. Computational chemistry grew in a new era of “desktop modeling,” which coincided with a growing demand for modeling software, especially from the pharmaceutical industry. Parameterization of models in computational chemistry is an arduous enterprise, and we argue that this activity leads, in this specific context, to tensions among scientists regarding the epistemic opacity transparency of parameterized methods and the software implementing them. We relate one flame war from the Computational Chemistry mailing List in order to assess in detail (...) the relationships between modeling methods, parameterization, software and the various forms of their enclosure or disclosure. Our claim is that parameterization issues are an important and often neglected source of epistemic opacity and that this opacity is entangled in methods and software alike. Models and software must be addressed together to understand the epistemological tensions at stake. (shrink)

If some aspects of human behavior are too murky to see into, others are too close and transparent to examine; one that has eluded both scientists and philosophers is how speakers of natural languages make words and expressions refer to specific objects in the world. Marshalling his expertise in philosophy, computers, and system science (State U. of.

Philosophers have recently focused on critical, epistemological challenges that arise from the opacity of deep neural networks. One might conclude from this literature that doing good science with opaque models is exceptionally challenging, if not impossible. Yet, this is hard to square with the recent boom in optimism for AI in science alongside a flood of recent scientific breakthroughs driven by AI methods. In this paper, I argue that the disconnect between philosophical pessimism and scientific optimism is driven by (...) a failure to examine how AI is actually used in science. I show that, in order to understand the epistemic justification for AI-powered breakthroughs, philosophers must examine the role played by deep learning as part of a wider process of discovery. The philosophical distinction between the 'context of discovery' and the 'context of justification' is helpful in this regard. I demonstrate the importance of attending to this distinction with two cases drawn from the scientific literature, and show that epistemic opacity need not diminish AI's capacity to lead scientists to significant and justifiable breakthroughs. (shrink)

Deep neural networks are said to be opaque, impeding the development of safe and trustworthy artificial intelligence, but where this opacity stems from is less clear. What are the sufficient properties for neural network opacity? Here, I discuss five common properties of deep neural networks and two different kinds of opacity. Which of these properties are sufficient for what type of opacity? I show how each kind of opacity stems from only one of these five (...) properties, and then discuss to what extent the two kinds of opacity can be mitigated by explainability methods. (shrink)

We explore the view that Frege's puzzle is a source of straightforward counterexamples to Leibniz's law. Taking this seriously requires us to revise the classical logic of quantifiers and identity; we work out the options, in the context of higher-order logic. The logics we arrive at provide the resources for a straightforward semantics of attitude reports that is consistent with the Millian thesis that the meaning of a name is just the thing it stands for. We provide models to show (...) that some of these logics are non-degenerate. (shrink)

Real human agents, even when they are rational by everyday standards, sometimes assign different credences to objectively equivalent statements, such as ‘Orwell is a writer’ and ‘E.A. Blair is a writer’, or credences less than 1 to necessarily true statements, such as not-yet-proven theorems of arithmetic. Anna Mahtani calls this the phenomenon of ‘opacity’. Opaque credences seem probabilistically incoherent, which goes against a key modelling assumption of probability theory. I sketch a modelling strategy for capturing opaque credence assignments without (...) abandoning probabilistic coherence. I draw on ideas from judgement-aggregation theory, where we face similar challenges of defining the ‘objects of judgement’. (shrink)

The Fukushima nuclear accident from 2011 provided an occasion for the public display of radiation maps generated using decision-support systems for nuclear emergency management. Such systems rely on computer models for simulating the atmospheric dispersion of radioactive materials and estimating potential doses in the event of a radioactive release from a nuclear reactor. In Germany, as in Japan, such systems are part of the national emergency response apparatus and, in case of accidents, they can be used by emergency task forces (...) for planning radioprotection countermeasures. In this context, the paper addresses the epistemology of dose projections by critically analyzing some of the sources of epistemic opacity and non-knowledge affecting them, and the different methods and practices used by German radioprotection experts to improve their trustworthiness and reliability. It will be argued that dose projections are part of an entire radioprotection regime or assemblage built around the belief that the effects of nuclear accidents can be effectively mitigated thanks to the simulation technologies underlying different protocols and practices of nuclear preparedness. And, as the Fukushima experience showed, some of these expectations will not be met in real emergencies due to the inherent uncertainties entailed by the use of dose projections when planning protective countermeasures. (shrink)

Harold Sonny Ladoo’s 1972 novel No Pain Like This Body has been analyzed for its seminal representation of the traumas experienced by a formerly indentured Indo-Trinidadian family in the early twentieth century. However, relatively little attention has been given to Ladoo’s experimentation with multiple languages, particularly English, Trinidadian Creole, and Hindi. This article argues that Ladoo’s multilingualism offers a guide for approaching the traumatic experiences he represents. While some aspects of the novel, such as its glossary, make the characters’ language (...) more comprehensible, others, such as the orthography Ladoo chooses to represent Creole speech, deliberately distance the reader. Using decolonial theorists of language, particularly Édouard Glissant's writing on multi­lingualism and opacity, this article considers Ladoo’s use of multilingualism both as a limit to readers’ understanding as well as an invitation to continued engagement with those aspects of the text that are resistant to easy comprehension. This article contrasts opacity as a reading methodology with some of the dominant paradigms for understanding linguistic difference in the field of comparative literature, which rely on linguistic and textual mastery. Ultimately, the article proposes reading multilingual texts through opacity as a model for decolonial reading in which creative, active engagement with the text can produce solidarity without requiring complete transparency. (shrink)

Concerns over epistemic opacity abound in contemporary debates on Artificial Intelligence (AI). However, it is not always clear to what extent these concerns refer to the same set of problems. We can observe, first, that the terms 'transparency' and 'opacity' are used either in reference to the computational elements of an AI model or to the models to which they pertain. Second, opacity and transparency might either be understood to refer to the properties of AI systems (...) or to the epistemic situation of human agents with respect to these systems. While these diagnoses are independently discussed in the literature, juxtaposing them and exploring possible interrelations will help to get a view of the relevant distinctions between conceptions of opacity and their empirical bearing. In pursuit of this aim, two pertinent conditions affecting computer models in general and contemporary AI in particular are outlined and discussed: opacity as a problem of computational tractability and opacity as a problem of the universality of the computational method. (shrink)

The article focuses on the value of opacity in communication. Jurgen Habermas's and M.M. Bakhtin's attitudes toward transparent or undistorted communication define almost antithetical approaches to the relationship between public discourse and autonomy. Habermas, both in his theory of communicative action and in his discourse ethics, assumes that transparent communication is possible and actually makes transparency a necessary condition for the legitimation of social norms. Yet, there is a sense in which the same kind of transparency that offers the (...) possibility of rational and autonomous selfhood to Habermas signifies vulnerability and tyranny to Bakhtin. In contrast to Habermas, Bakhtin assumes that utterances can never be perfectly transparent because the words that comprise them always carry meanings that exceed the intentions of speakers. This excess semantic value inevitably distorts speakers' intentions, even if only slightly. Bakhtin's suspicion of transparency leads him to develop a model of autonomy that revolves around the individual's ability to resist the emergence of transparency. However, Bakhtin attributes a positive ethical value to certain kinds of opacity because it is in the differences and distortions that Bakhtin situates the process through which individuals construct autonomy. Thus, in a kind of communicative paradox, opacity and ambiguity play the same liberating role in Bakhtin's thought that transparency and clarity play in Habermas's. (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)

I present the contours of an explanatory model of legitimacy that directs the focus away from normative questions and onto specific mechanisms of reality construction at play in constituting social orders. The key assumption informing the model is that stable orders rely fundamentally on their capacities to construct separate spheres of social reality, by which they exempt critical parts of reality from the burden of legitimation. I argue that an order's legitimacy ultimately depends on its ability to confine (...) the question of legitimacy by relegating authorship of reality to opaque sources that are separated epistemically from the institutional order. The effective reification of critical zones of reality is shown to be a functional precondition for institutional stability. I show that capitalist democracy, in particular, depends on this mode of constructing ‘passive legitimacy’, which constitutes a key obstacle for any attempt to transform it towards an ecologically and socially sustainable formation. (shrink)

This paper examines the logical and metaphysical consequences of denying Leibniz's Law, the principle that if t1= t2, then φ(t1) if and only if φ(t2). Recently, Caie, Goodman, and Lederman (2020) and Bacon and Russell (2019) have proposed sophisticated logical systems permitting violations of Leibniz's Law. We show that their systems conflict with widely held, attractive principles concerning the metaphysics of individuals. Only by adopting a highly revisionary picture, on which there is no finest-grained equivalence relation, can a well-motivated metaphysics (...) for rejecting Leibniz's Law be developed. We sketch one such picture—a metaphysics of stuff. Stuff ontologies can be initially motivated through ordinary language: stuff stands to mass nouns as objects stand to count nouns. The stuff ontology we propose takes stuff to be fundamental and views the world as composed of an infinite descending hierarchy of kinds and portions of stuff. We defend the coherence of this picture and offer a model theory demonstrating that it can be consistently formalized. (shrink)

In 2017 AlphaZero, a neural network-based chess engine shook the chess world by convincingly beating Stockfish, the highest-rated chess engine. In this paper, I describe the technical differences between the two chess engines and based on that, I discuss the impact of the modeling choices on the respective epistemic opacities. I argue that the success of AlphaZero’s approach with neural networks and reinforcement learning is counterbalanced by an increase in the epistemic opacity of the resulting model.

The problem of epistemic opacity in Artificial Intelligence is often characterised as a problem of intransparent algorithms that give rise to intransparent models. However, the degrees of transparency of an AI model should not be taken as an absolute measure of the properties of its algorithms but of the model’s degree of intelligibility to human users. Its epistemically relevant elements are to be specified on various levels above and beyond the computational one. In order to elucidate this (...) claim, I first contrast computer models and their claims to algorithm-based universality with cybernetics-style analogue models and their claims to structural isomorphism between elements of model and target system. While analogue models aim at perceptually or conceptually accessible model-target relations, computer models give rise to a specific kind of underdetermination in these relations that needs to be addressed in specific ways. I then undertake a comparison between two contemporary AI approaches that, although related, distinctly align with the above modelling paradigms and represent distinct strategies towards model intelligibility: Deep Neural Networks and Predictive Processing. I conclude that their respective degrees of epistemic transparency primarily depend on the underlying purposes of modelling, not on their computational properties. (shrink)

Simple idealized models seem to provide more understanding than opaque, complex, and hyper-realistic models. However, an increasing number of scientists are going in the opposite direction by utilizing opaque machine learning models to make predictions and draw inferences, suggesting that scientists are opting for models that have less potential for understanding. Are scientists trading understanding for some other epistemic or pragmatic good when they choose a machine learning model? Or are the assumptions behind why minimal models provide understanding misguided? (...) In this paper, using the case of deep neural networks, I argue that it is not the complexity or black box nature of a model that limits how much understanding the model provides. Instead, it is a lack of scientific and empirical evidence supporting the link that connects a model to the target phenomenon that primarily prohibits understanding. (shrink)

Abstract: How can the decisions of “autonomous” individuals provide a rationale for freedom and self-governance if a mechanical and causal sense of the self leads us to question the foundational nature of the individual? If most of our decisions originate in brain function below the level of consciousness, we live in a virtual world produced by mechanisms outside our control, arising from transparent self-models of which we are not aware. “Opacity,” the gift of not perceiving directly, of not automatically (...) believing what we are experiencing, is precisely what makes it possible for us to question our first-person perspectives. While we do not have direct access to the mechanisms that give rise to our sense of ourselves, our investigation can make it possible for us to analyze and understand those very mechanisms consciously. The deployment of that knowledge in the act of decision making and the infinite regress of interacting conscious individuals underlies our existential freedom. (shrink)

With the increasing degrees A.I.s and machines are applied across different social contexts, the need for implementing ethics in A.I.s is pressing. In this paper, we argue for a pluralist hybrid model for the implementation of moral A.I.s. We first survey current approaches to moral A.I.s and their inherent limitations. Then we propose the pluralist hybrid approach and show how these limitations of moral A.I.s can be partly alleviated by the pluralist hybrid approach. The core ethical decision-making capacity of (...) an A.I. based on the pluralist hybrid approach consists of two systems. The first is a deterministic algorithm system that incorporates different moral rules for making explicit moral decisions. The second is a machine learning system that calculates the value of the variables required by the application of moral principles. The pluralist hybrid system is better than the existing proposals as (i) it better addresses the moral disagreement problem of the top-down approach by including different moral principles, while (ii) it reduces the opacity of ethical decision-making compared with a bottom-up system by implementing explicit moral principles for moral decision making. (shrink)

I examine error diagnosis (model-model disagreement) in climate model intercomparisons including its difficulties, fruitful examples, and prospects for streamlining error diagnosis. I suggest that features of climate model intercomparisons pose a more significant challenge for error diagnosis than do features of individual model construction and complexity. Such features of intercomparisons include, e.g., the number of models involved, how models from different institutions interrelate, and what scientists know about each model. By considering numerous examples in (...) the climate modeling literature, I distill general strategies (e.g., employing physical reasoning and using dimension reduction techniques) used to diagnose model error. Based on these examples, I argue that an error repertoire could be beneficial for improving error diagnosis in climate modeling, although constructing one faces several difficulties. Finally, I suggest that the practice of error diagnosis demonstrates that scientists have a tacit-yet-working understanding of their models which has been under-appreciated by some philosophers. (shrink)

One of the main worries with machine learning model opacity is that we cannot know enough about how the model works to fully understand the decisions they make. But how much is model opacity really a problem? This chapter argues that the problem of machine learning model opacity is entangled with non-epistemic values. The chapter considers three different stages of the machine learning modeling process that corresponds to understanding phenomena: (i) model acceptance (...) and linking the model to the phenomenon, (ii) explanation, and (iii) attributions of understanding. At each of these stages, non-epistemic values can, in part, determine how much machine learning model opacity poses a problem. (shrink)

Over the course of the twentieth century, hydraulic engineering has come to rely primarily on the use of computational models. Disco and van den Ende hint towards the reasons for widespread adoption of computational models by pointing out that such models fulfill a crucial role as management tools in Dutch water management, and meet a more general desire to quantify water-related phenomena. The successful application of computational models implies blackboxing : “[w]hen a machine runs efficiently … one need focus only (...) on its inputs and outputs and not on its internal complexity. Thus, paradoxically, the more science and technology succeed, the more opaque and obscure they become”.. (shrink)

This paper provides an epistemological and methodological analysis of the recent practice of using neural language models to simulate brain language processing. It is argued that, on the one hand, this practice can be understood as an instance of the traditional simulative method in artificial intelligence, following a mechanistic understanding of the mind; on the other hand, that it modifies the simulative method significantly. Firstly, neural language models are introduced; a study case showing how neural language models are being applied (...) in cognitive neuroscience for simulative purposes is then presented; after recalling the main epistemological features of the simulative method in artificial intelligence, it is finally highlighted how the epistemic opacity of neural language models is tackled by using the brain itself to simulate the neural language model and to test hypotheses about it, in what is called here a co-simulation. (shrink)

The proliferation of AI systems across all domains of life as well as the complexification and opacity of algorithmic techniques, epitomised by the bourgeoning field of Deep Learning (DL), call for new methods in the Humanities for reflecting on the techno-human relation in a way that places the technical operation at its core. Grounded on the work of the philosopher of technology Gilbert Simondon, this paper puts forward individuation theory as a valuable approach to reflect on contemporary information technologies, (...) offering an analysis of the functioning of deep neural networks (DNNs), a type of data-driven computational models at the core of major breakthroughs in AI. The purpose of this article is threefold: (1) to demonstrate how a joint reading of Simondon’s mechanology and individuation theory, foregrounded in the Simondonian concept of information, can cast new light on contemporary algorithmic techniques by considering their situated emergence as opposed to technical lineage; (2) to suspend a predictive framing of AI systems, particularly DL techniques, so as to probe into their technical operation, accounting for the data-driven individuation of these models and the integration of potentials as functionality; and finally, (3) to argue that individuation theory might in fact de-individuate AI, in the sense of disassembling the already-there, the constituted, paving the way for questioning the potentialities for data and their algorithmic relationality to articulate the unfolding of everyday life. (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)

This chapter responds to Michael Tamir and Elay Shech’s chapter “Understanding from Deep Learning Models in Context.”.

The robust behavior of the patent citation network is a complex target of recent bottom-up models in science. This paper investigates the purpose and testing of three especially simple bottom-up models of the citation count distribution observed in the patent citation network. The complex causal webs in the models generate weakly emergent patterns of behavior, and this explains both the need for empirical observation of computer simulations of the models and the epistemic harmlessness of the resulting epistemic opacity.

It has often been argued that we face a trade-off between accuracy and opacity in deep learning models. The idea is that we can only harness the accuracy of deep learning models by simultaneously accepting that the grounds for the models’ decision-making are epistemically opaque to us. In this paper, we ask the following question: what are the prospects of making deep learning models transparent without compromising on their accuracy? We argue that the answer to this question depends on (...) which kind of opacity we have in mind. If we focus on the standard notion of opacity, which tracks the internal complexities of deep learning models, we argue that existing explainable AI (XAI) techniques show us that the prospects look relatively good. But, as it has recently been argued in the literature, there is another notion of opacity that concerns factors external to the model. We argue that there are at least two types of external opacity—link opacity and structure opacity—and that existing XAI techniques can to some extent help us reduce the former but not the latter. (shrink)

Pinyin is a phonological encoding system used to spell modern Chinese Mandarin due to the phonological opacity of Chinese characters. The present study examined the role of Pinyin spelling in the reading abilities of adolescents learning Chinese as a foreign language. A total of 158 Indonesian senior primary students were tested on Pinyin spelling, character production, listening comprehension, depth of vocabulary knowledge, and reading comprehension. Pinyin spelling skills were assessed by two measures, Pinyin Dictation and Pinyin Tagging. Path analysis (...) revealed that even after controlling for the effect of character production, Pinyin dictation performance influenced reading comprehension through the mediating effect of listening comprehension and the depth of vocabulary knowledge, and Pinyin tagging performance also influenced reading comprehension through the mediating effect of the depth of vocabulary knowledge. The results highlight the importance of Pinyin skills for Chinese reading abilities of CFL learners. As a reliable and explicit indicator of specifying Chinese phonological representations and processing, Pinyin spelling has a long-term and multifaceted influence on higher-level CFL abilities. (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)

Models developed using machine learning are increasingly prevalent in scientific research. At the same time, these models are notoriously opaque. Explainable AI aims to mitigate the impact of opacity by rendering opaque models transparent. More than being just the solution to a problem, however, Explainable AI can also play an invaluable role in scientific exploration. This paper describes how post-hoc analytic techniques from Explainable AI can be used to refine target phenomena in medical science, to identify starting points for (...) future investigations of causal relationships, and to generate possible explanations of target phenomena in cognitive science. In this way, this paper describes how Explainable AI—over and above machine learning itself—contributes to the efficiency and scope of data-driven scientific research. (shrink)

Computational systems biologists create and manipulate computational models of biological systems, but they do not always have straightforward epistemic access to the content and behavioural profile of such models because of their length, coding idiosyncrasies, and formal complexity. This creates difficulties both for modellers in their research groups and for their bioscience collaborators who rely on these models. In this paper we introduce a new kind of visualization that was developed to address just this sort of epistemic opacity. The (...) visualization is unusual in that it depicts the dynamics and structure of a computer model instead of that model’s target system, and because it is generated algorithmically. Using considerations from epistemology and aesthetics, we explore how this new kind of visualization increases scientific understanding of the content and function of computer models in systems biology to reduce epistemic opacity. (shrink)

This paper develops an account of the opacity problem in medical machine learning (ML). Guided by pragmatist assumptions, I argue that opacity in ML models is problematic insofar as it potentially undermines the achievement of two key purposes: ensuring generalizability and optimizing clinician–machine decision-making. Three opacity amelioration strategies are examined, with explainable artificial intelligence (XAI) as the predominant approach, challenged by two revisionary strategies in the form of reliabilism and the interpretability by design. Comparing the three strategies, (...) I argue that interpretability by design is most promising to overcome opacity in medical ML. Looking beyond the individual opacity amelioration strategies, the paper also contributes to a deeper understanding of the problem space and the solution space regarding opacity in medical ML. (shrink)

Many AI systems that make important decisions are black boxes: how they function is opaque even to their developers. This is due to their high complexity and to the fact that they are trained rather than programmed. Efforts to alleviate the opacity of black box systems are typically discussed in terms of transparency, interpretability, and explainability. However, there is little agreement about what these key concepts mean, which makes it difficult to adjudicate the success or promise of opacity (...) alleviation methods. I argue for a unified account of these key concepts that treats the concept of understanding as fundamental. This allows resources from the philosophy of science and the epistemology of understanding to help guide opacity alleviation efforts. A first significant benefit of this understanding account is that it defuses one of the primary, in-principle objections to post hoc explainable AI (XAI) methods. This “rationalization objection” argues that XAI methods provide mere rationalizations rather than genuine explanations. This is because XAI methods involve using a separate “explanation” system to approximate the original black box system. These explanation systems function in a completely different way than the original system, yet XAI methods make inferences about the original system based on the behavior of the explanation system. I argue that, if we conceive of XAI methods as idealized scientific models, this rationalization worry is dissolved. Idealized scientific models misrepresent their target phenomena, yet are capable of providing significant and genuine understanding of their targets. (shrink)

It is one thing to consider what human rights have been and another to inquire into what they could be. In this essay, I present a history of human rights vis-à-vis decolonization. I follow the scholarship of Samuel Moyn to suggest that human rights presented a “moral alternative” to political utopias. The question remains how to politicize the moral energy around human rights today. I argue that defending what Édouard Glissant calls a “right to opacity” could politicize the ethical (...) energy around human rights today. Glissant’s right to opacity outlines a blueprint for the praxis of human rights to shift from a “functional model” to a “critical model,” to use Enrique Dussel’s distinction. My ultimate aim is to show how social movements around human rights and decolonization could converge today. (shrink)

Psychedelic ingestion and meditative practice are both ancient methods for altering consciousness that became widely known in Western society in the second half of the 20th century. Do the similarities begin and end there, or do these methods – as many have claimed over the years – share some deeper common elements? In this chapter I take a neurophilosophical approach to this question and argue that there are, indeed, deeper commonalities. Recent empirical studies show that psychedelics and meditation modulate overlapping (...) brain networks involved in the sense of self, salience, and attention; moreover, psychedelics can occasion lasting increases in “mindfulness-related capacities” for taking a non-reactive stance on one’s inner experience (e.g. Sampedro et al. 2017). The self-binding theory of psychedelic ego dissolution (Letheby and Gerrans 2017) offers a plausible explanation of these findings: by disrupting self-related beliefs implemented in high-level cortical networks, both psychedelics and meditation can “unbind” mental contents from one’s self-model, moving these contents along the continuum from phenomenal transparency to opacity (cf. Metzinger 2003). In other words, both psychedelics and meditation can expose and weaken our foundational beliefs about our own identity, allowing us to disidentify with these beliefs and see them as “just thoughts”. There are connections between these ideas and recent arguments suggesting that psychedelic use may have epistemic benefits consistent with philosophical naturalism (Letheby 2015, 2016, 2019). I conclude with a proposal: these connections may help in thinking about the putative epistemic benefits of meditation practice from a naturalistic perspective. (shrink)

The artificial intelligence models with machine learning that exhibit the best predictive accuracy, and therefore, the most powerful ones, are, paradoxically, those with the most opaque black-box architectures. At the same time, the unstoppable computerization of advanced industrial societies demands the use of these machines in a growing number of domains. The conjunction of both phenomena gives rise to a control problem on AI that in this paper we analyze by dividing the issue into two. First, we carry out an (...) epistemological examination of the AI’s opacity in light of the latest techniques to remedy it. And second, we evaluate the rationality of delegating tasks in opaque agents. (shrink)

While AI systems are increasingly assuming roles traditionally occupied by human epistemic authorities (EAs), their epistemological status remains unclear. This paper aims to address this lacuna by assessing the potential for AI systems to be recognized as artificial epistemic authorities. In a first step, I examine the arguments against considering AI systems as EAs, in particular the established model of EAs as engaging in intentional belief transfer via testimony to laypeople – a process seemingly inapplicable to intentionless and beliefless (...) AI. Despite this, AI systems exhibit striking epistemic parallels with human EAs, including epistemic asymmetry and opacity that give rise to comparable challenges for both laypeople and AI users. These challenges include the identification problem – how to recognize reliable EAs/AI systems – and the deference problem – determining the appropriate epistemic stance towards EAs/AI systems. Faced with this dilemma, I discuss three possible solutions: 1. reject the concept of artificial EAs, 2. accept that AI can possess beliefs and intentions and thus align with the standard model, or 3. develop an alternative model that encompasses artificial EAs. I argue that while each option has its benefits and costs, a particularly strong case can be made for option 3. (shrink)

Computers have transformed science and help to extend the boundaries of human knowledge. However, does the validation and diffusion of results of computational inquiries and computer simulations call for a novel epistemological analysis? I discuss how the notion of novelty should be cashed out to investigate this issue meaningfully and argue that a consequentialist framework similar to the one used by Goldman to develop social epistemologySocial epistemology can be helpful at this point. I highlight computational, mathematical, representational, and social stages (...) on which the validity of simulation-based belief-generating processes hinges, and emphasize that their epistemic impact depends on the scientific practices that scientists adopt at these different stages. I further argue that epistemologists cannot ignore these partially novel issues and conclude that the epistemology of computational inquiries needs to go beyond that of models and scientific representations and has cognitive, social, and in the present case computational, dimensions. (shrink)

Decision-making assisted by algorithms developed by machine learning is increasingly determining our lives. Unfortunately, full opacity about the process is the norm. Would transparency contribute to restoring accountability for such systems as is often maintained? Several objections to full transparency are examined: the loss of privacy when datasets become public, the perverse effects of disclosure of the very algorithms themselves, the potential loss of companies’ competitive edge, and the limited gains in answerability to be expected since sophisticated algorithms usually (...) are inherently opaque. It is concluded that, at least presently, full transparency for oversight bodies alone is the only feasible option; extending it to the public at large is normally not advisable. Moreover, it is argued that algorithmic decisions preferably should become more understandable; to that effect, the models of machine learning to be employed should either be interpreted ex post or be interpretable by design ex ante. (shrink)

Decision-making assisted by algorithms developed by machine learning is increasingly determining our lives. Unfortunately, full opacity about the process is the norm. Would transparency contribute to restoring accountability for such systems as is often maintained? Several objections to full transparency are examined: the loss of privacy when datasets become public, the perverse effects of disclosure of the very algorithms themselves (“gaming the system” in particular), the potential loss of companies’ competitive edge, and the limited gains in answerability to be (...) expected since sophisticated algorithms usually are inherently opaque. It is concluded that, at least presently, full transparency for oversight bodies alone is the only feasible option; extending it to the public at large is normally not advisable. Moreover, it is argued that algorithmic decisions preferably should become more understandable; to that effect, the models of machine learning to be employed should either be interpreted ex post or be interpretable by design ex ante. (shrink)

In my dissertation, I argue that the enterprise of corrective justice requires answering questions about what is unjust and how we ought to set and pursue corrective justice goals. To answer these questions in a way that will allow us to correct for the persistent and entrenched injustices which result from processes of stratification in our society, I’ll put forward a two-tiered social contract theory, which will allow us to approach these questions in a way that will capture the agreement (...) we can have and allow us to structure the persistent disagreement that we cannot and ought not avoid. In the first tier—the cooperative position—the parties know only that they’re in what I’m calling the ten circumstances of cooperation—the circumstances in which modern cooperation takes place. These circumstances includes the social complexity and opacity inherent in complex cooperative arrangements which have epistemic effects on our social cognition in ways that make our view of our social arrangements at best partial, and work to obscure important features of our social reality, especially those affecting those at the bottom of our stratified social relations. This tier generates four cooperative ideals—proficient rationality, democratic equality, social understanding, and dignity—that function as a theory of justice which can gain an overlapping consensus in society but still allow for meaningful disagreement to take place. In the second tier—the corrective position—the parties have access to the circumstances of cooperation, four cooperative ideals, and a general and specific domination contract. These domination contracts model how stratified social relations operate to constrain people within social systems, including the ways that people think about those systems. I argue that the parties would affirm a principle of corrective justice as involving coalitions of differently-situated actors working together over time to dismantle unjust mechanisms, and five desiderata for applying that principle to circumstances as we confront them. They would do so because that is the only way to overcome the practical and epistemic barriers put in place by the domination contracts. (shrink)

In the early 20th century anthropologists collected sounds, images and artifacts to represent traditional cultures. Under the direction of Franz Boas, anthropologists working for the American Museum of Natural History's JesupNorth Pacific Expedition documented a variety of northeastern Siberian shamanisms. Demonstrations staged for the phonograph and the camera served as models for museum representations. These ethnographic inscriptions, together with the collection of texts and sacred objects, documented shamanistic traditions; yet ceremonial traditions remained partially obscured, resisting full intelligibility. The complexity of (...) variations suggests that mystification and opacity are central to shamanic efficacy and the continued vitality of performative ritual, and that Siberian shamanic practices are inherently resistant to anthropological modes of interpretation. (shrink)

Thought experiments and simulation experiments are compared and contrasted with each other. While the former rely on epistemic transparency as a working condition, in the latter complexity of model dynamics leads to epistemic opacity. The difference is elucidated by a discussion of the different kinds of iteration that are at work in both sorts of experiment.

In light of recent advances in machine learning for medical applications, the automation of medical diagnostics is imminent. That said, before machine learning algorithms find their way into clinical practice, various problems at the epistemic level need to be overcome. In this paper, we discuss different sources of uncertainty arising for clinicians trying to evaluate the trustworthiness of algorithmic evidence when making diagnostic judgments. Thereby, we examine many of the limitations of current machine learning algorithms (with deep learning in particular) (...) and highlight their relevance for medical diagnostics. Among the problems we inspect are the theoretical foundations of deep learning (which are not yet adequately understood), the opacity of algorithmic decisions, and the vulnerabilities of machine learning models, as well as concerns regarding the quality of medical data used to train the models. Building on this, we discuss different desiderata for an uncertainty amelioration strategy that ensures that the integration of machine learning into clinical settings proves to be medically beneficial in a meaningful way. (shrink)

Background Despite continuous performance improvements, especially in clinical contexts, a major challenge of Artificial Intelligence based Decision Support Systems (AI-DSS) remains their degree of epistemic opacity. The conditions of and the solutions for the justified use of the occasionally unexplainable technology in healthcare are an active field of research. In March 2024, the European Union agreed upon the Artificial Intelligence Act (AIA), requiring medical AI-DSS to be ad-hoc explainable or to use post-hoc explainability methods. The ethical debate does not (...) seem to settle on this requirement yet. This systematic review aims to outline and categorize the positions and arguments in the ethical debate. Methods We conducted a literature search on PubMed, BASE, and Scopus for English-speaking scientific peer-reviewed publications from 2016 to 2024. The inclusion criterion was to give explicit requirements of explainability for AI-DSS in healthcare and reason for it. Non-domain-specific documents, as well as surveys, reviews, and meta-analyses were excluded. The ethical requirements for explainability outlined in the documents were qualitatively analyzed with respect to arguments for the requirement of explainability and the required level of explainability. Results The literature search resulted in 1662 documents; 44 documents were included in the review after eligibility screening of the remaining full texts. Our analysis showed that 17 records argue in favor of the requirement of explainable AI methods (xAI) or ad-hoc explainable models, providing 9 categories of arguments. The other 27 records argued against a general requirement, providing 11 categories of arguments. Also, we found that 14 works advocate the need for context-dependent levels of explainability, as opposed to 30 documents, arguing for context-independent, absolute standards. Conclusions The systematic review of reasons shows no clear agreement on the requirement of post-hoc explainability methods or ad-hoc explainable models for AI-DSS in healthcare. The arguments found in the debate were referenced and responded to from different perspectives, demonstrating an interactive discourse. Policymakers and researchers should watch the development of the debate closely. Conversely, ethicists should be well informed by empirical and technical research, given the frequency of advancements in the field. (shrink)

Like any science marked by high uncertainty, climate science is characterized by a widespread use of expert judgment. In this paper, we first show that, in climate science, expert judgment is used to overcome uncertainty, thus playing a crucial role in the domain and even at times supplanting models. One is left to wonder to what extent it is legitimate to assign expert judgment such a status as an epistemic superiority in the climate context, especially as the production of expert (...) judgment is particularly opaque. To begin answering this question, we highlight the key components of expert judgment. We then argue that the justification for the status and use of expert judgment depends on the competence and the individual subjective features of the expert producing the judgment since expert judgment involves not only the expert's theoretical knowledge and tacit knowledge, but also their intuition and values. This goes against the objective ideal in science and the criteria from social epistemology which largely attempt to remove subjectivity from expertise. (shrink)

Machine learning systems are making considerable inroads in society owing to their ability to recognize and predict patterns. However, the decision-making logic of some widely used machine learning models, such as deep neural networks, is characterized by opacity, thereby rendering them exceedingly difficult for humans to understand and explain and, as a result, potentially risky to use. Considering the importance of addressing this opacity, this paper calls for research that studies empirically and theoretically how machine learning experts and (...) users seek to attain machine learning explainability. Focusing on automated trading, we take steps in this direction by analyzing a trading firm’s quest for explaining its deep neural network system’s actionable predictions. We demonstrate that this explainability effort involves a particular form of human–machine interaction that contains both anthropomorphic and technomorphic elements. We discuss this attempt to attain machine learning explainability in light of reflections on cross-species companionship and consider it an example of human–machine companionship. (shrink)

From Sartre to Levinas, continental philosophers have looked to the example of the Jew as the paradigmatic object of and model for ethical inquiry. Levinas, for example, powerfully dedicates his 1974 book Otherwise than Being to the victims of the Holocaust, and turns attention to the state of philosophy after Auschwitz. Such an ethics radically challenges prior notions of autonomy and comprehension-two key ideas for traditional ethical theory and, more generally, the Greek tradition. It seeks to respect the (...) class='Hi'>opacity of the other and avoid the dangers of hermeneutic violence. But how does such an ethics of the other translate into real, everyday life? What is at stake in thinking the other as Jew? Is the alterity of the Jew simply a counter to Greek universalism? Is a rhetoric of exceptionalism, with its unavoidable ontological residue, at odds with shifting political realities? Within this paradigm, what then becomes of the Arab or Muslim, the other of the Jew, the other of the other, so to speak? This line of ethical thought-in its desire to bear witness to past suffering and come to terms with subjectivity after Auschwitz-arguably brackets from analysis present operations of power. Would, then, a more sensitive historical approach expose the Palestinian as the other of the Israeli? Here, Zahi Zalloua offers a challenging intervention into how we configure the contemporary. (shrink)

Advancements and deployments of AI-based systems, especially Deep Learning-driven generative language models, have accomplished impressive results over the past few years. Nevertheless, these remarkable achievements are intertwined with a related fear that such technologies might lead to a general relinquishing of our lives’s control to AIs. This concern, which also motivates the increasing interest in the eXplainable Artificial Intelligence (XAI) research field, is mostly caused by the opacity of the output of deep learning systems and the way that it (...) is generated, which is largely obscure to laypeople. A dialectical interaction with such systems may enhance the users’ understanding and build a more robust trust towards AI. Commonly employed as specific formalisms for modelling intra-agent communications, dialogue games prove to be useful tools to rely upon when dealing with user’s explanation needs. The literature already offers some dialectical protocols that expressly handle explanations and their delivery. This paper fully formalises the novel Explanation–Question–Response (EQR) dialogue and its properties, whose main purpose is to provide satisfactory information (i.e., justified according to argumentative semantics) whilst ensuring a simplified protocol, in comparison with other existing approaches, for humans and artificial agents. (shrink)