Where does the mind stop and the rest of the world begin? The question invites two standard replies. Some accept the demarcations of skin and skull, and say that what is outside the body is outside the mind. Others are impressed by arguments suggesting that the meaning of our words "just ain't in the head", and hold that this externalism about meaning carries over into an externalism about mind. We propose to pursue a third position. We advocate a very different (...) sort of externalism: an _active externalism_, based on the active role of the environment in driving cognitive processes. (shrink)

‘Plant neurobiology’ has emerged in recent years as a multidisciplinary endeavor carried out mainly by steady collaboration within the plant sciences. The field proposes a particular approach to the study of plant intelligence by putting forward an integrated view of plant signaling and adaptive behavior. Its objective is to account for the way plants perceive and act in a purposeful manner. But it is not only the plant sciences that constitute plant neurobiology. Resources from philosophy and cognitive science are central (...) to such an interdisciplinary project, if plant neurobiology is to maintain its target well-focused. This manifesto outlines a road map for the establishment and development of a new subject—the Philosophy of Plant Neurobiology—, a new field of research emerging at the intersection of the philosophy of cognitive science and plant neurobiology. The discipline is herewith presented, introducing challenges and novel lines of engagement with the empirical investigation, and providing an explanatory framework and guiding principles that will hopefully ease the integration of research on the quest for plant intelligence. (shrink)

‘Plant neurobiology’ has emerged in recent years as a multidisciplinary endeavor carried out mainly by steady collaboration within the plant sciences. The field proposes a particular approach to the study of plant intelligence by putting forward an integrated view of plant signaling and adaptive behavior. Its objective is to account for the way plants perceive and act in a purposeful manner. But it is not only the plant sciences that constitute plant neurobiology. Resources from philosophy and cognitive science are central (...) to such an interdisciplinary project, if plant neurobiology is to maintain its target well-focused. This manifesto outlines a road map for the establishment and development of a new subject—the Philosophy of Plant Neurobiology—, a new field of research emerging at the intersection of the philosophy of cognitive science and plant neurobiology. The discipline is herewith presented, introducing challenges and novel lines of engagement with the empirical investigation, and providing an explanatory framework and guiding principles that will hopefully ease the integration of research on the quest for plant intelligence. (shrink)

This article considers the issue of opacity as a problem for socially consequential mechanisms of classification and ranking, such as spam filters, credit card fraud detection, search engines, news trends, market segmentation and advertising, insurance or loan qualification, and credit scoring. These mechanisms of classification all frequently rely on computational algorithms, and in many cases on machine learning algorithms to do this work. In this article, I draw a distinction between three forms of opacity: opacity as intentional corporate or state (...) secrecy, opacity as technical illiteracy, and an opacity that arises from the characteristics of machine learning algorithms and the scale required to apply them usefully. The analysis in this article gets inside the algorithms themselves. I cite existing literatures in computer science, known industry practices, and do some testing and manipulation of code as a form of lightweight code audit. I argue that recognizing the distinct forms of opacity that may be coming into play in a given application is a key to determining which of a variety of technical and non-technical solutions could help to prevent harm. (shrink)

Deep neural networks have become increasingly successful in applications from biology to cosmology to social science. Trained DNNs, moreover, correspond to models that ideally allow the prediction of new phenomena. Building in part on the literature on ‘eXplainable AI’, I here argue that these models are instrumental in a sense that makes them non-explanatory, and that their automated generation is opaque in a unique way. This combination implies the possibility of an unprecedented gap between discovery and explanation: When unsupervised models (...) are successfully used in exploratory contexts, scientists face a whole new challenge in forming the concepts required for understanding underlying mechanisms. (shrink)

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

Western philosophers have traditionally concentrated on theory as the means for expressing knowledge about a variety of phenomena. This absorbing book challenges this fundamental notion by showing how objects themselves, specifically scientific instruments, can express knowledge. As he considers numerous intriguing examples, Davis Baird gives us the tools to "read" the material products of science and technology and to understand their place in culture. Making a provocative and original challenge to our conception of knowledge itself, _Thing Knowledge _demands that we (...) take a new look at theories of science and technology, knowledge, progress, and change. Baird considers a wide range of instruments, including Faraday's first electric motor, eighteenth-century mechanical models of the solar system, the cyclotron, various instruments developed by analytical chemists between 1930 and 1960, spectrometers, and more. (shrink)

Bioethics, Volume 36, Issue 2, Page 121-133, February 2022.

Computer simulations have conventionally been understood to be either extensions of formal methods such as mathematical models or as special cases of empirical practices such as experiments. Here, I argue that computer simulations are best understood as instruments. Understanding them as such can better elucidate their actual role as well as their potential epistemic standing in relation to science and other scientific methods, practices and devices.

One of the most fruitful interdisciplinary boundaries in contemporary scholarship is that between philosophy and cognitive science. Now that solid empirical results about the activities of the human mind are available, it is no longer necessary for philosophers to practice armchair psychology. In this short, accessible, and entertaining book, Alvin Goldman presents a masterly survey of recent work in cognitive science that has particular relevance to philosophy. Besides providing a valuable review of the most suggestive work in cognitive and social (...) psychology, Goldman demonstrates conclusively that the best work in philosophy in a surprising number of different fields―including philosophy of science, epistemology, metaphysics, and ethics as well as philosophy of mind―must take into account empirical breakthroughs in psychology. One of those rare texts that will also be useful for professionals, Philosophical Applications of Cognitive Science is appropriate for students in a wide range of philosophy courses. It will also interest researchers and students in psychology who are intrigued by the wider theoretical implications of their work. (shrink)

One of the main difficulties in assessing artificial intelligence (AI) is the tendency for people to anthropomorphise it. This becomes particularly problematic when we attach human moral activities to AI. For example, the European Commission’s High-level Expert Group on AI (HLEG) have adopted the position that we should establish a relationship of trust with AI and should cultivate trustworthy AI (HLEG AI Ethics guidelines for trustworthy AI, 2019, p. 35). Trust is one of the most important and defining activities in (...) human relationships, so proposing that AI should be trusted, is a very serious claim. This paper will show that AI cannot be something that has the capacity to be trusted according to the most prevalent definitions of trust because it does not possess emotive states or can be held responsible for their actions—requirements of the affective and normative accounts of trust. While AI meets all of the requirements of the rational account of trust, it will be shown that this is not actually a type of trust at all, but is instead, a form of reliance. Ultimately, even complex machines such as AI should not be viewed as trustworthy as this undermines the value of interpersonal trust, anthropomorphises AI, and diverts responsibility from those developing and using them. (shrink)

Epistemic trust is crucial for science. This article aims to identify the kinds of assumptions that are involved in epistemic trust as it is required for the successful operation of science as a collective epistemic enterprise. The relevant kind of reliance should involve working from the assumption that the epistemic endeavors of others are appropriately geared towards the truth, but the exact content of this assumption is more difficult to analyze than it might appear. The root of the problem is (...) that methodological decisions in science typically involve a complex trade-off between the reliability of positive results, the reliability of negative results, and the investigation's power. Which balance between these is the ‘correct’ one can only be determined in light of an evaluation of the consequences of all the different possible outcomes of the inquiry. What it means for the investigation to be ‘appropriately geared towards the truth’ thus depends on certain value judgments. I conclude that in the optimal case, trusting someone in her capacity as an information provider also involves a reliance on her having the right attitude towards the possible consequences of her epistemic work. 1 Introduction2 Epistemic Reliance within the Sciences3 Methodological Conventionalism4 Trust in Science5 Conclusions. (shrink)

Because of its complexity, contemporary scientific research is almost always tackled by groups of scientists, each of which works in a different part of a given research domain. We believe that understanding scientific progress thus requires understanding this division of cognitive labor. To this end, we present a novel agent-based model of scientific research in which scientists divide their labor to explore an unknown epistemic landscape. Scientists aim to climb uphill in this landscape, where elevation represents the significance of the (...) results discovered by employing a research approach. We consider three different search strategies scientists can adopt for exploring the landscape. In the first, scientists work alone and do not let the discoveries of the community as a whole influence their actions. This is compared with two social research strategies, which we call the follower and maverick strategies. Followers are biased towards what others have already discovered, and we find that pure populations of these scientists do less well than scientists acting independently. However, pure populations of mavericks, who try to avoid research approaches that have already been taken, vastly outperform both of the other strategies. Finally, we show that in mixed populations, mavericks stimulate followers to greater levels of epistemic production, making polymorphic populations of mavericks and followers ideal in many research domains. (shrink)

Inspired by the literature on extended/scaffolded mind, a debate concerning the contribution of extra-bodily resources to our (extended) emotions is recently gaining traction. Within this debate, inspired by the literature on cognitive artifacts introduces the notion of “affective artifacts”, indicating those objects that exert persistent effects on our feelings, possibly altering our self. However, by focusing on feelings, this notion neglects other facets of emotional episodes. Following Scarnatino’s tripartition between feeling, appraisal, and motivational theories of emotion, I present three varieties (...) of affective artifacts: Feeling, Evaluative and Motivational Artifacts. (shrink)

What does it mean to trust the results of a computer simulation? This paper argues that trust in simulations should be grounded in empirical evidence, good engineering practice, and established theoretical principles. Without these constraints, computer simulation risks becoming little more than speculation. We argue against two prominent positions in the epistemology of computer simulation and defend a conservative view that emphasizes the difference between the norms governing scientific investigation and those governing ordinary epistemic practices.

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

Because idealizations frequently advance scientific understanding, many claim that falsehoods play an epistemic role. In this paper, we argue that these positions greatly overstate idealiza...

ABSTRACTSeveral philosophers have inquired into the metaphysical limits of conceptual engineering: ‘Can we engineer? And if so, to what extent?’. This paper is not concerned with answering these questions. It does concern itself, however, with the limits of conceptual engineering, albeit in a largely unexplored sense: it cares about the normative, rather than about the metaphysical limits thereof. I first defend an optimistic claim: I argue that the ameliorative project has, so far, been too modest; there is little value theoretic (...) reason to restrict the project to remedying deficient representational devices, rather than go on a more ambitious quest: conceptual improvement. That being said, I also identify a limitation to the optimistic claim: I show that the ‘should’ in ameliorative projects suffers from a ‘wrong-kind-of-reasons’ problem. Last but not least, I sketch a proposal of normative constraining meant to address both the above results. The proposal gives primacy to epistemic constraints: accordingly, a concept should be ameliorated only insofar as this does not translate into epistemic loss. (shrink)

In this paper I use philosophical accounts on the relationship between trust and knowledge in science to apprehend this relationship on the Web. I argue that trust and knowledge are fundamentally entangled in our epistemic practices. Yet despite this fundamental entanglement, we do not trust blindly. Instead we make use of knowledge to rationally place or withdraw trust. We use knowledge about the sources of epistemic content as well as general background knowledge to assess epistemic claims. Hence, although we may (...) have a default to trust, we remain and should remain epistemically vigilant; we look out and need to look out for signs of insincerity and dishonesty in our attempts to know. A fundamental requirement for such vigilance is transparency: in order to critically assess epistemic agents, content and processes, we need to be able to access and address them. On the Web, this request for transparency becomes particularly pressing if (a) trust is placed in unknown human epistemic agents and (b) if it is placed in non-human agents, such as algorithms. I give examples of the entanglement between knowledge and trust on the Web and draw conclusions about the forms of transparency needed in such systems to support epistemically vigilant behaviour, which empowers users to become responsible and accountable knowers. (shrink)

Artefacts do not always do what they are supposed to, due to a variety of reasons, including manufacturing problems, poor maintenance, and normal wear-and-tear. Since software is an artefact, it should be subject to malfunctioning in the same sense in which other artefacts can malfunction. Yet, whether software is on a par with other artefacts when it comes to malfunctioning crucially depends on the abstraction used in the analysis. We distinguish between “negative” and “positive” notions of malfunction. A negative malfunction, (...) or dysfunction, occurs when an artefact token either does not or cannot do what it is supposed to. A positive malfunction, or misfunction, occurs when an artefact token may do what is supposed to but, at least occasionally, it also yields some unintended and undesirable effects. We argue that software, understood as type, may misfunction in some limited sense, but cannot dysfunction. Accordingly, one should distinguish software from other technical artefacts, in view of their design that makes dysfunction impossible for the former, while possible for the latter. (shrink)

In this paper I would like to propose the notion of “affective artifact”, building on an analogy with theories of cognitive artifacts and referring to the development of a situated affective science. Affective artifacts are tentatively defined as objects that have the capacity to alter the affective condition of an agent, and that in some cases play an important role in defining that agent’s self. The notion of affective artifacts will be presented by means of examples supported by empirical findings, (...) by discussing a tentative definition and classification, and by considering several related but differing notions. Within the framework of situated affectivity, the notion of affective artifacts will represent a further step in the enterprise of understanding how the environment helps us scaffold our affective processes. I will conclude that affective artifacts play a key role in the philosophy of cognitive science, the philosophy of technology and in the debate about the self. (shrink)

In this paper I argue that the search for explainable models and interpretable decisions in AI must be reformulated in terms of the broader project of offering a pragmatic and naturalistic account of understanding in AI. Intuitively, the purpose of providing an explanation of a model or a decision is to make it understandable to its stakeholders. But without a previous grasp of what it means to say that an agent understands a model or a decision, the explanatory strategies will (...) lack a well-defined goal. Aside from providing a clearer objective for XAI, focusing on understanding also allows us to relax the factivity condition on explanation, which is impossible to fulfill in many machine learning models, and to focus instead on the pragmatic conditions that determine the best fit between a model and the methods and devices deployed to understand it. After an examination of the different types of understanding discussed in the philosophical and psychological literature, I conclude that interpretative or approximation models not only provide the best way to achieve the objectual understanding of a machine learning model, but are also a necessary condition to achieve post hoc interpretability. This conclusion is partly based on the shortcomings of the purely functionalist approach to post hoc interpretability that seems to be predominant in most recent literature. (shrink)

In this paper I argue that the search for explainable models and interpretable decisions in AI must be reformulated in terms of the broader project of offering a pragmatic and naturalistic account of understanding in AI. Intuitively, the purpose of providing an explanation of a model or a decision is to make it understandable to its stakeholders. But without a previous grasp of what it means to say that an agent understands a model or a decision, the explanatory strategies will (...) lack a well-defined goal. Aside from providing a clearer objective for XAI, focusing on understanding also allows us to relax the factivity condition on explanation, which is impossible to fulfill in many machine learning models, and to focus instead on the pragmatic conditions that determine the best fit between a model and the methods and devices deployed to understand it. After an examination of the different types of understanding discussed in the philosophical and psychological literature, I conclude that interpretative or approximation models not only provide the best way to achieve the objectual understanding of a machine learning model, but are also a necessary condition to achieve post hoc interpretability. This conclusion is partly based on the shortcomings of the purely functionalist approach to post hoc interpretability that seems to be predominant in most recent literature. (shrink)

In this paper I critically examine the notion of explanation used in Artificial Intelligence in general, and in the theory of belief revision in particular. I focus on two of the best known accounts in the literature: Pagnucco’s abductive expansion functions and Gärdenfors’ counterfactual analysis. I argue that both accounts are at odds with the way in which this notion has historically been understood in philosophy. They are also at odds with the explanatory strategies used in actual scientific practice. At (...) the end of the paper I outline the desiderata for an epistemologically motivated, scientifically informed belief revision model for explanation. (shrink)

According to the Value-Neutrality Thesis, technology is morally and politically neutral, neither good nor bad. A knife may be put to bad use to murder an innocent person or to good use to peel an apple for a starving person, but the knife itself is a mere instrument, not a proper subject for moral or political evaluation. While contemporary philosophers of technology widely reject the VNT, it remains unclear whether claims about values in technology are just a figure of speech (...) or nontrivial empirical claims with genuine factual content and real-world implications. This paper provides the missing argument. I argue that by virtue of their material properties, technological artifacts are part of the normative order rather than external to it. I illustrate how values can be empirically identified in technology. The reason why value-talk is not trivial or metaphorical is that due to the endurance and longevity of technological artifacts, values embedded in them have long-term implications that surpass their designers and builders. I further argue that taking sides in this debate has real-world implications in the form of moral constraints on the development of technology. (shrink)

This paper offers an analysis of the nature of epistemic trust. With increased philosophical attention to social epistemology in general and testimony in particular, the role for an epistemic or intellectual version of trust has loomed large in recent debates. But, too often, epistemologists talk about trust without really providing a sustained examination of the concept. After some introductory comments, I begin by addressing various components key to trust simpliciter. In particular, I examine what we might think of when we (...) consider what it means to place trust in someone. Once we have examined trust in general, we can modify this discussion to derive an epistemic version of trust—placing trust in someone for an epistemic reason. I argue that ET includes four components: belief, communication, reliance, and confidence. The first two sets are distinctively epistemic and the second set of conditions form the core of any kind of trust. Put together,.. (shrink)

Although decision‐making algorithms are not new to medicine, the availability of vast stores of medical data, gains in computing power, and breakthroughs in machine learning are accelerating the pace of their development, expanding the range of questions they can address, and increasing their predictive power. In many cases, however, the most powerful machine learning techniques purchase diagnostic or predictive accuracy at the expense of our ability to access “the knowledge within the machine.” Without an explanation in terms of reasons or (...) a rationale for particular decisions in individual cases, some commentators regard ceding medical decision‐making to black box systems as contravening the profound moral responsibilities of clinicians. I argue, however, that opaque decisions are more common in medicine than critics realize. Moreover, as Aristotle noted over two millennia ago, when our knowledge of causal systems is incomplete and precarious—as it often is in medicine—the ability to explain how results are produced can be less important than the ability to produce such results and empirically verify their accuracy. (shrink)

Reasons are given to justify the claim that computer simulations and computational science constitute a distinctively new set of scientific methods and that these methods introduce new issues in the philosophy of science. These issues are both epistemological and methodological in kind.

A comparison is made between some epistemological issues arising in computer networks and standard features of social epistemology. A definition of knowledge for computational devices is provided and the topics of nonconceptual content and testimony are discussed.

Computational methods such as computer simulations, Monte Carlo methods, and agent-based modeling have become the dominant techniques in many areas of science. Extending Ourselves contains the first systematic philosophical account of these new methods, and how they require a different approach to scientific method. Paul Humphreys draws a parallel between the ways in which such computational methods have enhanced our abilities to mathematically model the world, and the more familiar ways in which scientific instruments have expanded our access to the (...) empirical world. This expansion forms the basis for a new kind of empiricism, better suited to the needs of science than the older anthropocentric forms of empiricism. Human abilities are no longer the ultimate standard of epistemological correctness.Humphreys also includes arguments for the primacy of properties rather than objects, the need to consider technological constraints when appraising scientific methods, and a detailed account of how the path from computational template to scientific application is constructed. This last feature allows us to hold a form of selective realism in which anti-realist arguments based on formal reconstructions of theories can be avoided. One important consequence of the rise of computational methods is that the traditional organization of the sciences is being replaced by an organization founded on computational templates.Extending Ourselves will be of interest to philosophers of science, epistemologists, and to anyone interested in the role played by computers in modern science. (shrink)

In his recent article ‘Limits of trust in medical AI,’ Hatherley argues that, if we believe that the motivations that are usually recognised as relevant for interpersonal trust have to be applied to interactions between humans and medical artificial intelligence, then these systems do not appear to be the appropriate objects of trust. In this response, we argue that it is possible to discuss trust in medical artificial intelligence, if one refrains from simply assuming that trust describes human–human interactions. To (...) do so, we consider an account of trust that distinguishes trust from reliance in a way that is compatible with trusting non-human agents. In this account, to trust a medical AI is to rely on it with little monitoring and control of the elements that make it trustworthy. This attitude does not imply specific properties in the AI system that in fact only humans can have. This account of trust is applicable, in particular, to all cases where a physician relies on the medical AI predictions to support his or her decision making. (shrink)

Real engines of the artificial intelligence revolution, machine learning models, and algorithms are embedded nowadays in many services and products around us. As a society, we argue it is now necessary to transition into a phronetic paradigm focused on the ethical dilemmas stemming from the conception and application of AIs to define actionable recommendations as well as normative solutions. However, both academic research and society-driven initiatives are still quite far from clearly defining a solid program of study and intervention. In (...) this contribution, we will focus on selected ethical investigations around AI by proposing an incremental model of trust that can be applied to both human-human and human-AI interactions. Starting with a quick overview of the existing accounts of trust, with special attention to Taddeo’s concept of “e-trust,” we will discuss all the components of the proposed model and the reasons to trust in human-AI interactions in an example of relevance for business organizations. We end this contribution with an analysis of the epistemic and pragmatic reasons of trust in human-AI interactions and with a discussion of kinds of normativity in trustworthiness of AIs. (shrink)

The use of black box algorithms in medicine has raised scholarly concerns due to their opaqueness and lack of trustworthiness. Concerns about potential bias, accountability and responsibility, patient autonomy and compromised trust transpire with black box algorithms. These worries connect epistemic concerns with normative issues. In this paper, we outline that black box algorithms are less problematic for epistemic reasons than many scholars seem to believe. By outlining that more transparency in algorithms is not always necessary, and by explaining that (...) computational processes are indeed methodologically opaque to humans, we argue that the reliability of algorithms provides reasons for trusting the outcomes of medical artificial intelligence. To this end, we explain how computational reliabilism, which does not require transparency and supports the reliability of algorithms, justifies the belief that results of medical AI are to be trusted. We also argue that several ethical concerns remain with black box algorithms, even when the results are trustworthy. Having justified knowledge from reliable indicators is, therefore, necessary but not sufficient for normatively justifying physicians to act. This means that deliberation about the results of reliable algorithms is required to find out what is a desirable action. Thus understood, we argue that such challenges should not dismiss the use of black box algorithms altogether but should inform the way in which these algorithms are designed and implemented. When physicians are trained to acquire the necessary skills and expertise, and collaborate with medical informatics and data scientists, black box algorithms can contribute to improving medical care. (shrink)

Several philosophical issues in connection with computer simulations rely on the assumption that results of simulations are trustworthy. Examples of these include the debate on the experimental role of computer simulations :483–496, 2009; Morrison in Philos Stud 143:33–57, 2009), the nature of computer data Computer simulations and the changing face of scientific experimentation, Cambridge Scholars Publishing, Barcelona, 2013; Humphreys, in: Durán, Arnold Computer simulations and the changing face of scientific experimentation, Cambridge Scholars Publishing, Barcelona, 2013), and the explanatory power of (...) computer simulations :277–292, 2008; Durán in Int Stud Philos Sci 31:27–45, 2017). The aim of this article is to show that these authors are right in assuming that results of computer simulations are to be trusted when computer simulations are reliable processes. After a short reconstruction of the problem of epistemic opacity, the article elaborates extensively on computational reliabilism, a specified form of process reliabilism with computer simulations located at the center. The article ends with a discussion of four sources for computational reliabilism, namely, verification and validation, robustness analysis for computer simulations, a history of successful implementations, and the role of expert knowledge in simulations. (shrink)

The debate between externalists and internalists in epistemology can be viewed as a disagreement about whether there are epistemic rights without corresponding duties or obligations. Taking an epistemic right to believe P as an authorization to not only accept P as true but to use P as a positive reason for accepting other propositions, the debate is about whether there are unjustified justifiers. It is about whether there are propositions that provide for others what nothing need provide for them—viz., reasons (...) for thinking them true. (shrink)

In the past few years, machine learning (ML) tools have been implemented with success in the medical context. However, several practitioners have raised concerns about the lack of transparency—at the algorithmic level—of many of these tools; and solutions from the field of explainable AI (XAI) have been seen as a way to open the ‘black box’ and make the tools more trustworthy. Recently, Alex London has argued that in the medical context we do not need machine learning tools to be (...) interpretable at the algorithmic level to make them trustworthy, as long as they meet some strict empirical desiderata. In this paper, we analyse and develop London’s position. In particular, we make two claims. First, we claim that London’s solution to the problem of trust can potentially address another problem, which is how to evaluate the reliability of ML tools in medicine for regulatory purposes. Second, we claim that to deal with this problem, we need to develop London’s views by shifting the focus from the opacity of algorithmic details to the opacity of the way in which ML tools are trained and built. We claim that to regulate AI tools and evaluate their reliability, agencies need an explanation of how ML tools have been built, which requires documenting and justifying the technical choices that practitioners have made in designing such tools. This is because different algorithmic designs may lead to different outcomes, and to the realization of different purposes. However, given that technical choices underlying algorithmic design are shaped by value-laden considerations, opening the black box of the design process means also making transparent and motivating (technical and ethical) values and preferences behind such choices. Using tools from philosophy of technology and philosophy of science, we elaborate a framework showing how an explanation of the training processes of ML tools in medicine should look like. (shrink)

Information providing and gathering increasingly involve technologies like search ‎engines, which actively shape their epistemic surroundings. Yet, a satisfying account ‎of the epistemic responsibilities associated with them does not exist. We analyze ‎automatically generated search suggestions from the perspective of social ‎epistemology to illustrate how epistemic responsibilities associated with a ‎technology can be derived and assigned. Drawing on our previously developed ‎theoretical framework that connects responsible epistemic behavior to ‎practicability, we address two questions: first, given the different technological ‎possibilities available (...) to searchers, the search technology, and search providers, ‎who should bear which responsibilities? Second, given the technology’s ‎epistemically relevant features and potential harms, how should search terms be ‎autocompleted? Our analysis reveals that epistemic responsibility lies mostly with ‎search providers, which should eliminate three categories of autosuggestions: those ‎that result from organized attacks, those that perpetuate damaging stereotypes, and ‎those that associate negative characteristics with specific individuals.‎. (shrink)

Conventional accounts of epistemic opacity, particularly those that stem from the definitive work of Paul Humphreys, typically point to limitations on the part of epistemic agents to account for the distinct ways in which systems, such as computational methods and devices, are opaque. They point, for example, to the lack of technical skill on the part of an agent, the failure to meet standards of best practice, or even the nature of an agent as reasons why epistemically relevant elements of (...) a process may be inaccessible. In this paper I argue that there are certain instances of epistemic opacity— particularly in computational methods such a computer simulations and machine learning processes—that do not arise from, are not responsive to, and are therefore not explained by the epistemic limitations of an agent. I call these instances agent-neutral and agent-independent instances of epistemic opacity respectively. As a result, I also argue that conventional accounts of epistemic opacity offer a limited understanding of the full spectrum of kinds and sources of epistemic opacity, particularly of the kind found in computational methods. In particular, as I will show below, the limitations of these accounts are reflected in the way they fail to provide satisfactory explanations when faced with certain instances of opacity. (shrink)