The machine conception of the organism (MCO) is one of the most pervasive notions in modern biology. However, it has not yet received much attention by philosophers of biology. The MCO has its origins in Cartesian natural philosophy, and it is based on the metaphorical redescription of the organism as a machine. In this paper I argue that although organisms and machines resemble each other in some basic respects, they are actually very different kinds of systems. I submit (...) that the most significant difference between organisms and machines is that the former are intrinsically purposive whereas the latter are extrinsically purposive. Using this distinction as a starting point, I discuss a wide range of dissimilarities between organisms and machines that collectively lay bare the inadequacy of the MCO as a general theory of living systems. To account for the MCO’s prevalence in biology, I distinguish between its theoretical, heuristic, and rhetorical functions. I explain why the MCO is valuable when it is employed heuristically but not theoretically, and finally I illustrate the serious problems that arise from the rhetorical appeal to the MCO. (shrink)

Machine learning algorithms may radically improve our ability to diagnose and treat disease. For moral, legal, and scientific reasons, it is essential that doctors and patients be able to understand and explain the predictions of these models. Scalable, customisable, and ethical solutions can be achieved by working together with relevant stakeholders, including patients, data scientists, and policy makers.

In, “The Turing Triage Test”, published in Ethics and Information Technology, I described a hypothetical scenario, modelled on the famous Turing Test for machine intelligence, which might serve as means of testing whether or not machines had achieved the moral standing of people. In this paper, I: (1) explain why the Turing Triage Test is of vital interest in the context of contemporary debates about the ethics of AI; (2) address some issues that complexify the application of this test; (...) and, (3) in doing so, defend a way of thinking about the question of the moral standing of intelligent machines, which takes the idea of “seriousness” seriously. This last objective is, in fact, my primary one and is motivated by the sense that, to date, much of the “philosophy” of AI has suffered from a profound failure to properly distinguish between things that we can say and things that we can really mean. (shrink)

The notion of computation has changed the world more than any previous expressions of knowledge. However, as know-how in its particular algorithmic embodiment, computation is closed to meaning. Therefore, computer-based data processing can only mimic life’s creative aspects, without being creative itself. AI’s current record of accomplishments shows that it automates tasks associated with intelligence, without being intelligent itself. Mistaking the abstract for the concrete has led to the religion of “everything is an output of computation”—even the humankind that conceived (...) the computer. The hypostatized role of computers explains the increased dependence on them. The convergence machine called deep learning is only the most recent form through which the deterministic theology of the machine claims more than what it actually is: extremely effective data processing. A proper understanding of complexity, as well as the need to distinguish between the reactive nature of the artificial and the anticipatory nature of the living are suggested as practical responses to the challenges posed by machine theology. (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)

Recent work on interpretability in machine learning and AI has focused on the building of simplified models that approximate the true criteria used to make decisions. These models are a useful pedagogical device for teaching trained professionals how to predict what decisions will be made by the complex system, and most importantly how the system might break. However, when considering any such model it’s important to remember Box’s maxim that "All models are wrong but some are useful." We focus (...) on the distinction between these models and explanations in philosophy and sociology. These models can be understood as a "do it yourself kit" for explanations, allowing a practitioner to directly answer "what if questions" or generate contrastive explanations without external assistance. Although a valuable ability, giving these models as explanations appears more difficult than necessary, and other forms of explanation may not have the same trade-offs. We contrast the different schools of thought on what makes an explanation, and suggest that machine learning might benefit from viewing the problem more broadly. (shrink)

In this paper, I reconstruct Robert Nozick's experience machine objection to hedonism about well-being. I then explain and briefly discuss the most important recent criticisms that have been made of it. Finally, I question the conventional wisdom that the experience machine, while it neatly disposes of hedonism, poses no problem for desire-based theories of well-being.

This study focuses in testing the power of reciprocity and leadership as collective action incentive structures and cooperation economic benefits in explaining collective action initiation in the context of a post-communist transition economy. The paper is based on a structured survey targeting Albanian export-oriented farmers. Different from most previous studies, this paper uses both regression analysis and machine learning procedure which is better suited for analysing non-linear relationships. The empirical findings are at odds with common sense that non-cooperation is (...) the dominant strategy, because the presence of tolerant reciprocators and leadership resources provide promising incentive structures for collective action development. KEYWORDS: Albania, collective action, collective action benefits, incentive structures, leadership, power, reciprocity. (shrink)

A properly programmed artificially intelligent agent may eventually have one duty, the duty to satisfice expected welfare. We explain this claim and defend it against objections.

This paper provides an analysis of the way in which two foundational principles of medical ethics–the trusted doctor and patient autonomy–can be undermined by the use of machine learning (ML) algorithms and addresses its legal significance. This paper can be a guide to both health care providers and other stakeholders about how to anticipate and in some cases mitigate ethical conflicts caused by the use of ML in healthcare. It can also be read as a road map as to (...) what needs to be done to achieve an acceptable level of explainability in an ML algorithm when it is used in a healthcare context. (shrink)

To create fair and accountable AI and robotics, we need precise regulation and better methods to certify, explain, and audit inscrutable systems.

In this article I explore various facets of Nozick’s famous thought experiment involving the experience machine. Nozick’s original target is hedonism—the view that the only intrinsic prudential value is pleasure. But the argument, if successful, undermines any experientialist theory, i.e. any theory that limits intrinsic prudential value to mental states. I first highlight problems arising from the way Nozick sets up the thought experiment. He asks us to imagine choosing whether or not to enter the machine and uses (...) our choice (or rather the choice he assumes most people will have) as evidence against experientialist theories. But for this strategy to succeed it must be possible to distinguish between self-interested and non-self-interested reasons for declining to enter the machine, and there is no obvious way to do this without begging the question against the hedonist. In successive sections I then (a) consider a common misconception of Nozick’s conclusion (that he thinks machine life is the worst life), (b) consider different intuitions about what is important to well-being but missing from machine life, and finally (c) explain what “the experience requirement” is, and describe its relationship to debates about experientialist theories. (shrink)

There has been a push in recent years to provide better explanations for how AIs make their decisions. Most of this push has come from the ethical concerns that go hand in hand with AIs making decisions that affect humans. Outside of the strictly ethical concerns that have prompted the study of explainable AIs (XAIs), there has been research interest in the mere possibility of creating XAIs in various domains. In general, the more accurate we make our models the harder (...) they are to explain. Go playing AIs like AlphaGo and KataGo provide fantastic examples of this phenomenon. In this paper, I discuss a non-exhaustive list of the leading theories of explanation and what each of these theories would say about the explainability of AI-played moves of Go. Finally, I consider the possibility of ever explaining AI-played Go moves in a way that meets the four principles of XAI. I conclude, somewhat pessimistically, that Go is not as imminently explainable as other domains. As such, the probability of having an XAI for Go that meets the four principles is low. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping (...) causal patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

Background: Emergentism as an ontology of consciousness leaves unanswered the question as to its mechanism. Aim: I aim to solve the Body-Mind problem by explaining how conscious organisms emerged on an evolutionary basis at various times in accordance with an accepted scientific principle, through a mechanism that cannot be understood, in principle. Proposal: The reason for this cloak of secrecy is found in a seeming contradiction in the behaviour of information with respect to the first two laws of thermodynamics. Information, (...) the microstate of particles within an isolated system’s macrostate, can, like first law energy, be neither created nor destroyed, yet the information in the system, like second law entropy, will inevitably increase. To explain information increasing without being created, Laplace’s demon is invoked, able to predict where each particle is destined. This doesn’t work for emerging events like consciousness, which are unpredictable. This can be understood in terms of the derivation of entropy, and the emergence of classical physics, from the Relativistic Transactional Interpretation of Quantum Mechanics. I propose that the increased entropy in a time-irreversible, unpredictable (emergent) isolated system requires the simultaneous deletion of information concerning the steps, or calculations, involved. Conclusion: Thus, the steps leading to consciousness are immediately destroyed and must remain a mystery. Implications include that entropy, not panpsychism, is the Universal principle generative of consciousness, that our being conscious proves that we are not predetermined, and that consciousness requires assuming an “entropy debt” that can only be repaid by living organisms, prohibiting the emergence of conscious machines. (shrink)

There is a well-documented Pre-Reflective Hostility against Machine Art (PRHMA), exemplified by the sentiments of fear and anxiety. How can it be explained? The present paper attempts to find the answer to this question by surveying a considerable amount of research on machine art. It is found that explanations of PRHMA based on the (alleged) fact that machine art lacks an element that is (allegedly) found in human art (for example, autonomy) do not work. Such explanations cannot (...) account for the sentiments of fear and anxiety present in PRHMA, because the art receiver could simply turn to human art for finding the element she is looking for. By contrast, an explanation based on the idea that machine art is “symbolically” a threat to human survival can be successful, since the art receiver’s turning from machine art to human art does not eliminate the (alleged) “symbolic” threat machine art poses for human survival. If there is a pre-reflective belief or feeling that machine art is such a threat, then it is perfectly understandable why humans exhibit a pre-reflective hostility against machine art. (shrink)

One of the most important ongoing debates in the philosophy of mind is the debate over the reality of the first-person character of consciousness.[1] Philosophers on one side of this debate hold that some features of experience are accessible only from a first-person standpoint. Some members of this camp, notably Frank Jackson, have maintained that epiphenomenal properties play roles in consciousness [2]; others, notably John R. Searle, have rejected dualism and regarded mental phenomena as entirely biological.[3] In the opposite camp (...) are philosophers who hold that all mental capacities are in some sense computational - or, more broadly, explainable in terms of features of information processing systems.[4] Consistent with this explanatory agenda, members of this camp normally deny that any aspect of mind is accessible solely from a first-person standpoint. This denial sometimes goes very far - even as far as Dennett's claim that the phenomenology of conscious experience does not really exist. (shrink)

his article will focus on the mechanistic origins of the computer metaphor, which forms the conceptual framework for the methodology of the cognitive sciences, some areas of artificial intelligence and the philosophy of mind. The connection between the history of computing technology, epistemology and the philosophy of mind is expressed through the metaphorical dictionaries of the philosophical discourse of a particular era. The conceptual clarification of this connection and the substantiation of the mechanistic components of the computer metaphor is the (...) main goal of this article. The statement is substantiated that the invention of mechanical computing devices, having a long history in the European engineering tradition, formed the prerequisites for the emergence of machine functionalism in the modern philosophy of mind. The idea of multiple implementation stems from the principle that a formal symbol system prescribes rules for the use of rational abstractions through the physical architecture of a computational engine. The article considers the reasons for the conceptual shift and reveals the semantic foundations for the metaphorical transfer of the properties of abstract objects from the theory of automata to the field of modern philosophy of mind. The criticism and ways of protecting the philosophical program of machine functionalism are analyzed by changing the content of the metaphor “Mind as machine”. The reasons for the stability of the information-computer approach in cognitive sciences are also disclosed and explained. (shrink)

This paper is a warning that objections based on thought experiments can be misleading because they may elicit judgments that, unbeknownst to the judger, have been seriously skewed by psychological biases. The fact that most people choose not to plug in to the Experience Machine in Nozick’s (1974) famous thought experiment has long been used as a knock-down objection to hedonism because it is widely thought to show that real experiences are more important to us than pleasurable experiences. This (...) paper argues that the commonplace choice to remain in reality when offered a life in the Experience Machine is best explained by status quo bias – the irrational preference for things to remain the same. An alternative thought experiment, empirical evidence, and discussion of how psychological biases can affect our judgments are provided to support this argument. (shrink)

We address the question of whether it is possible to operate a time machine by manipulating matter and energy so as to manufacture closed timelike curves. This question has received a great deal of attention in the physics literature, with attempts to prove no- go theorems based on classical general relativity and various hybrid theories serving as steps along the way towards quantum gravity. Despite the effort put into these no-go theorems, there is no widely accepted definition of a (...) time machine. We explain the conundrum that must be faced in providing a satisfactory definition and propose a resolution. Roughly, we require that all extensions of the time machine region contain closed timelike curves; the actions of the time machine operator are then sufficiently "potent" to guarantee that closed timelike curves appear. We then review no-go theorems based on classical general relativity, semi-classical quantum gravity, quantum field theory on curved spacetime, and Euclidean quantum gravity. Our verdict on the question of our title is that no result of sufficient generality to underwrite a confident "yes" has been proven. Our review of the no-go results does, however, highlight several foundational problems at the intersection of general relativity and quantum physics that lend substance to the search for an answer. (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)

I distinguish two theses regarding technological successors to current humans (posthumans): an anthropologically bounded posthumanism (ABP) and an anthropologically unbounded posthumanism (AUP). ABP proposes transcendental conditions on agency that can be held to constrain the scope for “weirdness” in the space of possible posthumans a priori. AUP, by contrast, leaves the nature of posthuman agency to be settled empirically (or technologically). Given AUP there are no “future proof” constraints on the strangeness of posthuman agents. -/- In Posthuman Life I defended (...) AUP via a critique of Donald Davidson’s work on intentionality and a “naturalistic deconstruction” of transcendental phenomenology (See also Roden 2013). In this paper I extend this critique to Robert Brandom’s account of the relationship between normativity and intentionality in Making It Explicit (MIE) and in other writings. -/- Brandom’s account understands intentionality in terms of the capacity to undertake and ascribe inferential normative commitments. It makes “first class agency” dependent on the ability to participate in discursive social practices. It implies that posthumans – insofar as they qualify as agents at all – would need to be social and discursive beings. -/- The problem with this approach, I will argue, is that it replicates a problem that Brandom discerns in Dennett’s intentional stance approach. It tells us nothing about the conditions under which a being qualifies as a potential interpreter and thus little about the conditions for meaning, understanding or agency. -/- I support this diagnosis by showing that Brandom cannot explain how a non-sapient community could bootstrap itself into sapience by setting up a basic deontic scorekeeping system without appealing (along with Davidson and Dennett) to the ways in which an idealized observer would interpret their activity. -/- This strongly suggests that interpretationist and pragmatist accounts cannot explain the semantic or the intentional without regressing to assumptions about ideal interpreters or background practices whose scope they are incapable of delimiting. It follows that Anthropologically Unbounded Posthumanism is not seriously challenged by the claim that agency and meaning are “constituted” by social practices. -/- AUP implies that we can infer no claims about the denizens of “Posthuman Possibility Space” a priori, by reflecting on the pragmatic transcendental conditions for semantic content. We thus have no reason to suppose that posthuman agents would have to be subjects of discourse or, indeed, members of communities. The scope for posthuman weirdness can be determined by recourse to engineering alone. (shrink)

It has become customary to conceptualize the living cell as an intricate piece of machinery, different to a man-made machine only in terms of its superior complexity. This familiar understanding grounds the conviction that a cell's organization can be explained reductionistically, as well as the idea that its molecular pathways can be construed as deterministic circuits. The machine conception of the cell owes a great deal of its success to the methods traditionally used in molecular biology. However, the (...) recent introduction of novel experimental techniques capable of tracking individual molecules within cells in real time is leading to the rapid accumulation of data that are inconsistent with an engineering view of the cell. This paper examines four major domains of current research in which the challenges to the machine conception of the cell are particularly pronounced: cellular architecture, protein complexes, intracellular transport, and cellular behaviour. It argues that a new theoretical understanding of the cell is emerging from the study of these phenomena which emphasizes the dynamic, self-organizing nature of its constitution, the fluidity and plasticity of its components, and the stochasticity and non-linearity of its underlying processes. (shrink)

The advance of mechanism in science and philosophy in the 17th century created a great interest to machines or automata. Leibniz was no exception - in an early memoir Drôle de pensée he wrote admiringly about a machine that could walk on water, exhibited in Paris. The idea of automatic processing in general had a large role in his thought, as can be seen, for example, in his invention of the binary code and the so-called Calculemus!-model for solving controversies. (...) In metaphysics, the idea of an automata was expressed most clearly in the 1695 article New System of the Nature of Substances and their Communication, and the ensuing correspondence with, among others, Foucher, Bayle, Lamy, Jaquelot and Masham. In the article Leibniz discusses the soul as a spiritual machine in the context of pre-established harmony, arguing that God can "give to a substance at the outset a nature or internal force which could produce in it an orderly way (as in spiritual or formal automaton; but a free one, in the case of a substance which is endowed with a share of reason) everything that is going to happen to it, that is to say, all the appearances or expressions it is going to have, and all without the help of any created thing." The basic idea of Leibniz's spiritual machine is that the soul or entelechy is an autonomous and spontaneous unity, consisting of internal active force and producing its own perceptions (both confused and distinct). It is a self-moving machine, driven by its perceptions and appetites, but it follows (without being conscious of it) a lawful series or programme created by God. The entelechy remains the same despite undergoing an infinite number of changes. To Foucher he explained that each state of the individual substance is a consequence of its preceding one, "as if there were only God and the substance in the world". Despite this, the spiritual machine is related not only to God, but through the pre-established harmony to bodies or natural machines. However, as it is not dependent on them, the spiritual machine is superiot to natural machines. In this paper I concentrate on three moral themes related to the spiritual machine: moral deliberation, moral identity and the goal of moral action. All these themes are more or less implicit in the New System, but are discussed in more detail in the ensuing correspondence after the publication of the article. Finally, the themes are given an extended discussion especially in the second book of New Essays on Human Understanding (1704, published 1765), written partly at the same time as some of the correspondence and unpublished related material. I will first give an overview of Leibniz's position concerning the spiritual machine or automaton in the New System and then go on to explore the three moral themes one by one, as they feature in the subsequent correspondence and in the New Essays. I will show that the three topics are essential to the function of the spiritual machine – to Sophie Charlotte Leibniz wrote: "...let us say that everything in bodies happens mechanically, or in accordance with laws of motion, and that everything in the soul happens morally, or in accordance with perceived good or evil." There are many texts that anticipate the views in the New System (notably the unpublished memoir De Affectibus of 1679) and Leibniz continued to discuss its themes later on (in addition to correpondence, for example, in Theodicy of 1710 and Monadology and Principles of Nature and Grace of 1714). Here I focus on the short-term development of Leibniz's views on the topic (around 1695-1705), but I argue that his basic view of the spiritual machine did not change much after finishing the dialogue with Locke in 1704. (shrink)

Social machines are a prominent focus of attention for those who work in the field of Web and Internet science. Although a number of online systems have been described as social machines, there is, as yet, little consensus as to the precise meaning of the term “social machine.” This presents a problem for the scientific study of social machines, especially when it comes to the provision of a theoretical framework that directs, informs, and explicates the scientific and engineering activities (...) of the social machine community. The present paper outlines an approach to understanding social machines that draws on recent work in the philosophy of science, especially work in so-called mechanical philosophy. This is what might be called a mechanistic view of social machines. According to this view, social machines are systems whose phenomena are explained via an appeal to socio-technical mechanisms. We show how this account is able to accommodate a number of existing attempts to define the social machine concept, thereby yielding an important opportunity for theoretical integration. (shrink)

The recent emergence of machine-manipulated media raises an important societal question: How can we know whether a video that we watch is real or fake? In two online studies with 15,016 participants, we present authentic videos and deepfakes and ask participants to identify which is which. We compare the performance of ordinary human observers with the leading computer vision deepfake detection model and find them similarly accurate, while making different kinds of mistakes. Together, participants with access to the model’s (...) prediction are more accurate than either alone, but inaccurate model predictions often decrease participants’ accuracy. To probe the relative strengths and weaknesses of humans and machines as detectors of deepfakes, we examine human and machine performance across video-level features, and we evaluate the impact of preregistered randomized interventions on deepfake detection. We find that manipulations designed to disrupt visual processing of faces hinder human participants’ performance while mostly not affecting the model’s performance, suggesting a role for specialized cognitive capacities in explaining human deepfake detection performance. -/- . (shrink)

Some physicians, in their care of patients at risk of misusing opioids, use machine learning (ML)-based prediction drug monitoring programmes (PDMPs) to guide their decision making in the prescription of opioids. This can cause a conflict: a PDMP Score can indicate a patient is at a high risk of opioid abuse while a patient expressly reports oppositely. The prescriber is then left to balance the credibility and trust of the patient with the PDMP Score. Pozzi1 argues that a prescriber (...) who downgrades the credibility of a patient’s testimony based on a low PDMP Score is epistemically and morally unjustified and contributes to a form of testimonial injustice. This results in patients being silenced, excluded from decision-making processes and subjected to structural injustices. Additionally, the use of ML systems in medical practices raises concerns about perpetuating existing inequalities, overestimating their capabilities and displacing human authority. However, almost the very same critiques apply to human-based systems. Formalisation, ML systems included, should instead be viewed positively,2 and precisely as a powerful means to begin eroding these and other problems in ethically sensitive domains. In this case, the epistemic virtues of formalisation include promoting transparency, consistency and replicability in decision making. Rigorous ML systems can also help ensure that models …. (shrink)

Artificial Psychology (AP) is a highly multidisciplinary field of study in psychology. AP tries to solve problems which occur when psychologists do research and need a robust analysis method. Conventional statistical approaches have deep rooted limitations. These approaches are excellent on paper but often fail to model the real world. Mind researchers have been trying to overcome this by simplifying the models being studied. This stance has not received much practical attention recently. Promoting and improving artificial intelligence helps mind researchers (...) to find a holistic model of mental models. This development achieves this goal by using multiple perspectives and multiple data sets together with interactive, and realistic models. In this book, the methodology of approximate inference in psychological research from a theoretical and practical perspective has been considered. Quantitative variable-oriented methodology and qualitative case-oriented methods are both used to explain the set-oriented methodology and this book combines the precision of quantitative methods with information from qualitative methods. This is a book that many researchers can use to expand and deepen their psychological research and is a book which can be useful to postgraduate students. The reader does not need an in-depth knowledge of mathematics or statistics because statistical and mathematical intuitions are key here and they will be learned through practice. What is important is to understand and use the new application of the methods for finding new, dynamic and realistic interpretations. This book incorporates theoretical fuzzy inference and deep machine learning algorithms in practice. This is the kind of book that we wished we had had when we were students. This book covers at least some of the most important issues in mind research including uncertainty, fuzziness, continuity, complexity and high dimensionality which are inherent to mind data. These are elements of artificial psychology. This book implements models using R software. (shrink)

Le déficit d’explicabilité des techniques d’apprentissage machine (AM) pose des problèmes opérationnels, juridiques et éthiques. Un des principaux objectifs de notre projet est de fournir des explications éthiques des sorties générées par une application fondée sur de l’AM, considérée comme une boîte noire. La première étape de ce projet, présentée dans cet article, consiste à montrer que la validation de ces boîtes noires diffère épistémologiquement de celle mise en place dans le cadre d’une modélisation mathématique et causale d’un phénomène (...) physique. La différence majeure est qu’une méthode d’AM ne prétend pas représenter une causalité entre les paramètres d’entrées, qui peuvent être de plus de haute dimensionnalité, et ceux de sortie. Nous montrons dans cet article l’intérêt de mettre en œuvre les distinctions épistémologiques entre les différentes fonctions épistémiques d’un modèle, d’une part, et entre la fonction épistémique et l’usage d’un modèle, d’autre part. Enfin, la dernière partie de cet article présente nos travaux en cours sur l’évaluation d’une explication, qui peut être plus persuasive qu’informative, ce qui peut ainsi causer des problèmes d’ordre éthique. (shrink)

The use of counterfactuals for considerations of algorithmic fairness and explainability is gaining prominence within the machine learning community and industry. This paper argues for more caution with the use of counterfactuals when the facts to be considered are social categories such as race or gender. We review a broad body of papers from philosophy and social sciences on social ontology and the semantics of counterfactuals, and we conclude that the counterfactual approach in machine learning fairness and (...) social explainability can require an incoherent theory of what social categories are. Our findings suggest that most often the social categories may not admit counterfactual manipulation, and hence may not appropriately satisfy the demands for evaluating the truth or falsity of counterfactuals. This is important because the wide- spread use of counterfactuals in machine learning can lead to misleading results when applied in high-stakes domains. Accordingly, we argue that even though counterfactuals play an essential part in some causal inferences, their use for questions of algorithmic fairness and social explanations can create more problems than they resolve. Our positive result is a set of tenets about using counterfactuals for fairness and explanations in machine learning. (shrink)

Le déficit d’explicabilité des techniques d’apprentissage machine (AM) pose des problèmes opérationnels, juridiques et éthiques. Un des principaux objectifs de notre projet est de fournir des explications éthiques des sorties générées par une application fondée sur de l’AM, considérée comme une boîte noire. La première étape de ce projet, présentée dans cet article, consiste à montrer que la validation de ces boîtes noires diffère épistémologiquement de celle mise en place dans le cadre d’une modélisation mathéma- tique et causale d’un (...) phénomène physique. La différence majeure est qu’une méthode d’AM ne prétend pas représenter une causalité entre les paramètres d’entrées et ceux de sortie. Après avoir proposé une clarification et une adaptation des notions d’inter- prétabilité et d’explicabilité telles qu’on les rencontre dans la littérature déjà abondante sur le sujet, nous montrons dans cet article l’intérêt de mettre en œuvre les distinctions épistémologiques entre les différentes fonctions épistémiques d’un modèle, d’une part, et entre la fonction épistémique et l’usage d’un modèle, d’autre part. Enfin, la dernière partie de cet article présente nos travaux en cours sur l’évaluation d’une explication, qui peut être plus persuasive qu’informative, ce qui peut ainsi causer des problèmes d’ordre éthique. (shrink)

The success of deep learning in natural language processing raises intriguing questions about the nature of linguistic meaning and ways in which it can be processed by natural and artificial systems. One such question has to do with subword segmentation algorithms widely employed in language modeling, machine translation, and other tasks since 2016. These algorithms often cut words into semantically opaque pieces, such as ‘period’, ‘on’, ‘t’, and ‘ist’ in ‘period|on|t|ist’. The system then represents the resulting segments in a (...) dense vector space, which is expected to model grammatical relations among them. This representation may in turn be used to map ‘period|on|t|ist’ (English) to ‘par|od|ont|iste’ (French). Thus, instead of being modeled at the lexical level, translation is reformulated more generally as the task of learning the best bilingual mapping between the sequences of subword segments of two languages; and sometimes even between pure character sequences: ‘p|e|r|i|o|d|o|n|t|i|s|t’ → ‘p|a|r|o|d|o|n|t|i|s|t|e’. Such subword segmentations and alignments are at work in highly efficient end-to-end machine translation systems, despite their allegedly opaque nature. The computational value of such processes is unquestionable. But do they have any linguistic or philosophical plausibility? I attempt to cast light on this question by reviewing the relevant details of the subword segmentation algorithms and by relating them to important philosophical and linguistic debates, in the spirit of making artificial intelligence more transparent and explainable. (shrink)

Here I will re-purpose Nozick’s (1974) “Experience Machine” thought experiment against hedonism into an argument against Veritic Epistemic Consequentialism. According to VEC, the right action, epistemically speaking, is the one that results in at least as favorable a ratio of true to false belief as any other action available. A consequence of VEC is that it would be epistemically right to outsource all your cognitive endeavors to a matrix-like “True Belief Machine” that uploads true beliefs through artificial stimulation. (...) Rather than reflecting, inferring or inquiring, all of your beliefs would be the product of updates coming from the machine. This implication of VEC is counter intuitive. I will spend much of the paper considering defensive maneuvers and showing them to be lacking. The problem with going in for the True Belief Machine seems to be that the subject forfeits her autonomy. There has been much work lately on the virtue of intellectual autonomy and some of it has the potential to explain what is epistemically amiss about plugging into the machine. However, I will argue that the proponent of VEC cannot appeal to it without abandoning her preferred order of explanation. (shrink)

Drawing from Arjen Kleinherenbrink's recent book, Against Continuity: Gilles Deleuze's Speculative Realism (2019), this paper undertakes a detailed review of Kleinherenbrink's fourfold "externality thesis" vis-à-vis Deleuze's machine ontology. Reading Deleuze as a philosopher of the actual, this paper renders Deleuzean syntheses as passive contemplations, pulling other (passive) entities into an (active) experience and designating relations as expressed through contraction. In addition to reviewing Kleinherenbrink's book (which argues that the machine ontology is a guiding current that emerges in Deleuze's (...) work after Difference and Repetition) alongside much of Deleuze's oeuvre, we relate and juxtapose Deleuze's machine ontology to positions concerning externality held by a host of speculative realists. Arguing that the machine ontology has its own account of interaction, change, and novelty, we ultimately set to prove that positing an ontological "cut" on behalf of the virtual realm is unwarranted because, unlike the realm of actualities, it is extraneous to the structure of becoming-that is, because it cannot be homogenous, any theory of change vis-à-vis the virtual makes it impossible to explain how and why qualitatively different actualities are produced. (shrink)

Ecological realism and cognitivism are the two major current contenders in the field of cognitive perceptual theory. This thesis examines these theories, and the debate between them. It shows that the debate, as it exists in the literature, is inconclusive, primarily because of problems in the current formulations of the two contending theories. The most obvious difficulties in the two theories are removed, leaving reconstructed versions of both. The debate is then re-examined in the context of the reconstructed theories. It (...) is shown that ecological realism is a special case of cognitivism dealing with the detection of properties of objects in the environment by resonant transduction. It is also shown that neither theory, as it stands, can adequately describe changes in the perceptual abilities of even very simple animals. (shrink)

The increasing application of artificial intelligence (AI) to healthcare raises both hope and ethical concerns. Some advanced machine learning methods provide accurate clinical predictions at the expense of a significant lack of explainability. Alex John London has defended that accuracy is a more important value than explainability in AI medicine. In this article, we locate the trade-off between accurate performance and explainable algorithms in the context of distributive justice. We acknowledge that accuracy is cardinal from outcome-oriented justice (...) because it helps to maximize patients’ benefits and optimizes limited resources. However, we claim that the opaqueness of the algorithmic black box and its absence of explainability threatens core commitments of procedural fairness such as accountability, avoidance of bias, and transparency. To illustrate this, we discuss liver transplantation as a case of critical medical resources in which the lack of explainability in AI-based allocation algorithms is procedurally unfair. Finally, we provide a number of ethical recommendations for when considering the use of unexplainable algorithms in the distribution of health-related resources. (shrink)

This paper is devoted to some generalizations of explanatory potential of connectionist approaches to theoretical problems of the philosophy of mind. Are considered both strong, and weaknesses of neural network models. Connectionism has close methodological ties with modern neurosciences and neurophilosophy. And this fact strengthens its positions, in terms of empirical naturalistic approaches. However, at the same time this direction inherits weaknesses of computational approach, and in this case all system of anticomputational critical arguments becomes applicable to the connectionst models (...) of mind. The last developments in the field of deep learning gave rich empirical material for cognitive sciences. Multilayered networks, mathematical models of associative dynamics of learning, self-organizing neuronets and all that allow to explain the principles of human conceptual organizing and after this to emulate these processes in computer systems. At all engineering achievements of this technology there is a traditional criticism from representatives of cognitive psychology who cannot accept a thesis about learning ability of a neuronet on the basis of redistribution of scales. Process of learning of natural intelligence, according to cognitive models, happens due to attraction of knowledge broadcast in a symbolical form (mental representations, concepts) at the expense of the systems of output knowledge expressed in the propositional contents. Some philosophical aspects of «neural metaphor» in modern cognitive sciences create the problem field which demands comprehensive understanding, the first step towards which is taken in this work. (shrink)

As soon as you believe an imagination to be nonfictional, this imagination becomes your ontological theory of the reality. Your ontological theory (of the reality) can describe a system as the reality. However, actually this system is only a theory/conceptual-space/imagination/visual-imagery of yours, not the actual reality (i.e., the thing-in-itself). An ontological theory (of the reality) actually only describes your (subjective/mental) imagination/visual-imagery/conceptual-space. An ontological theory of the reality, is being described as a situation model (SM). There is no way to prove/disprove (...) that there is only one reality, or there are two realities (i.e., “subjective reality” and “objective reality”). So, every ontology talk/theory/imagination about the two realities is only a talk/theory/imagination – we will never know whether it is true or not. The conventionally-called “physical/objective reality/world” around my conventionally-called “physical/objective body” is actually a geometric mathematical model (being generated/mathematically-modeled by my brain) – it's actually a subset/component/part/element of my brain’s mind/consciousness/manifest-image. Our cosmos is an autonomous objective parallel computing automaton (aka state machine) which evolves by itself automatically/unintentionally – wave-particle duality and Heisenberg’s uncertainty principle can be explained under this SM of my brain. Each elementary particle (as a building block of our cosmos) is an autonomous mathematical entity itself (i.e., a thing in itself). Our cosmos has the same nature as a Game of Life system – both are autonomous objective parallel-computing automata. Cosmos (as a state machine) is indistinguishable from a digital simulation – my consciousness (as something nonphysical) is not cosmos (as a state machine). If we are happy to accept randomness/stochasticity, then it is obviously possible that all other worlds in the many-worlds interpretation (MWI) actually do not exist (objectively). As one metaphysical option, we can treat all other worlds as subjective only (even if they are actually objective). Under the context of this metaphysical option, we are only living in one world (i.e., the world we are currently living in; this world) – we are not living in many worlds at the same time parallelly. Under the context of this metaphysical option, there is only one possible future. The relationship among any number of elementary particles, is governed/described by Schrodinger equation. If (in theory) Schrodinger equation can’t be used to reliably forecast whether I will go to McDonald for dinner in this world (based on the current state of all elementary particles of the cosmos), then what can Schrodinger equation do? The conventionally-called “space” does not exist objectively. “Time” and “matter” are not physical. Consciousness is the subjective-form (aka quale) of the mathematical models (of the objective cosmos) which are intracorporeally/subjectively used by the control logic of a Turing machine’s program fatedly. A Turing machine’s mind/consciousness/manifest-image or deliberate decisions/choices should not be able to actually/objectively change/control/drive the (autonomous or fated) worldline of any elementary particle within this world (i.e., the world we are currently living in, under the context of MWI). Besides the Schrodinger equation (or another mathematical equation/function which is yet to be discovered) which is a valid/correct/factual causality of our cosmos/state-machine, every other causality (of our cosmos/state-machine) is either invalid/incorrect/counterfactual or can be proved by deductive inference based on the Schrodinger equation (or the aforementioned yet-to-be-discovered mathematical equation/function) only. Closed causality entails no causality. Consciousness plays no causal role (“epiphenomenalism”), or in other words, any cognitive/behavioural activity can in principle be carried out without consciousness (“conscious inessentialism”). If the “loop quantum gravity” theory is correct, then time/space does not actually/objectively exist in the objective-evolution of the objective cosmos, or in other words, we should not use the subjective/mental concept of “time”, “state” or “space” to describe/imagine the objective-evolution of our cosmos. (shrink)

This is our journal for developing Deduction Theory and studying Deep Learning and Artificial intelligence. Deduction Theory is a Theory of Deducing World’s Relativity by Information Coupling and Asymmetry. We focus on information processing, see intelligence as an information structure that relatively close object-oriented, probability-oriented, unsupervised learning, relativity information processing and massive automated information processing. We see deep learning and machine learning as an attempt to make all types of information processing relatively close to probability information processing. We will (...) discuss about how to understand Deep Learning and Artificial intelligence and why Deep Learning is shown better performance than the other methods by metaphysical logic. (shrink)

Self-consciousness relates to important themes, such as sentience and personhood, and is often the cornerstone of moral theories (Warren, 1997). However, not much attention is given to future moral creatures of the earth: robots. This may be due to the unsettled status of their experience, which is why I have chosen to find the necessary features of self-consciousness in them. Philosophy of mind is also my interest which I have developed since I rejected the idea of souls and could not (...) account for my own phenomenal experiences. Among the current answers to the hard problem of consciousness (Chalmers, 1995) one stood out to me – Daniel Dennett’s multiple drafts model of consciousness presented in “Consciousness Explained” (Dennett, 1991) and his other works. Daniel Dennett is a self-described illusionist (Dennett, 2016) and a functionalist. His position is synonymous to holding self-conscious experiences, such as deliberating over one’s pain or reflecting over one’s actions insufficient to postulate existence of a metaphysical self. He believes instead that experiences are functional realizations which do not require any self, and the feeling of being conscious or self-conscious is actually a ”user-illusion” (Dennett, 1991), a term describing the unjustified belief of being in control of one’s actions. In his views, a selfhood as it is popularly conceived is contradictory. Therefore, he rejects realism about mental events, a position also called naïve realism (Siegel, 2011), on the basis that it is unverifiable and there are insufficient grounds to accept mental events (Dennett, 1988). He adopts a similar scepticism towards other theories of minds, hence nearing eliminative materialism in self-consciousness and related theme of pain (Dennett, 1978). His multiple drafts model of consciousness seeks to ground our beliefs about self-consciousness in the specific architecture of mind, which is particular to every species and kind of being. As such it is a particularly flexible theory that examines experiences of non-humans without falling into binary thinking – believing that a being has as rich inner life as humans do or none. This is why it is a great starting point for establishing necessary conditions for consciousness in robots. In this essay, I will often use words computer, programme, robot interchangeably for the purpose of clarity and take as the working definition “autonomous machines capable of sensing their environment, making decision and performing tasks similar to human actions”. I chose to explore self-consciousness instead of committing to the issue of consciousness, because consciousness can be defined in a weak sense, in which even a thermostat is “conscious” of the temperature, and consciousness in a strong sense, which invokes the internalism and externalism debate that is less connected to the nature of experiences than self-consciousness. My exploration pertains first and foremost to the philosophy of mind, of which I will examine the most relevant positions and decide to commit to functionalism (Block, 1980). Then I will advocate for explaining away of some of phenomena of self-consciousness and believing self-consciousness to be a sort of illusion. From there I will conclude that robots could fall for the same illusion of self-consciousness and therefore become self-conscious. I will argue, that self-consciousness can be established using a modified Turing Test, which would determine if a robot believed in a meaningful narrative about itself (Ricoeur, 1991) and hence be self-conscious. This will lead to answering the question of this essay: what are the features necessary for a self-conscious robot in the light of “Consciousness Explained” by Daniel Dennett? (shrink)

Radical democratic thinking is becoming intrigued by the material situatedness of its political agents and by the role of nonhuman participants in political interaction. At stake here is the displacement of narrow anthropocentrism that currently guides democratic theory and practice, and its repositioning into what we call ‘the nonhuman condition’. This Critical Exchange explores the nonhuman condition. It asks: What are the implications of decentering the human subject via a new materialist reading of radical democracy? Does this reading dilute political (...) agency? Or should this be seen, on the contrary, as an invitation for new voices and demands to enter into democratic assemblages? How might engagement with the more-than-human disrupt or extend theories of radical democracy? Hans Asenbaum and Amanda Machin discuss the human democratic subject both through radical democratic and new materialist lenses. They suggest that a new materialist decentred subject does not lose agency but further gains political responsibility in radical democracy. Jean-Paul Gagnon highlights the possibility of becoming through loss. In losing our anthropocentric arrogance—our understanding of being other and better than animal—we become more connected and discover our place in human-nonhuman assemblages. Melissa Orlie agrees that something can be gained through renouncing dominant human fantasies and presents a ‘radical democratic naturalism’. By renouncing land violence and acknowledging nature’s subjectivity, we can further nonexploitative radical democratic politics. This is where the notion of ‘tidalectic’ processes introduced by James L. Smith is helpful. Focusing on water and the Indigenous knowledges around rivers, he considers how democracy and community can be reimagined as plural, cyclical and attentive to the nonhuman with which the human is inevitably entangled. Diana Leong offers a different approach. Considering the nonhuman from the perspective of Black Studies demands an awareness to how blackness has long been constituted as non- or partially human. (shrink)

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

In a recent article, Erasmus, Brunet, and Fisher (2021) argue that Artificial Neural Networks (ANNs) are explainable. They survey four influential accounts of explanation: the Deductive-Nomological model, the Inductive-Statistical model, the Causal-Mechanical model, and the New-Mechanist model. They argue that, on each of these accounts, the features that make something an explanation is invariant with regard to the complexity of the explanans and the explanandum. Therefore, they conclude, the complexity of ANNs (and other Machine Learning models) does not make (...) them less explainable. In this reply, it is argued that Erasmus et al. left out one influential account of explanation from their discussion: the Unificationist model. It is argued that, on the Unificationist model, the features that makes something an explanation is sensitive to complexity. Therefore, on the Unificationist model, ANNs (and other Machine Learning models) are not explainable. It is emphasized that Erasmus et al.’s general strategy is correct. The literature on explainable Artificial Intelligence can benefit by drawing from philosophical accounts of explanation. However, philosophical accounts of explanation do not settle the problem of whether ANNs are explainable because they do not unanimously declare that explanation is invariant with regard to complexity. (shrink)

Machine intelligence already helps medical staff with a number of tasks. Ethical decision-making, however, has not been handed over to computers. In this proof-of-concept study, we show how an algorithm based on Beauchamp and Childress’ prima-facie principles could be employed to advise on a range of moral dilemma situations that occur in medical institutions. We explain why we chose fuzzy cognitive maps to set up the advisory system and how we utilized machine learning to train it. We report (...) on the difficult task of operationalizing the principles of beneficence, non-maleficence and patient autonomy, and describe how we selected suitable input parameters that we extracted from a training dataset of clinical cases. The first performance results are promising, but an algorithmic approach to ethics also comes with several weaknesses and limitations. Should one really entrust the sensitive domain of clinical ethics to machine intelligence? (shrink)

Machines seem to be getting smarter and smarter and much better at human jobs, yet true AI is utterly implausible. This article explains the reasons why this is the case.

Despite Kant's belief in a universal causal determinism among phenomena and his rejection of any noumenal agency in brutes, he nevertheless rejected Descartes's hypothesis that brutes are machines. Explaining Kant's response to Descartes forms the basis for this discussion of the nature of consciousness and matter in Kant's system. Kant's numerous remarks on animal psychology-as found in his lecture notes and reflections on metaphysics and anthropology-suggest a theory of consciousness and self-consciousness at odds with that traditionally ascribed to him.

The opacity of some recent Machine Learning (ML) techniques have raised fundamental questions on their explainability, and created a whole domain dedicated to Explainable Artificial Intelligence (XAI). However, most of the literature has been dedicated to explainability as a scientific problem dealt with typical methods of computer science, from statistics to UX. In this paper, we focus on explainability as a pedagogical problem emerging from the interaction between lay users and complex technological systems. We defend an (...) empirical methodology based on field work, which should go beyond the in-vitro analysis of UX to examine in-vivo problems emerging in the field. Our methodology is also comparative, as it chooses to steer away from the almost exclusive focus on ML to compare its challenges with those faced by more vintage algorithms. Finally, it is also philosophical, as we defend the relevance of the philosophical literature to define the epistemic desiderata of a good explanation. This study was conducted in collaboration with Etalab, a Task Force of the French Prime Minister in charge of Open Data & Open Government Policies, dealing in particular with the enforcement of the right to an explanation. In order to illustrate and refine our methodology before going up to scale, we conduct a preliminary work of case studies on the main different types of algorithms used by the French administration: computation, matching algorithms and ML. We study the merits and drawbacks of a recent approach to explanation, which we baptize input-output black box reasoning or BBR for short. We begin by presenting a conceptual framework including the distinctions necessary to a study of pedagogical explainability. We proceed to algorithmic case studies, and draw model-specific and model-agnostic lessons and conjectures. (shrink)

Definition of the problem The umbrella term “explicability” refers to the reduction of opacity of artificial intelligence (AI) systems. These efforts are challenging for medical AI applications because higher accuracy often comes at the cost of increased opacity. This entails ethical tensions because physicians and patients desire to trace how results are produced without compromising the performance of AI systems. The centrality of explicability within the informed consent process for medical AI systems compels an ethical reflection on the trade-offs. Which (...) levels of explicability are needed to obtain informed consent when utilizing medical AI? Arguments We proceed in five steps: First, we map the terms commonly associated with explicability as described in the ethics and computer science literature, i.e., disclosure, intelligibility, interpretability, and explainability. Second, we conduct a conceptual analysis of the ethical requirements for explicability when it comes to informed consent. Third, we distinguish hurdles for explicability in terms of epistemic and explanatory opacity. Fourth, this then allows to conclude the level of explicability physicians must reach and what patients can expect. In a final step, we show how the identified levels of explicability can technically be met from the perspective of computer science. Throughout our work, we take diagnostic AI systems in radiology as an example. Conclusion We determined four levels of explicability that need to be distinguished for ethically defensible informed consent processes and showed how developers of medical AI can technically meet these requirements. (shrink)

Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence (XAI), a fast-growing research area that is so far lacking in firm theoretical foundations. Building on work in logic, probability, and causality, we establish the central role of necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We provide a sound and complete algorithm for computing explanatory factors (...) with respect to a given context, and demonstrate its flexibility and competitive performance against state of the art alternatives on various tasks. (shrink)

As the demand for explainable deep learning grows in the evaluation of language technologies, the value of a principled grounding for those explanations grows as well. Here we study the state-of-the-art in explanation for neural models for natural-language processing (NLP) tasks from the viewpoint of philosophy of science. We focus on recent evaluation work that finds brittleness in explanations obtained through attention mechanisms.We harness philosophical accounts of explanation to suggest broader conclusions from these studies. From this analysis, we assert the (...) impossibility of causal explanations from attention layers over text data. We then introduce NLP researchers to contemporary philosophy of science theories that allow robust yet non-causal reasoning in explanation, giving computer scientists a vocabulary for future research. (shrink)