I argue that any plausible decision theory for agents with incomplete preferences which obeys the Never Worse Principle will violate Transitivity. The Never Worse Principle says that if one option never does worse than another, you shouldn’t disprefer it. Transitivity says that if you prefer X to Y and you prefer Y to Z, then you should prefer X to Z. Violating Transitivity allows one to be money pumped. Although agents with incomplete preferences are already, in virtue of having incomplete (...) preferences, vulnerable to being money pumped, I argue that the money pump argument for Transitivity is more serious than the one for Completeness. (shrink)

This article explores the concept of metaphysically opaque grounding, a largely neglected form of metaphysical grounding that challenges the commonly held assumptions that grounding is an especially intimate and powerful connection between facts and that it is necessarily connected with the essences of things. I provide a definition of opaque grounding, identify some interesting philosophical views that are committed to it, and explore some consequences for the general theory of grounding. Finally, I briefly address some natural initial doubts (...) about opaque grounding and find them unwarranted. The upshot is that the notion deserves more attention than it has previously received. (shrink)

This article aims to contribute to the largely neglected issue of whether metaphysical grounding – the relation of one fact’s obtaining in virtue of the obtaining of some other (or others) – can be given a reductive account. I introduce the notion of metaphysically opaque grounding, a form of grounding which constitutes a less metaphysically intimate connection than in standard cases. I then argue that certain important and interesting views in metaphysics are committed to there being cases of (...) class='Hi'>opaque grounding and demonstrate that four representative accounts of grounding available in the literature are unable to accommodate such cases. This is argued to constitute a problem for those accounts that is likely to extend to other possible reductive accounts of grounding that employ the popular strategy of explaining grounding in terms of other hyperintensional phenomena. Unless the reductionist is willing to opt for some sophisticated modalist account, the possibility of opaque grounding cases thus provides indirect support for primitivism about grounding, a view that has previously been widely embraced but rarely supported by argument. (shrink)

This paper offers an analytic perspective on epistemic dependence that is grounded in theoretical discussion and field observation at the same time. When in the course of knowledge creation epistemic labor is divided, collaborating scientists come to depend upon one another epistemically. Since instances of epistemic dependence are multifarious in scientific practice, I propose to distinguish between two different forms of epistemic dependence, opaque and translucent epistemic dependence. A scientist is opaquely dependent upon a colleague if she does not (...) possess the expertise necessary to independently carry out, and to profoundly assess, the piece of scientific labor which her colleague is contributing. If the scientist does possess the necessary expertise, I argue, her dependence is translucent. However, the distinction between opaque and translucent epistemic dependence does not exhaust dependence relations in scientific practice, because many dependence relations are neither entirely opaque nor translucent. I will discuss why this is the case, and show how we can make sense of the gray zone between opaque and translucent epistemic dependence. (shrink)

If updating with E has the same result across all epistemically possible worlds, then the agent has no uncertainty as to the behavior of the update, and we may call it a transparent update. If an agent is uncertain about the behavior of an update, we may call it opaque. In order to model the uncertainty an agent has about the result of an update, the same update must behave differently across different possible worlds. In this paper, I study (...) opaque updates using a simple system of dynamic epistemic logic suitably modified for that purpose. The paper highlights the connection between opaque updates and the dynamic-epistemic principles Perfect-Recall and No-Miracles. I argue that opaque updates are central to contemporary discussions in epistemology, in particular to externalist theories of knowledge and to the related problem of epistemic bootstrapping, or easy knowledge. Opaque updates allow us to explicitly investigate a dynamic form of uncertainty, using simple and precise logical tools. (shrink)

In 2020, the European Union announced the award of the contract for the biometric part of the new database for border control, the Entry Exit System, to two companies: IDEMIA and Sopra Steria. Both companies had been previously involved in the development of databases for border and migration management. While there has been a growing amount of publicly available documents that show what kind of technologies are being implemented, for how much money, and by whom, there has been limited engagement (...) with digital methods in this field. Moreover, critical border and security scholarship has largely focused on qualitative and ethnographic methods. Building on a data feminist approach, we propose a transdisciplinary methodology that goes beyond binaries of qualitative/quantitative and opacity/transparency, examines power asymmetries and makes the labour of coding visible. Empirically, we build and analyse a dataset of the contracts awarded by two European Union agencies key to its border management policies – the European Agency for Large-Scale Information Systems (eu-LISA) and the European Border and Coast Guard Agency (Frontex). We supplement the digital analysis and visualisation of networks of companies with close reading of tender documents. In so doing, we show how a transdisciplinary methodology can be a device for making datafication ‘intelligible’ at the European Union borders. (shrink)

Summary Verbs of creation (create, make, paint) are not transparent. The object created does not exist during the event time but only thereafter. We may call this type of opacity temporal opacity. I is to be distinguished from modal opacity, which is found in verbs like owe or seek. (Dowty, 1979) offers two analyses of creation verbs. One analysis predicts that no object of the sort created exists before the time of the creation. The other analysis says that the object (...) exists throughout the act of creation. I investigate three theories: Theory I says that no object of the sort created and which is caused by the very act of creation exists before the creation. In this theory, verbs of creation must embed a property. Theory II can regard the indefinite object of a creation verb as a quantifier and gives it wide scope with respect to the verb. The theory has to make sure that the objects quantified over exist only after the event. While Theory I and II start from the assumption that the extension of all nouns depend on time, Theory III says that Individual Level predicates do not depend on time. This ontology will enable us to treat verbs of creation as first order relations. The theory will entail that a picture does not mean the same as there is a picture. The paper discusses various approaches to the problem: Krifka, Parsons, Landman, Kratzer and Zucchi. (shrink)

The paper is about the interpretation of opaque verbs like “seek”, “owe”, and “resemble” which allow for unspecific readings of their (indefinite) objects. It is shown that the following two observations create a problem for semantic analysis: (a) The opaque position is upward monotone: “John seeks a unicorn” implies “John seeks an animal”, given that “unicorn” is more specific than “animal”. (b) Indefinite objects of opaque verbs allow for higher-order, or “underspecific”, readings: “Jones is looking for something (...) Smith is looking for” can express that there is something unspecific that both Jones and Smith are looking for. Given (a) and (b), it would seem that the following inference is hard to escape, if the premisses are construed unspecifically and the conclusion is taken on its under- specific reading: Jones is looking for a sweater. Smith is looking for a pen. Smith is looking for something Jones is looking for. (shrink)

_Self-Reflection for the Opaque Mind_ attempts to solve a grave problem about critical self-reflection. Psychological studies indicate not just that we are bad at detecting our own "ego-threatening" thoughts; they also suggest that we are ignorant of even our ordinary thoughts. However, self-reflection presupposes an ability to know one’s own thoughts. So if ignorance is the norm, why attempt self-reflection? While admitting the psychological data, this book argues that we are infallible in a limited range of self-discerning judgments—that in (...) some cases, these judgments are self-fulfilling or self-verifying. Even so, infallibility does not imply indubitability, and the author does not wish to provide a "foundation" for empirical knowledge. The point is rather to explain how self-reflection as a rational activity is possible. The book will be of interest to scholars working on the issue of self-reflection across a wide range of disciplines, including philosophy, psychology, and cognitive science. (shrink)

ABSTRACTThis paper addresses the problem of opaque sweetening and argues that one should use stochastic dominance in comparing lotteries even when dealing with incomplete orderings that allow for non-comparable outcomes.

As narrative conceptions of selfhood have gained more acceptance within various disciplines including philosophy, psychology, and the cognitive sciences, so too have these conceptions been critically appraised. Chief among those who are suspicious of the overall viability of ‘narrative identity’ is the philosopher, Galen Strawson. In this paper, I develop five arguments underlying Strawson’s critique of narrative identity, and respond to each argument from the perspective of the hermeneutic phenomenology of Paul Ricœur. Though intuitive, I demonstrate that none of Strawson’s (...) arguments are cogent. The confrontation between these two figures highlights a deep conceptual disagreement about our epistemic access to the self, which has thus far gone unrecognized in the Anglo-American discussion, so that it raises a new problem for the metaphysics of personal identity. (shrink)

In a novel move against Russellianism, Heck (2014) has argued that reports of the form S believes that p are semantically opaque on the grounds that there are no other means in English to report psychologically individuated beliefs, such as those Lois Lane reports using the names ‘Superman’ and ‘Clark Kent.’ I show that there are several other ways to meet this need. I focus on quotational reports of the form S believes “p,” which philosophers have overlooked or mischaracterized. (...) I argue that reports of this form are no more metalinguistic than those using the complementizer that. In both, the meaning of the report depends on the meaning, not the form, of the complement. Some distinguishing features of non-metalinguistic quotational reports are that their complements force “deictic shift” and have no transparent interpretation. Free direct speech and thought bubbles have the same properties. The “q-ascriptions” McCullagh (2017a) proposed to defend Russellianism against Heck are not in common use and on his account are unsuitably metalinguistic. The fact that quotational belief reports are opaque without being metalinguistic makes the Russellian thesis that the meaning of a name is its referent untenable, thereby undermining Russellianism about belief reports. (shrink)

In this paper, we outline an approach to giving extensional truth-theoretic semantics for what have traditionally been seen as opaque sentential contexts. We outline an approach to providing a compositional truth-theoretic semantics for opaque contexts which does not require quantifying over intensional entities of any kind, and meets standard objections to such accounts. The account we present aims to meet the following desiderata on a semantic theory T for opaque contexts: (D1) T can be formulated in a (...) first-order extensional language; (D2) T does not require quantification over intensional entities­i.e., meanings, propositions, properties, relations, or the like­in its treatment of opaque contexts; (D3) T captures the entailment relations that hold in virtue of form between sentences in the language for which it is a theory; (D4) T has a finite number of axioms. If the approach outlined here is correct, it resolves a longstanding complex of problems in metaphysics, the philosophy of mind and the philosophy of language. (shrink)

Under what conditions does machine learning (ML) model opacity inhibit the possibility of explaining and understanding phenomena? In this article, I argue that nonepistemic values give shape to the ML opacity problem even if we keep researcher interests fixed. Treating ML models as an instance of doing model-based science to explain and understand phenomena reveals that there is (i) an external opacity problem, where the presence of inductive risk imposes higher standards on externally validating models, and (ii) an internal opacity (...) problem, where greater inductive risk demands a higher level of transparency regarding the inferences the model makes. (shrink)

The Visible Medium. Opaque Interface and Non-Mimetic Immersivity -/- The relation of reciprocal co-implication that Merleau-Ponty formulates—and on which he insists throughout his work—between sense and the sensible, perception and expression, and then visible and invisible, transforms the way in which one conceives of the medium. Merleau-Ponty’s aesthetics reveals an idea of the medium as a support that erases itself in the act of conveying the signifi cation and also shakes the direct correlation between transparency and mimetic simulation. Understood (...) as the sensible thickness of the body opening onto the world, then as depth and écart that catalyzes vision, the medium, then, furnishes one of the defi nitions of fl esh, as the element of auto-mediation: connective tissue or fabric of communication that is at once écart and internal difference. Merleau-Ponty conceives of the medium as both that which renders and that which is rendered visible. It is therefore no longer an intermediary; it ceases to be an invisible mediator and becomes the opaque element that reveals in fi ligree the movement of gestaltic difference. It is from such a Merleau-Pontian conception of medium that one can begin to elaborate the complex issues posed by mediality in the post-medial age. The idea of a “visible medium” permits us to break with the confl ation of the simulation’s immersive effect and performance, which often informs the rhetoric concerning medias and new technologies, in order rather to think of an “opaque interface” or an “non-mimetic immersivity.” From such a conception of mediality we can equally understand the phenomenon of numerical convergence, not as the accomplishment of the suppression or dematerialization of the medium, as is the case with traditional theories, but as the point of departure for a return to the body as the condition of possibility for every aesthetic experience. (shrink)

This paper is about the semantic analysis of opaque verbs such as seek and owe, which allow for unspecific readings of their indefinite objects.1 One may be looking for a good car without there being any car that one is looking for; or, one may be looking for a good car in that a specific car exists that one is looking for. It thus appears that there are two interpretations of these verbs – a specific and an unspecific one (...) – and one may wonder how they are related. The present paper is a contribution to this question. (shrink)

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

Opaqueness of pp collisions is evaluated at three CERN-ISR energies. Comparisons with predictions of the factorizable eikonal models and the scaling hypothesis are made. It appears that results are in favor of the factorizable models.

Exploring intentionality from an externalist perspective, I distinguish three kinds of intentionality in the case of seeing, which I call transparent, translucent, and opaque respectively. I then extend the distinction from seeing to knowing, and then to believing. Having explicated the three-fold distinction, I then critically explore some important consequences that follow from granting that (i) there are transparent and translucent intentional states and (ii) these intentional states are mental states. These consequences include: first, that existential opacity is neither (...) the mark of intentionality nor of the mental; second, that Sellars has not shown that all intentionality is non-relational; third, that a key Quinean argument for semantic indeterminacy rests on a false premise; fourth, that perceptual experience is intentional on Alston’s Theory of Appearing; fifth, that either some mental causation is more than internal physiological causation or some mental states are epiphenomenal. (shrink)

We are currently witnessing a sharp rise in the use of algorithmic decision-making tools. In these instances, a new wave of policy concerns is set forth. This article strives to map out these issues, separating the wheat from the chaff. It aims to provide policy makers and scholars with a comprehensive framework for approaching these thorny issues in their various capacities. To achieve this objective, this article focuses its attention on a general analytical framework, which will be applied to a (...) specific subset of the overall discussion. The analytical framework will reduce the discussion to two dimensions, every one of which addressing two central elements. These four factors call for a distinct discussion, which is at times absent in the existing literature. The two dimensions are the specific and novel problems the process assumedly generates and the specific attributes which exacerbate them. While the problems are articulated in a variety of ways, they most likely could be reduced to two broad categories: efficiency and fairness-based concerns. In the context of this discussion, such problems are usually linked to two salient attributes the algorithmic processes feature—its opaque and automated nature. (shrink)

This essay examines women's spectacularly visible status in feminised mass cultural domains in the first decades of the twentieth century. Feminine spectacles are commonly understood to invite viewers to access women's bodies, yet early twentieth-century spectacles paradoxically called renewed attention to women's illegibility. Women's visual prominence made apparent their ‘unknowability’, recasting an ancient ideational heritage in modern terms. Representations of women as opaque in the early twentieth century constituted a challenge to ocularcentrism and reveal the centrality of femininity in (...) mass mediations of epistemology and ontology. Drawing on written accounts of women's opacity in the fashion and beauty press, I argue that attention to spectacles of unknowability can be productive for feminist modernist studies. The texturing of histories of feminine spectacle challenges some tenacious dichotomies that continue to inform accounts of women's place in the modern, including those of subject and object, and visibility and invisibility. Focusing on opacity leads us to a productive account of the variable visibility of women in the modern, which foregrounds the multiple historical relations of different groups of women to regimes of visibility and keeps in view the diverse ways that differently classed and raced women were positioned vis-à-vis spectacle. The essay draws on feminist and postcolonial theory to suggest that an attunement to the unknowable not only nuances our understanding of a discrete historical period, but can lead the feminist researcher to confront and expand her own gaze in the era of capitalist modernity. (shrink)

This paper examines the paradoxical transparency involved in training machine-learning models. Existing literature typically critiques the opacity of machine-learning models such as neural networks or collaborative filtering, a type of critique that parallels the black-box critique in technology studies. Accordingly, people in power may leverage the models’ opacity to justify a biased result without subjecting the technical operations to public scrutiny, in what Dan McQuillan metaphorically depicts as an “algorithmic state of exception”. This paper attempts to differentiate the black-box abstraction (...) that wraps around complex computational systems from the opacity of machine-learning models. It contends that the degree of asymmetry in knowledge is greater in the former than the latter. In the case of software systems, software codes are difficult to understand as only software experts with sufficient domain knowledge are equipped to formulate a sound critique. In contrast, the meanings of trained parameters in a machine-learning model are obscure even to the data scientists who configure and train the model. Hence, the asymmetry of knowledge lies only in how data examples are collected, the choice and configuration of machine-learning models, and the specification of features in model design. Under the trend of algorithmic decision-making proliferating with machine-learning heuristics, the paper contends that the more symmetric distribution of knowledge in machine learning could lead to a more transparent production process if proper policies are in place. (shrink)

Charles Travis; V*—Are Belief Ascriptions Opaque?, Proceedings of the Aristotelian Society, Volume 85, Issue 1, 1 June 1985, Pages 73–100, https://doi.org/10.10.

In his recent article, ``Self-Consciousness'’, George Bealer has set outa novel and interesting argument against functionalism in the philosophyof mind. I shall attempt to show, however, that Bealer's argument cannotbe sustained.In arguing for this conclusion, I shall be defending three main theses.The first is connected with the problem of defining theoreticalpredicates that occur in theories where the following two features arepresent: first, the theoretical predicate in question occurswithin both extensional and non-extensional contexts; secondly, thetheory in question asserts that the relevant (...) theoretical states enterinto causal relations. What I shall argue is that a Ramsey-styleapproach to the definition of such theoretical terms requires twodistinct quantifiers: one which ranges over concepts, and theother which ranges over properties in the world.My second thesis is a corollary: since the theories on whichBealer is focusing have both of the features just mentioned, and sincethe method that he employs to define theoretical terms in his argumentagainst functionalism does not involve both quantifiers that range overproperties and quantifiers that range over concepts, that method isunsound.My final thesis is that when a sound method is used, Bealer's argumentagainst functionalism no longer goes through.The structure of my discussion is as follows. I begin by setting out twoarguments – the one, a condensed version of Bealer's argument, andthe other, an argument that parallels Bealer's argument very closely.The parallel argument leads to a conclusion, however, that, rather thanbeing merely somewhat surprising, seems very implausible indeed. Forwhat the second argument establishes, if sound, is that there can betheoretical terms that apply to objects by virtue of their first-orderphysical properties, but whose meaning cannot be defined via aRamsey-style approach.Having set out the two parallel arguments, I then go on to focus uponthe second, to determine what is wrong with it. My diagnosis will bethat the problem with the argument arises from the fact that it involvesdefining a theoretical term that occurs both inside and outside ofopaque contexts, for the method employed fails to take into account thefact that the types of entities that are involved in the relevanttruthmakers are different when a sentence occurs within an extensionalcontext from those involved when a sentence occurs within anon-extensional context.I then go on to discuss how one should define a theoretical term thatoccurs within such theories, and I argue that in such a case one needstwo quantifiers, ranging over different types of entities – on theone hand, over properties and relations, and the other, over concepts. Ithen show that, when such an approach is followed, the argument inquestion collapses.I then turn to Bealer's argument against functionalism, and I show,first, that precisely the same method of defining theoretical terms canbe applied there, and, secondly, that, when this is done, it turns outthat that argument is also unsound.Next, I consider two responses that Bealer might make to my argument,and I argue that those responses would not succeed.Finally, I conclude by asking exactly where the problem lies in the caseof Bealer's argument. My answer will be that it is not simply the factthat one is dealing with a theoretical term that occurs in bothextensional and non-extensional contexts. It is rather the combinationof that feature together with the fact that the theory in questionasserts that the relevant type of theoretical state enters into causalrelations. For the first of these features means that the Ramseysentence for the theory must involve quantification over concepts, whilethe presence of the second feature means that the Ramsey sentence mustinvolve quantification over properties in the world, and so no attemptto offer a Ramsey-style account of the meaning of the relevanttheoretical term can succeed unless one employs both quantification overconcepts and quantification over properties. Bealer, however, in hisargument against functionalism, uses a method of defining theoreticalterms that does not involve both types of quantification, and it isprecisely because of this that his argument does not in the end succeed. (shrink)

I will argue that there are (at least) two species of musical silence, which cannot be distinguished by attending to how these silences sound. I term these two kinds of musical silence ‘capacious’ and ‘opaque’. Both capacious and opaque musical silences might occur in the midst of the ongoing production of sound or might exist in the complete absence of sound. Both kinds of silence can, in certain conditions, be sonically identical, but both are always received by the (...) listening ear in importantly different ways. I will discuss our ability to distinguish between these silences, even when they ‘sound’ the same. I will argue that the accounts of musical silence and silent music offered by (among others) Levinson, Kania, and Davies are vitiated by their failure to distinguish between these kinds of musical silence. I distinguish these kinds of silence by the kinds of listening activity associated with them, and the ideal structure of such listening practices. (shrink)

This dissertation investigates grounding, the relation of non-causal determination whereby one fact obtains in virtue of some other fact or facts. Although considerations of grounding have been central throughout Western philosophy, the last 15-20 years have seen a renaissance of systematic work on grounding in analytic philosophy. The aim of the dissertation is to contribute to our understanding of the nature of grounding and its relation to other central phenomena in metaphysics. -/- Chapter 1 of the dissertation provides a brief (...) presentation of grounding and introduces some distinctions of relevance to later chapters. Chapter 2 discusses the question of whether grounding can be reduced. Four reductive accounts – in terms of fundamentality and supervenience, essence, reduction, and metaphysical laws respectively – are considered. All of these accounts are found to be subject to objections and are consequently rejected. -/- Chapter 3 considers the relation between grounding and another important phenomenon in metaphysics, namely ontological dependence. The question of the relation between the two is approached through an investigation of their modal connection (or lack thereof). It is argued that on the common assumption that grounding is a factive relation, ontological dependence can be shown to be neither necessary nor sufficient for grounding. Furthermore, to get any form of modal connection between the two relations, one needs to move to a non-factive conception of grounding or consider generic ontological dependence and accept a potentially controversial principle about the modal behaviour of grounding. Even then, the modal connection between grounding and ontological dependence is weak and subject to qualifications. -/- Chapter 4 develops and defends the novel notion of metaphysically opaque grounding. It is commonly assumed that grounding is an especially intimate and powerful connection and enables a form of explanation that is particularly strict and illuminating. An arguably related idea is that grounding is necessarily connected with the core features of things – their essences or natures. Metaphysically opaque grounding constitutes a form of grounding that falsifies both of these ideas. Chapter 4 makes the notion of metaphysically opaque grounding precise, motivates the idea that there are genuine cases of such grounding, explores its consequences for the theory of grounding, and defends the notion from potential objections, concluding that there are currently no strong reasons to rule out opaque grounding. (shrink)

Matheus Valente presents a number of examples designed to show that my theory of reference fixing for demonstratives violates the desideratum that demonstrative thoughts should be transparent to speakers. In this note I argue that the alleged desideratum is not really such and defend my theory against other criticisms made by Valente.

Lots of contextually sensitive expressions appear to have context invariant meanings that do not by themselves suffice to secure semantic values for those expressions in context. For example, suppose I say 1. She is smart. where I do not demonstrate any female, I don’t intend that some female is the semantic value of my use of ‘she’, no female is uniquely salient in the context of utterance, and no female has been under discussion. It would appear in such a case (...) that the context invariant meaning of ‘she’ does not secure a semantic value in context for my use of ‘she’, resulting in infelicity. After all, what would that semantic value in context be? This appears to show that the context invariant meaning of ‘she’ does not by itself secure semantic values in context for it. The class of expressions that are like ‘she’ in this respect is quite large. It arguably includes simple and complex demonstratives, tense, expressions taking implicit arguments, gradable adjectives, quantifiers, ‘only’, possessives, conditionals, modals and more. I call these expressions supplementives to highlight the fact that their context invariant meanings need to be supplemented in context for them to secure semantic values in context. I claim that supplementives differ from each other in the following two ways: 1. The degree to which normal speakers are explicitly aware that the expression is contextually sensitive. 2. The degree to which ordinary speakers are explicitly aware of what sort of semantic value in context the expression takes. I hold the view that semantic values in context for supplementives are fixed by recognizable speaker intentions. However, I argue that given the differences between supplementives with respect to 1 and 2, the intentions fixing the semantic values in context of supplementives that differ with respect to 1 and 2 will themselves be different, while still all being speaker intentions that some entity o be the semantic value of the use of the supplementive in context. (shrink)

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