks to deepen our understanding of the possibility of AI cognition and its implications for the broader philosophical questions of knowledge and intelligence. (shrink)

Background: Use of patients’ medical data for secondary purposes such as health research, audit, and service planning is well established in the UK. However, the governance environment, as well as public understanding about this work, have lagged behind. We aimed to systematically review the literature on UK and Irish public views of patient data used in research, critically analysing such views though an established biomedical ethics framework, to draw out potential strategies for future good practice guidance and inform ethical (...) and privacy debates. Methods: We searched three databases using terms such as patient, public, opinion, and electronic health records. Empirical studies were eligible for inclusion if they surveyed healthcare users, patients or the public in UK and Ireland and examined attitudes, opinions or beliefs about the use of patient data for medical research. Results were synthesised into broad themes using a framework analysis. Results: Out of 13,492 papers and reports screened, 20 papers or reports were eligible. While there was a widespread willingness to share patient data for research for the common good, this very rarely led to unqualified support. The public expressed two generalised concerns about the potential risks to their privacy. The first of these concerns related to a party’s competence in keeping data secure, while the second was associated with the motivation a party might have to use the data. Conclusions: The public evaluates trustworthiness of research organisations by assessing their competence in data-handling and motivation for accessing the data. Public attitudes around data-sharing exemplified several principles which are also widely accepted in biomedical ethics. This provides a framework for understanding public attitudes, which should be considered in the development in any guidance for regulators and data custodians. We propose four salient questions which decision makers should address when evaluating proposals for the secondary use of data. (shrink)

Research misconduct remains an important problem in health research despite decades of local, national, regional, and international efforts to eliminate it. The ultimate goal of every health research project, irrespective of setting, is to produce trustworthy findings to address local as well as global health issues. To be able to lead or participate meaningfully in international research collaborations, individual and institutional capacities for research integrity are paramount. Accordingly, this paper concerns itself not only with individuals’ research skills but also (...) with institutional and national policies and governance. Such policies and governance provide an ethical scaffold for the production of knowledge and structure incentives. This paper’s operational definition of research therefore draws from Institute of Medicine’s articulation of health research as an inquiry that aims to produce knowledge about the structure, processes, or effects of personal health services; and from an existing health systems framework. The paper reviews the research regulatory environment and the ethics apparatus in Ghana, and describes a project jointly undertaken by Ghanaian researchers in collaboration with New York University to assess the perceived adequacy of current institutional practices, opportunities, and incentives for promoting RI. (shrink)

In this chapter, we argue that trust can be better understood in relation to people’s attempts to deal with vulnerability in social interactions. Different situations afford different forms of adaptation that correspond to different forms of trust. We describe three forms of trust: trust as a decision, trust as a performance and trust as an uncontrollable force. We show how these different types of trust differ with respect to assumptions about trust, trustworthiness and agency as well as with respect to (...) the accompanying emotions and outline a tentative process model of trust. We conclude the chapter by reviewing the contribution of the suggested framework as well as indicating questions for further research. (shrink)

This paper discusses the epistemology of the Internet of Things [IoT] by focusing on the topic of trust. It presents various frameworks of trust, and argues that the ethical framework of trust is what constitutes our responsibility to reveal desired norms and standards and embed them in other frameworks of trust. The first section briefly presents the IoT and scrutinizes the scarce philosophical work that has been done on this subject so far. The second section suggests that the field (...) of epistemology is not sufficiently capable of dealing with technologies, and presents a possible solution to this problem. It is argued that knowledge is not only social phenomena, but also a technological one, and that in order to address epistemological issues in technology, we need to carefully depart from traditional epistemic analysis and form a new approach that is technological (termed here Techno-Epistemology). The third and fourth sections engage in an epistemic analysis of trust by dividing it in to various frameworks. The last section argues that these various frameworks of trust can be understood to form a trustworthy large-scale socio-technological system, emphasizing the place of ethical trust as constituting our commitment to give proper accounts for all of the other frameworks. (shrink)

This paper presents the theory of epistemic safety in relation to three problems: similarity, closure, and generality. Within the neo-Moorean framework of skepticism, the epistemic safety theory complements contextualist theories, where a difference is established between sceptical-thought and everyday contexts. In this way, it is claimed that conviction–i.e., when the bases upon which a belief is constructed remain unquestioned–is an intellectual virtue that makes trustworthy processes in near worlds possible. Finally, the aim of the paper is to highlight (...) the modal difference between beliefs that presuppose near worlds as their core and those that take nearby words as their core. (shrink)

Australia is a leading AI nation with strong allies and partnerships. Australia has prioritised the development of robotics, AI, and autonomous systems to develop sovereign capability for the military. Australia commits to Article 36 reviews of all new means and method of warfare to ensure weapons and weapons systems are operated within acceptable systems of control. Additionally, Australia has undergone significant reviews of the risks of AI to human rights and within intelligence organisations and has committed to producing ethics guidelines (...) and frameworks in Security and Defence. Australia is committed to OECD’s values-based principles for the responsible stewardship of trustworthy AI as well as adopting a set of National AI ethics principles. While Australia has not adopted an AI governance framework specifically for Defence; Defence Science has published ‘A Method for Ethical AI in Defence’ (MEAID) technical report which includes a framework and pragmatic tools for managing ethical and legal risks for military applications of AI. (shrink)

Since anonymous agents can spread misinformation with impunity, many people advocate for greater accountability for internet speech. This paper provides a veritistic argument that accountability mechanisms can cause significant epistemic problems for internet encyclopedias and social media communities. I show that accountability mechanisms can undermine both the dissemination of true beliefs and the detection of error. Drawing on social psychology and behavioral economics, I suggest alternative mechanisms for increasing the trustworthiness of internet communication.

Situationists such as John Doris, Gilbert Harman, and Maria Merritt suppose that appeal to reliable behavioral dispositions can be dispensed with without radical revision to morality as we know it. This paper challenges this supposition, arguing that abandoning hope in reliable dispositions rules out genuine trust and forces us to suspend core reactive attitudes of gratitude and resentment, esteem and indignation. By examining situationism through the lens of trust we learn something about situationism (in particular, the radically revisionary moral implications (...) of its adoption) as well as something about trust (in particular, that the conditions necessary for genuine trust include a belief in a capacity for robust dispositions). (shrink)

Abstract: We examine how trustworthy behaviour can be achieved in the financial sector. The task is to ensure that firms are motivated to pursue long-term interests of customers rather than pursuing short-term profits. Firms’ self-interested pursuit of reputation, combined with regulation, is often not sufficient to ensure that this happens. We argue that trustworthy behaviour requires that at least some actors show a concern for the wellbeing of clients, or a respect for imposed standards, and that the behaviour (...) of these actors is copied in such a way that it becomes a behavioural norm. We briefly suggest what such behavioural norms might need to be if trustworthy behaviour is to be achieved, and consider how they might be supported; we describe the research that is necessary in order to understand these norms in more detail. We argue that the norms of traders are different from the norms of those engaged in other activities, since they are inevitably self-interested, and we consider the risk that traders’ norms might undermine those of other actors. We analyse the task for governance in dealing with this problem, and the role which leadership by a corporate board and management might play in doing this. We describe the need for further research in to describe how this might be done. (shrink)

This essay analyses the Confucian conception of xin, an attribute that broadly resembles what we would ordinarily call trustworthiness. More specifically, it provides an analysis of the psychology of someone who is xin and highlights a feature of the Confucian conception of trustworthiness: the trustworthy person has to ensure that there is a match between her self-presentation and the way she is. My goal is not to argue against any of the existing accounts of trustworthiness but to draw on (...) Confucian insights so as to shed light on features of trustworthiness that are overlooked in current discussions. I hope to show that the Confucian conception of trustworthiness puts more emphasis on the way a trustworthy person actively tries to make sure another’s dependency on her is not unwarranted than on how the trustworthy person responds to the one who gives trust. (shrink)

Trust can be thought of as a three place relation: A trusts B to do X. Trustworthiness has two components: competence (does the trustee have the relevant skills, knowledge and abilities to do X?) and willingness (is the trustee intending or aiming to do X?). This chapter is about the willingness component, and the different motivations that a trustee may have for fulfilling trust. The standard assumption in economics is that agents are self-regarding, maximizing their own consumption of goods and (...) services. This is too restrictive. In particular, people may be concerned with the outcomes of others, and they may be concerned to follow ethical principles. I distinguish weak trustworthiness, which places no restrictions on B’s motivation for doing X, from strong trustworthiness, where the behaviour must have a particular non-selfish motivation, in finance the fiduciary commitment to promote the interests of the truster. I discuss why strong trustworthiness may be more efficient and also normatively preferable to weak. In finance, there is asymmetric information between buyer and seller, which creates a need for trustworthy assessment of products. It also creates an ambiguity about whether the relationship is one of buyer and seller, governed by caveat emptor, or a fiduciary relationship of advisor and client. This means that there are two possible reasons why trust may be breached: because the trustee didn’t realise that the truster framed the relationship as a fiduciary one, or because the trustee did realise but actively sought to take advantage of the trust. Correspondingly, there are two possible types of agent: normal people who are not always self-regarding and who are trust responsive (if they believe that they are being trusted then they are likely to fulfill that trust), and knaves, after Hume’s character who is always motivated by his own private interest. We can increase the trustworthiness of normal people by getting them to re-frame the situation as one of trust, so they will be strongly trustworthy (i.e. change of institutional culture), and by providing non-monetary incentives (the correct choice of incentive will depend on exactly what their non-selfish motivation is). Knaves need sanctions, which can make them weakly trustworthy. However, this is a delicate balance because sanctions can crowd out normative frames. We can also increase the trustworthiness of financiers by making finance less attractive to knaves; changing the mix of types in finance could help support the necessary cultural change. (shrink)

A widespread assumption in debates about trust and trustworthiness is that the evaluative norms of principal interest on the trustor’s side of a cooperative exchange regulate trusting attitudes and performances whereas those on the trustee’s side regulate dispositions to respond to trust. The aim here will be to highlight some unnoticed problems with this asymmetrical picture – and in particular, how it elides certain key evaluative norms on both the trustor’s and trustee’s side the satisfaction of which are critical to (...) successful cooperative exchanges – and to show that a symmetrical, ‘achievement-first’ approach to theorising about trust and trustworthiness (and their relation to each other) has important advantages by comparison. The view I develop is guided by a structural analogy with practical reasoning. Just as practical reasoning is working as it should only when there is realisation (knowledge and action) of states (belief and intention) with reverse directions of fits (mind-to-world and world-to-mind), likewise, cooperation between trustor and trustee is functioning as it should only when there is an analogous kind of realisation on both sides of the cooperative exchange – viz., when the trustor ‘matches’ her achievement in trusting (an achievement in fitting reliance to reciprocity) with the trustee’s achievement in responding to trust (an achievement in fitting reciprocity to reliance). An upshot of viewing cooperation between trustor and trustee as exhibiting achievement-theoretic structure is that we will be better positioned to subsume trustworthiness (and its cognates on the trustee’s side), like trust, under a wider suite of evaluative norms that regulate attempts, dispositions, and achievements symmetrically on both sides of a cooperative exchange, with ‘matching achievements’ as the gold standard. (shrink)

Overconfidence is typically damaging to one’s reputation as a trustworthy source of information. Previous research shows that the reputational cost associated with conveying a piece of false information is higher for confident than unconfident speakers. When judging speaker trustworthiness, individuals do not exclusively rely on past accuracy but consider the extent to which speakers expressed a degree of confidence that matched the accuracy of their claims (their “confidence-accuracy calibration”). The present study experimentally examines the interplay between confidence, accuracy and (...) a third factor, namely evidence, in the assessment of speaker trustworthiness. Experiment 1 probes the hypothesis that overconfidence does not backfire when a confident but inaccurate claim is justified: the trustworthiness of a confident speaker who turns out to be wrong is restored if the confidence expressed is based on strong evidence (good confidence-evidence calibration). Experiment 2 investigates the hypothesis that confidence can backfire if a confident and accurate claim is not justified: the trustworthiness of a confident speaker who turns out to be right is damaged if the confidence expressed is based on weak evidence (bad confidence-evidence calibration). Our results support both hypotheses and thus suggest that “confidence-evidence calibration” plays a crucial role in the assessment of speaker trustworthiness. (shrink)

This paper examines how science advice can provide policy recommendations in a trustworthy manner. Despite their major political importance, expert recommendations are understudied in the philosophy of science and social epistemology. Matthew Bennett has recently developed a notion of what he calls recommendation trust, according to which well-placed trust in experts’ policy recommendations requires that recommendations are aligned with the interests of the trust-giver. While interest alignment might be central to some cases of public trust, this paper argues against (...) the significance of interest-alignment to meritorious public trust. First, political bodies and citizens can have a basic kind of well-placed recommendation trust in science advice based on an all-things-considered judgement regarding the possession of relevant competencies, responsible conduct, and a proper institutional design. Moreover, scientists’ policy recommendations can be seen as open-ended and as dynamic proposals that enable inter-institutional reasoning and political deliberation. Finally, by providing conditional recommendations, scientists can expand the scope of feasible policy options from which political bodies can choose, thus making the condition of interest alignment even less significant to the trustworthy provision of recommendations. (shrink)

This paper proposes a formal account of AI trustworthiness, connecting both intrinsic and perceived trustworthiness in an operational schematization. We argue that trustworthiness extends beyond the inherent capabilities of an AI system to include significant influences from observers' perceptions, such as perceived transparency, agency locus, and human oversight. While the concept of perceived trustworthiness is discussed in the literature, few attempts have been made to connect it with the intrinsic trustworthiness of AI systems. Our analysis introduces a novel schematization to (...) quantify trustworthiness by assessing the discrepancies between expected and observed behaviors and how these affect perceived uncertainty and trust. The paper provides a formalization for measuring trustworthiness, taking into account both perceived and intrinsic characteristics. By detailing the factors that influence trust, this study aims to foster more ethical and widely accepted AI technologies, ensuring they meet both functional and ethical criteria. (shrink)

Simion and Kelp explain the trustworthiness of an AI based on that AI’s disposition to meet its obligations. Roughly speaking, according to Simion and Kelp, an AI is trustworthy regarding its task if and only if that AI is obliged to complete the task and its disposition to complete the task is strong enough. Furthermore, an AI is obliged to complete a task in the case where the task is the AI’s etiological function or design function. This account has (...) a strength in that it can provide a unificatory rationale for the trustworthy-making properties of AIs. According to this account, being explainable, being safe, and being transparent are the trustworthy-making properties of an AI because an AI can fulfill its etiological or design functions in the case where it is explainable, safe, and transparent. This paper shows that though Simion and Kelp’s account has a strength, this account is not satisfactory for two reasons. The first reason is that an AI’s trustworthiness is not determined just by the AI’s disposition to meet obligations, and the second reason is that it is difficult to explain how an AI’s etiological function and design function have to do with that AI’s obligations. To provide a full-fledged account, Simion and Kelp should dismiss these concerns. (shrink)

Cyberspace relies on information technologies to mediate relations between different people, across different communication networks and is reliant on the supporting technology. These interactions typically occur without physical proximity and those working depending on cybersystems must be able to trust the overall human–technical systems that support cyberspace. As such, detailed discussion of cybersecurity policy would be improved by including trust as a key value to help guide policy discussions. Moreover, effective cybersystems must have resilience designed into them. This paper argues (...) that trustworthy cybersystems are a key element to resilient systems, and thus are core to cybersecurity policy. The paper highlights the importance of trustworthiness for resilient cybersystems. The importance of trustworthiness is shown through a discussion of three events where trustworthiness was the target or casualty of cyberattacks: Stuxnet, hacking of communications and the Edward Snowden revelations. The impact of losing trust is highlighted, to underpin the argument that a resilient cybersystem ought to design in trustworthiness. The paper closes off by presenting a general set of policy implications arising from recognition of the interplay between trust, trustworthiness and resilience for effective cybersecurity. (shrink)

Trustworthiness is typically regarded as a desirable feature of national identification systems (NISs); but the variegated nature of the trustor communities associated with such systems makes it difficult to see how a single system could be equally trustworthy to all actual and potential trustors. This worry is accentuated by common theoretical accounts of trustworthiness. According to such accounts, trustworthiness is relativized to particular individuals and particular areas of activity, such that one can be trustworthy with regard to some (...) individuals in respect of certain matters, but not trustworthy with regard to all trustors in respect of every matter. The present article challenges this relativistic approach to trustworthiness by outlining a new account of trustworthiness, dubbed the expectation-oriented account. This account allows for the possibility of an absolutist (or one-place) approach to trustworthiness. Such an account, we suggest, is the approach that best supports the effort to develop NISs. To be trustworthy, we suggest, is to minimize the error associated with trustor expectations in situations of social dependency (commonly referred to as trust situations), and to be trustworthy in an absolute sense is to assign equal value to all expectation-related errors in all trust situations. In addition to outlining the features of the expectation-oriented account, we describe some of the implications of this account for the design, development, and management of trustworthy NISs. (shrink)

Where does entitlement to trust come from? When we trust someone to φ, do we need to have reason to trust them to φ or do we start out entitled to trust them to φ by default? Reductivists think that entitlement to trust always “reduces to” or is explained by the reasons that agents have to trust others. In contrast, anti-reductivists think that, in a broad range of circumstances, we just have entitlement to trust. even if we don’t have positive (...) reasons to do so. In this paper, we argue for a version of anti-reductivism. Roughly, we argue that we have default entitlement to trust someone to φ so long as there is an operative norm that requires S to φ. At least in such circumstances (and absent defeaters), we don’t need any positive reasons to trust S to φ. (shrink)

We argue that, to be trustworthy, Computa- tional Intelligence (CI) has to do what it is entrusted to do for permissible reasons and to be able to give rationalizing explanations of its behavior which are accurate and gras- pable. We support this claim by drawing par- allels with trustworthy human persons, and we show what difference this makes in a hypo- thetical CI hiring system. Finally, we point out two challenges for trustworthy CI and sketch a mechanism (...) which could be used to gener- ate sufficiently accurate as well as graspable rationalizing explanations for CI behavior. (shrink)

The aim of this paper is to discuss the “Framework for M&S with Agents” (FMSA) proposed by Zeigler et al. [2000, 2009] in regard to the diverse epistemological aims of agent simulations in social sciences. We first show that there surely are great similitudes, hence that the aim to emulate a universal “automated modeler agent” opens new ways of interactions between these two domains of M&S with agents. E.g., it can be shown that the multi-level conception at the core (...) of the FMSA is similar in both contexts: notions of “levels of system specifi cation”, “behavior of models”, “simulator”and “endomorphic agents” can be partially translated in the terms linked to the “denotational hierarchy” (DH) recently introduced in a multi-level centered epistemology of M&S. Second, we suggest considering the question of “credibility” of agent M&S in social sciences when we do not try to emulate but only to simulate target systems. Whereas a stringent and standardized treatment of the heterogeneous internal relations (in the DH) between systems of formalisms is the key problem and the essential challenge in the scope of Agent M&S driven engineering, it is urgent too to address the problem of the external relations (and of the external validity, hence of the epistemic power and credibility) of such levels of formalisms in the specific domains of agent M&S in social sciences, especially when we intend to introduce the concepts of activity tracking. (shrink)

The use of AI algorithms in criminal trials has been the subject of very lively ethical and legal debates recently. While there are concerns over the lack of accuracy and the harmful biases that certain algorithms display, new algorithms seem more promising and might lead to more accurate legal decisions. Algorithms seem especially relevant for bail decisions, because such decisions involve statistical data to which human reasoners struggle to give adequate weight. While getting the right legal outcome is a strong (...) desideratum of criminal trials, advocates of the relational theory of procedural justice give us good reason to think that fairness and perceived fairness of legal procedures have a value that is independent from the outcome. According to this literature, one key aspect of fairness is trustworthiness. In this paper, I argue that using certain algorithms to assist bail decisions could increase three diferent aspects of judges’ trustworthiness: (1) actual trustworthiness, (2) rich trustworthiness, and (3) perceived trustworthiness. (shrink)

I reply to Piazza's objection to my reductio against phenomenal conservatism.

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)

Several attempts have been made to apply the choice-sensitive theory of distributive justice, luck egalitarianism, in the context of health and healthcare. This article presents a framework for this discussion by highlighting different normative decisions to be made in such an application, some of the objections to which luck egalitarians must provide answers and some of the practical implications associated with applying such an approach in the real world. It is argued that luck egalitarians should address distributions of health (...) rather than healthcare, endorse an integrationist theory that combines health concerns with general distributive concerns and be pluralist in their approach. It further suggests that choice-sensitive policies need not be the result of applying luck egalitarianism in this context. (shrink)

In his recent article "Carnapian Frameworks" (Synthese, 2021), Gabriel Broughton criticizes my discussions of Carnap on ontology and puts forward his own interpretation of what Carnap’s external/internal distinction amounts to. I here first argue that Broughton’s main claims about me are based on a misinterpretation. Then I turn to some issues of broader interest. I argue that Broughton’s own, potentially interesting interpretation of Carnap’s external/internal distinction does not work. And in light of Broughton’s discussion I present a sharpened version of (...) what I have earlier said about Carnap’s this distinction. (shrink)

Philosophers of science are increasingly interested in engaging with scientific communities, policy makers, and members of the public; however, the nature of this engagement has not been systematically examined. Instead of delineating a specific kind of engaged philosophy of science, as previous accounts have done, this article draws on literature from outside the discipline to develop a framework for analyzing different forms of broadly engaged philosophy of science according to two key dimensions: social interaction and epistemic integration. Clarifying the (...) many forms of engagement available to philosophers of science can advance future scholarship on engagement and promote more strategic engagement efforts. (shrink)

In this paper, I examine the ethics of e - trust and e - trustworthiness in the context of health care, looking at direct computer-patient interfaces (DCPIs), information systems that provide medical information, diagnosis, advice, consenting and/or treatment directly to patients without clinicians as intermediaries. Designers, manufacturers and deployers of such systems have an ethical obligation to provide evidence of their trustworthiness to users. My argument for this claim is based on evidentialism about trust and trustworthiness: the idea that trust (...) should be based on sound evidence of trustworthiness. Evidence of trustworthiness is a broader notion than one might suppose, including not just information about the risks and performance of the system, but also interactional and context-based information. I suggest some sources of evidence in this broader sense that make it plausible that designers, manufacturers and deployers of DCPIs can provide evidence to users that is cognitively simple, easy to communicate, yet rationally connected with actual trustworthiness. (shrink)

A theme running through Katherine Hawley’s recent works on trust and trustworthiness is that thinking about the relations between these and Miranda Fricker’s notion of testimonial injustice offers a perspective from which we can see several limitations of Fricker’s own account of testimonial injustice. This paper clarifies the aspects of Fricker’s account that Hawley’s criticisms target, focusing on her objections to Fricker’s proposal that its primary harm involves a kind of epistemic objectification and her characterization of testimonial injustice in terms (...) of prejudicial credibility deficits. I also offer an assessment of the potency of Hawley’s objections, concluding that they do point to genuine limitations of Fricker’s account, but that we can appreciate Hawley’s points without adopting her trust-centric approach to testimonial injustice, or her specific commitment-based accounts of trust and trustworthiness. However, in the last section I examine the positive picture of testimonial injustice that emerges from Hawley’s discussion of unequally distributed obstacles to both being trustworthy and maintaining a reputation for trustworthy testimony in her recent book How To Be Trustworthy, considering how this picture contrasts with Fricker’s account and where it might have important affinities with other accounts in the literature. (shrink)

Since the global financial crisis in 2008, corporations have faced a crisis of trust, with growing sentiment against ‘elites and ‘big business’ and a feeling that ‘something ought to be done’ to re-establish public regard for corporations. Trust and trustworthiness are deeply moral significant. They provide the ‘glue or lubricant’ that begets reciprocity, decreases risk, secures dignity and respect, and safeguards against the subordination of the powerless to the powerful. However, in deciding how to restore trust, it is difficult to (...) determine precisely what should be done, by whom, and who will bear the cost, especially if any action involves a risk to overall market efficiency and corporate profitability. The paper explores whether corporations have a moral duty to be trustworthy, to bear the cost of being so and thus contribute to resolving the current crisis of trust. It also considers where the state and other social actors have strong reason to protect and enforce such moral rights, while acknowledging that other actors have similar obligations to be trustworthy. It outlines five ‘salient factors’ that trigger specific rights to trustworthiness and a concomitant duty on corporations to be trustworthy: market power, subordination (threat and intimidation), the absence of choice, the need to preserve systemic trust, and corporate political power which might undermine a state’s legitimacy. Absent these factors and corporations do not have a general duty to be trustworthy, since a responsible actor in fair market conditions should be able to choose between the costs and benefits of dealing with generally trustworthy corporations. (shrink)

Abstract: There are epistemic manipulators in the world. These people are actively attempting to sacrifice epistemic goods for personal gain. In doing so, manipulators have led many competent epistemic agents into believing contrarian theories that go against well-established knowledge. In this paper, I explore one mechanism by which manipulators get epistemic agents to believe contrarian theories. I do so by looking at a prominent empirical model of trustworthiness. This model identifies three major factors that epistemic agents look for when trying (...) to determine who is trustworthy. These are (i) ability, (ii) benevolence, and (iii) moral integrity. I then show how manipulators can manufacture the illusion that they possess these factors. This leads epistemic agents to view manipulators as trustworthy sources of information. Additionally, I argue that fact-checking will be an ineffective – or even harmful – practice when correcting the beliefs of epistemic agents who have been tricked by this illusion of epistemic trustworthiness. I suggest that in such cases we should use an alternative correction, which I call trust undercutting. (shrink)

Social epistemologists often operationalize the task of indirectly assessing experts’ trustworthiness to identifying whose beliefs are more reliably true on matters in an area of expertise. Not only does this neglect the philosophically rich space between belief formation and testimonial utterances, it also reduces trustworthiness to reliability. In ethics of trust, by contrast, explicitly relational views of trust include things like good will and responsiveness. One might think that relational aspects can be safely set aside for social epistemology of trust (...) in experts, that such considerations may be relevant for personal relationships but not for expert trustworthiness. Against these claims I argue for the social-epistemic relevance of relational aspects of trust in experts, and to that end I discuss three sorts of considerations – responsively positive, neutral, and negative factors – that can make a difference for expert trustworthiness. (shrink)

Islamic legal maxims (qawāʿʿid fiqhiyyah) provide necessary basis for extracting legal injunctions on the unprecedented cases (fiqh al-nawāzil) and make it possible for the jurists to forego the need of memorizing copious fiqh treatises. In light of this fact, this article attempts to design a framework for harm elimination, utilizing the related legal maxims, which will be arguably of great use in developing an outlook that enables a person to tackle the many challenges he or she finds in the (...) course of removal of harm. After explaining the concept and definition of Islamic legal maxims in general, this research surveys in detail the universal legal maxim “harm must be eliminated”, covering its role and significance in the theories of fiqh and its scope and application in contemporary issues. Employing descriptive, analytical and critical methods, this study categorizes the legal maxims related to the harm elimination into three: (1) maxims related to prevention of harm before its occurrence, (2) maxims related to elimination of harm after it has taken place and (3) maxims related to minimization of harm if complete removal is impractical. Likewise, this research analyses the sub-maxims of harm elimination, discussing their legal bases, various purposes for which they operate, related uṣūlī principles and legal examples, providing at the end a flowchart that represents a sequence of five steps useful in the course of removal of harm. (shrink)

Expert-informed public policy often depends on a degree of public trust in the relevant expert authorities. But if lay citizens are not themselves authorities on the relevant area of expertise, how can they make good judgements about the trustworthiness of those who claim such authority? I argue that the answer to this question depends on the kind of trust under consideration. Specifically, I maintain that a distinction between epistemic trust and recommendation trust has consequences for novices judging the trustworthiness of (...) experts. I argue for this by identifying the unique difficulties that emerge when a novice is asked not just to believe expert testimony, but to follow expert recommendations. I outline criteria for novice judgements of expert trustworthiness that have been proposed by Elizabeth Anderson and show that novel problems emerge for her criteria when we shift focus from epistemic trust to recommendation trust. More is needed when we are asked not just to believe the experts but to act as they recommend, because novices looking for trustworthy expert recommendations need to establish whether the recommended course of action supports what is important to them and accords with their values. (shrink)

Philosophical accounts of trustworthiness typically define trustworthiness as an agent being reliable in virtue of a specific motivation such as goodwill. The underlying thought motivating this view is that to be trustworthy is to be more than merely reliable. If motivational accounts are correct, this is a problem for non-motivational accounts of trustworthiness, as motivations are not required for trustworthiness. In this paper, I defend the non-motivational approach to trustworthiness and show that the motivational approach is inadequate. I do (...) this by making a novel distinction between trusting-to and trusting-as relations. A trusting-to relation is a relation in which a trustor ‘X’ trusts the trustee ‘Y’ to do something. Trusting-as relations are an overlooked relation implicit in all trusting-to relations. They describe the social relationship that holds between X and Y. I will argue that trusting-as relations determine whether any specific motivations are required for trustworthiness trusting-to relations. Thus, I show that acknowledging trusting-as relations enables us to provide a satisfactory explanation of the motivation intuition without making specific motivations constitutive features of trust. -/- . (shrink)

Philosophers have appealed to a wide variety of different factors in providing a metaphysics of race: appearance, ancestry, systems of oppression, shared ways of life, and so-called “racial essences”. I distinguish four importantly different questions about racial groups that one may be answering in appealing these factors. I then show that marking these distinctions proves quite fruitful, revealing ways of strengthening existing arguments for the non-existence of racial groups, new avenues for addressing challenges to biological and constructionist accounts of race, (...) and a range of hybrid positions that have been largely overlooked in the literature. (shrink)

Following the publication of numerous ethical principles and guidelines, the concept of 'Trustworthy AI' has become widely used. However, several AI ethicists argue against using this concept, often backing their arguments with decades of conceptual analyses made by scholars who studied the concept of trust. In this paper, I describe the historical-philosophical roots of their objection and the premise that trust entails a human quality that technologies lack. Then, I review existing criticisms about 'Trustworthy AI' and the consequence (...) of ignoring these criticisms: if the concept of 'Trustworthy AI' is kept being used, we risk attributing responsibilities to agents who cannot be held responsible, and consequently, deteriorate social structures which regard accountability and liability. Nevertheless, despite suggestions to shift the paradigm from 'Trustworthy AI' to 'Reliable AI', I argue that, realistically, this concept will be kept being used. I end by arguing that, ultimately, AI ethics is also about power, social justice, and scholarly activism. Therefore, I propose that community-driven and social justice-oriented ethicists of AI and trust scholars further focus on (a) democratic aspects of trust formation; and (b) draw attention to critical social aspects highlighted by phenomena of distrust. This way, it will be possible to further reveal shifts in power relations, challenge unfair status quos, and suggest meaningful ways to keep the interests of citizens. (shrink)

Many policies and ethical guidelines recommend developing “trustworthy AI”. We argue that developing morally trustworthy AI is not only unethical, as it promotes trust in an entity that cannot be trustworthy, but it is also unnecessary for optimal calibration. Instead, we show that reliability, exclusive of moral trust, entails the appropriate normative constraints that enable optimal calibration and mitigate the vulnerability that arises in high-stakes hybrid decision-making environments, without also demanding, as moral trust would, the anthropomorphization of (...) AI and thus epistemically dubious behavior. The normative demands of reliability for inter-agential action are argued to be met by an analogue to procedural metacognitive competence (i.e., the ability to evaluate the quality of one’s own informational states to regulate subsequent action). Drawing on recent empirical findings that suggest providing reliability scores (e.g., F1-scores) to human decision-makers improves calibration in the AI system, we argue that reliability scores provide a good index of competence and enable humans to determine how much they wish to rely on the system. (shrink)

In this paper we raise the question whether technological artifacts can properly speaking be trusted or said to be trustworthy. First, we set out some prevalent accounts of trust and trustworthiness and explain how they compare with the engineer’s notion of reliability. We distinguish between pure rational-choice accounts of trust, which do not differ in principle from mere judgments of reliability, and what we call “motivation-attributing” accounts of trust, which attribute specific motivations to trustworthy entities. Then we consider (...) some examples of technological entities that are, at first glance, best suited to serve as the objects of trust: intelligent systems that interact with users, and complex socio-technical systems. We conclude that the motivation-attributing concept of trustworthiness cannot be straightforwardly applied to these entities. Any applicable notion of trustworthy technology would have to depart significantly from the full-blown notion of trustworthiness associated with interpersonal trust. (shrink)

The main thesis of this paper is that when we trust the results of scientific research, that trust is inevitably directed at least in part at collective bodies rather than at single researchers, and that accordingly, reasonable assessments of epistemic trustworthiness in science must attend to these collective bodies. In order to support this claim, I start by invoking the collaborative nature of most of today’s scientific research. I argue that the trustworthiness of a collaborative research group does not supervene (...) on the trustworthiness of its individual members and point out some specific problems for the assessment of epistemic trustworthiness that arise from the specific nature of today’s collaborative research. Next, I argue that the social diffusion of trustworthiness goes even further; we always also need an assessment of the trustworthiness of the respective research community as a whole. Communities, I claim, play an essential role in the epistemic quality management of science. To see why this role is indispensible, we have to appreciate the full complexity of determining what is desirable in a method of inquiry. The relevant features of a method include three different dimensions: the reliability of positive results, the reliability of negative results, and the method’s power. Every methodological choice involves a trade-off between these three dimensions. The right balance between them (the “distribution of inductive risks”, or DIR) depends on value judgments about the costs of false results and the benefits of correct ones. Conventional methodological standards of research communities impose constraints on admissible DIR and thereby harmonize the implicit value judgments. Trusting that the research community has set the limitations on DIR in a suitable way is thus always part of placing our trust in a scientific result. (shrink)

When people must either save a greater number of people from a smaller harm or a smaller number from a greater harm, do their choices reflect a reasonable moral outlook? We pursue this question with the help of an experiment. In our experiment, two-fifths of subjects employ a similarity heuristic. When alternatives appear dissimilar in terms of the number saved but similar in terms of the magnitude of harm prevented, this heuristic mandates saving the greater number. In our experiment, this (...) leads to choices that are inconsistent with all standard theories of justice. We argue that this demonstrates the untrustworthiness of distributive judgments in cases that elicit similarity-based choice. (shrink)

This paper analyses the phenomenology and epistemology of chatbots such as ChatGPT and Bard. The computational architecture underpinning these chatbots are large language models (LLMs), which are generative artificial intelligence (AI) systems trained on a massive dataset of text extracted from the Web. We conceptualise these LLMs as multifunctional computational cognitive artifacts, used for various cognitive tasks such as translating, summarizing, answering questions, information-seeking, and much more. Phenomenologically, LLMs can be experienced as a “quasi-other”; when that happens, users anthropomorphise them. (...) For most users, current LLMs are black boxes, i.e., for the most part, they lack data transparency and algorithmic transparency. They can, however, be phenomenologically and informationally transparent, in which case there is an interactional flow. Anthropomorphising and interactional flow can, in some users, create an attitude of (unwarranted) trust towards the output LLMs generate. We conclude this paper by drawing on the epistemology of trust and testimony to examine the epistemic implications of these dimensions. Whilst LLMs generally generate accurate responses, we observe two epistemic pitfalls. Ideally, users should be able to match the level of trust that they place in LLMs to the degree that LLMs are trustworthy. However, both their data and algorithmic opacity and their phenomenological and informational transparency can make it difficult for users to calibrate their trust correctly. The effects of these limitations are twofold: users may adopt unwarranted attitudes of trust towards the outputs of LLMs (which is particularly problematic when LLMs hallucinate), and the trustworthiness of LLMs may be undermined. (shrink)