Recent work in explainable artificial intelligence (XAI) attempts to render opaque AI systems understandable through a divide-and-conquer strategy. However, this fails to illuminate how trained AI systems work as a whole. Precisely this kind of functional understanding is needed, though, to satisfy important societal desiderata such as safety. To remedy this situation, we argue, AI researchers should seek mechanistic interpretability, viz. apply coordinated discovery strategies familiar from the life sciences to uncover the functional organisation of complex AI systems. Additionally, theorists (...) should accommodate for the unique costs and benefits of such strategies in their portrayals of XAI research. (shrink)

This paper addresses the black-box problem in artificial intelligence (AI), and the related problem of explainability of AI in the legal context. We argue, first, that the black box problem is, in fact, a superficial one as it results from an overlap of four different – albeit interconnected – issues: the opacity problem, the strangeness problem, the unpredictability problem, and the justification problem. Thus, we propose a framework for discussing both the black box problem and the explainability of AI. We (...) argue further that contrary to often defended claims the opacity issue is not a genuine problem. We also dismiss the justification problem. Further, we describe the tensions involved in the strangeness and unpredictability problems and suggest some ways to alleviate them. (shrink)

In this paper, we contribute to research on enterprise artificial intelligence (AI), specifically to organizations improving the customer experiences and their internal processes through using the type of AI called machine learning (ML). Many organizations are struggling to get enough value from their AI efforts, and part of this is related to the area of explainability. The need for explainability is especially high in what is called black-box ML models, where decisions are made without anyone understanding how an AI reached (...) a particular decision. This opaqueness creates a user need for explanations. Therefore, researchers and designers create different versions of so-called eXplainable AI (XAI). However, the demands for XAI can reduce the accuracy of the predictions the AI makes, which can reduce the perceived usefulness of the AI solution, which, in turn, reduces the interest in designing the organizational task structure to benefit from the AI solution. Therefore, it is important to ensure that the need for XAI is as low as possible. In this paper, we demonstrate how to achieve this by optimizing the task structure according to sociotechnical systems design principles. Our theoretical contribution is to the underexplored field of the intersection of AI design and organizational design. We find that explainability goals can be divided into two groups, pattern goals and experience goals, and that this division is helpful when defining the design process and the task structure that the AI solution will be used in. Our practical contribution is for AI designers who include organizational designers in their teams, and for organizational designers who answer that challenge. (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)

Artificial intelligence (AI) systems have demonstrated impressive performance across a variety of clinical tasks. However, notoriously, sometimes these systems are 'black boxes'. The initial response in the literature was a demand for 'explainable AI'. However, recently, several authors have suggested that making AI more explainable or 'interpretable' is likely to be at the cost of the accuracy of these systems and that prioritising interpretability in medical AI may constitute a 'lethal prejudice'. In this paper, we defend the value of interpretability (...) in the context of the use of AI in medicine. Clinicians may prefer interpretable systems over more accurate black boxes, which in turn is sufficient to give designers of AI reason to prefer more interpretable systems in order to ensure that AI is adopted and its benefits realised. Moreover, clinicians may be justified in this preference. Achieving the downstream benefits from AI is critically dependent on how the outputs of these systems are interpreted by physicians and patients. A preference for the use of highly accurate black box AI systems, over less accurate but more interpretable systems , may itself constitute a form of lethal prejudice that may diminish the benefits of AI to - and perhaps even harm - patients. (shrink)

Advancements in machine learning have fuelled the popularity of using AI decision algorithms in procedures such as bail hearings, medical diagnoses and recruitment. Academic articles, policy texts, and popularizing books alike warn that such algorithms tend to be opaque: they do not provide explanations for their outcomes. Building on a causal account of transparency and opacity as well as recent work on the value of causal explanation, I formulate a moral concern for opaque algorithms that is yet to receive a (...) systematic treatment in the literature: when such algorithms are used in life-changing decisions, they can obstruct us from effectively shaping our lives according to our goals and preferences, thus undermining our autonomy. I argue that this concern deserves closer attention as it furnishes the call for transparency in algorithmic decision-making with both new tools and new challenges. (shrink)

Explainable artificial intelligence (XAI) aims to help people understand black box algorithms, particularly of their outputs. But what are these explanations and when is one explanation better than another? The manipulationist definition of explanation from the philosophy of science offers good answers to these questions, holding that an explanation consists of a generalization that shows what happens in counterfactual cases. Furthermore, when it comes to explanatory depth this account holds that a generalization that has more abstract variables, is broader in (...) scope and/or more accurate is better. By applying these definitions and contrasting them with alternative definitions in the XAI literature I hope to help clarify what a good explanation is for AI. (shrink)

Explainable artificial intelligence (XAI) is an emerging, multidisciplinary field of research that seeks to develop methods and tools for making AI systems more explainable or interpretable. XAI researchers increasingly recognise explainability as a context-, audience- and purpose-sensitive phenomenon, rather than a single well-defined property that can be directly measured and optimised. However, since there is currently no overarching definition of explainability, this poses a risk of miscommunication between the many different researchers within this multidisciplinary space. This is the problem we (...) seek to address in this paper. We outline a framework, called Explanatory Pragmatism, which we argue has two attractive features. First, it allows us to conceptualise explainability in explicitly context-, audience- and purpose-relative terms, while retaining a unified underlying definition of explainability. Second, it makes visible any normative disagreements that may underpin conflicting claims about explainability regarding the purposes for which explanations are sought. Third, it allows us to distinguish several dimensions of AI explainability. We illustrate this framework by applying it to a case study involving a machine learning model for predicting whether patients suffering disorders of consciousness were likely to recover consciousness. (shrink)

Different people have different perceptions about artificial intelligence (AI). It is extremely important to bring together all the alternative frames of thinking—from the various communities of developers, researchers, business leaders, policymakers, and citizens—to properly start acknowledging AI. This article highlights the ‘fruitful collaboration’ that sociology and AI could develop in both social and technical terms. We discuss how biases and unfairness are among the major challenges to be addressed in such a sociotechnical perspective. First, as intelligent machines reveal their nature (...) of ‘magnifying glasses’ in the automation of existing inequalities, we show how the AI technical community is calling for transparency and explainability, accountability and contestability. Not to be considered as panaceas, they all contribute to ensuring human control in novel practices that include requirement, design and development methodologies for a fairer AI. Second, we elaborate on the mounting attention for technological narratives as technology is recognized as a social practice within a specific institutional context. Not only do narratives reflect organizing visions for society, but they also are a tangible sign of the traditional lines of social, economic, and political inequalities. We conclude with a call for a diverse approach within the AI community and a richer knowledge about narratives as they help in better addressing future technical developments, public debate, and policy. AI practice is interdisciplinary by nature and it will benefit from a socio-technical perspective. (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)

With automation of routine decisions coupled with more intricate and complex information architecture operating this automation, concerns are increasing about the trustworthiness of these systems. These concerns are exacerbated by a class of artificial intelligence that uses deep learning, an algorithmic system of deep neural networks, which on the whole remain opaque or hidden from human comprehension. This situation is commonly referred to as the black box problem in AI. Without understanding how AI reaches its conclusions, it is an open (...) question to what extent we can trust these systems. The question of trust becomes more urgent as we delegate more and more decision-making to and increasingly rely on AI to safeguard significant human goods, such as security, healthcare, and safety. Models that “open the black box” by making the non-linear and complex decision process understandable by human observers are promising solutions to the black box problem in AI but are limited, at least in their current state, in their ability to make these processes less opaque to most observers. A philosophical analysis of trust will show why transparency is a necessary condition for trust and eventually for judging AI to be trustworthy. A more fruitful route for establishing trust in AI is to acknowledge that AI is situated within a socio-technical system that mediates trust, and by increasing the trustworthiness of these systems, we thereby increase trust in AI. (shrink)

Should decision-making algorithms be held to higher standards of transparency than human beings? The way we answer this question directly impacts what we demand from explainable algorithms, how we govern them via regulatory proposals, and how explainable algorithms may help resolve the social problems associated with decision making supported by artificial intelligence. Some argue that algorithms and humans should be held to the same standards of transparency and that a double standard of transparency is hardly justified. We give two arguments (...) to the contrary and specify two kinds of situations for which higher standards of transparency are required from algorithmic decisions as compared to humans. Our arguments have direct implications on the demands from explainable algorithms in decision-making contexts such as automated transportation. (shrink)

There is general consensus that explainable artificial intelligence is valuable, but there is significant divergence when we try to articulate why, exactly, it is desirable. This question must be distinguished from two other kinds of questions asked in the XAI literature that are sometimes asked and addressed simultaneously. The first and most obvious is the ‘how’ question—some version of: ‘how do we develop technical strategies to achieve XAI?’ Another question is specifying what kind of explanation is worth having in the (...) first place. As difficult and important as the challenges are in answering these questions, they are distinct from a third question: why do we want XAI at all? There is vast literature on this question as well, but I wish to explore a different kind of answer. The most obvious way to answer this question is by describing a desirable outcome that would likely be achieved with the right kind of explanation, which would make the explanation valuable instrumentally. That is, XAI is desirable to attain some other value, such as fairness, trust, accountability, or governance. This family of arguments is obviously important, but I argue that explanations are also intrinsically valuable, because unexplainable systems can be dehumanizing. I argue that there are at least three independently valid versions of this kind of argument: an argument from participation, from knowledge, and from actualization. Each of these arguments that XAI is intrinsically valuable is independently compelling, in addition to the more obvious instrumental benefits of XAI. (shrink)

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

Previous research in Explainable Artificial Intelligence (XAI) suggests that a main aim of explainability approaches is to satisfy specific interests, goals, expectations, needs, and demands regarding artificial systems (we call these “stakeholders' desiderata”) in a variety of contexts. However, the literature on XAI is vast, spreads out across multiple largely disconnected disciplines, and it often remains unclear how explainability approaches are supposed to achieve the goal of satisfying stakeholders' desiderata. This paper discusses the main classes of stakeholders calling for explainability (...) of artificial systems and reviews their desiderata. We provide a model that explicitly spells out the main concepts and relations necessary to consider and investigate when evaluating, adjusting, choosing, and developing explainability approaches that aim to satisfy stakeholders' desiderata. This model can serve researchers from the variety of different disciplines involved in XAI as a common ground. It emphasizes where there is interdisciplinary potential in the evaluation and the development of explainability approaches. (shrink)

Despite the widespread use of automated decision-making (ADM) systems, they are often developed without involving the public or those directly affected, leading to concerns about systematic biases that may perpetuate structural injustices. Existing formal fairness approaches primarily focus on statistical outcomes across demographic groups or individual fairness, yet these methods reveal ambiguities and limitations in addressing fairness comprehensively. This paper argues for a holistic approach to algorithmic fairness that integrates procedural fairness, considering both decision-making processes and their outcomes. Procedural fairness (...) emphasizes the importance of fair decision-making procedures, which aligns with theories of relational justice that stress the quality of social relations and power dynamics. We highlight the need for substantive procedural fairness to ensure better outcomes and address forward-looking responsibilities. Additionally, we propose leveraging Public Engagement, a core dimension within the well-established Responsible Research and Innovation framework, to enhance procedural fairness in ADM systems. Our contribution underscores the value of Public Engagement in fostering fairer ADM processes, thereby expanding the current focus beyond technical outcome-based approaches to encompass broader procedural considerations. (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)

Artificial intelligence has historically been conceptualized in anthropomorphic terms. Some algorithms deploy biomimetic designs in a deliberate attempt to effect a sort of digital isomorphism of the human brain. Others leverage more general learning strategies that happen to coincide with popular theories of cognitive science and social epistemology. In this paper, I challenge the anthropomorphic credentials of the neural network algorithm, whose similarities to human cognition I argue are vastly overstated and narrowly construed. I submit that three alternative supervised learning (...) methods—namely lasso penalties, bagging, and boosting—offer subtler, more interesting analogies to human reasoning as both an individual and a social phenomenon. Despite the temptation to fall back on anthropomorphic tropes when discussing AI, however, I conclude that such rhetoric is at best misleading and at worst downright dangerous. The impulse to humanize algorithms is an obstacle to properly conceptualizing the ethical challenges posed by emerging technologies. (shrink)

Criminal liability for acts committed by AI systems has recently become a hot legal topic. This paper includes three different contributions. The first contribution is an analysis of the extent to which an AI system can satisfy the requirements for criminal liability: accomplishing an actus reus, having the corresponding mens rea, possessing the cognitive capacities needed for responsibility. The second contribution is a discussion of criminal activity accomplished by an AI entity, with reference to a recent case involving an online (...) bot, the Random Darknet Shopper. This discussion will provide the context for the analysis of commonalities and differences between criminal activities by humans and by artificial systems. The third contribution concerns the evaluation of different ways of addressing criminal activities by AI systems in a regulatory perspective. (shrink)

The literature addressing bias and fairness in AI models (fair-AI) is growing at a fast pace, making it difficult for novel researchers and practitioners to have a bird’s-eye view picture of the field. In particular, many policy initiatives, standards, and best practices in fair-AI have been proposed for setting principles, procedures, and knowledge bases to guide and operationalize the management of bias and fairness. The first objective of this paper is to concisely survey the state-of-the-art of fair-AI methods and resources, (...) and the main policies on bias in AI, with the aim of providing such a bird’s-eye guidance for both researchers and practitioners. The second objective of the paper is to contribute to the policy advice and best practices state-of-the-art by leveraging from the results of the NoBIAS research project. We present and discuss a few relevant topics organized around the NoBIAS architecture, which is made up of a Legal Layer, focusing on the European Union context, and a Bias Management Layer, focusing on understanding, mitigating, and accounting for bias. (shrink)

Unsupervised learning algorithms are widely used for many important statistical tasks with numerous applications in science and industry. Yet despite their prevalence, they have attracted remarkably little philosophical scrutiny to date. This stands in stark contrast to supervised and reinforcement learning algorithms, which have been widely studied and critically evaluated, often with an emphasis on ethical concerns. In this article, I analyze three canonical unsupervised learning problems: clustering, abstraction, and generative modeling. I argue that these methods raise unique epistemological and (...) ontological questions, providing data-driven tools for discovering natural kinds and distinguishing essence from contingency. This analysis goes some way toward filling the lacuna in contemporary philosophical discourse on unsupervised learning, as well as bringing conceptual unity to a heterogeneous field more often described by what it isnot(i.e., supervised or reinforcement learning) than by what itis. I submit that unsupervised learning is not just a legitimate subject of philosophical inquiry but perhaps the most fundamental branch of all AI. However, an uncritical overreliance on unsupervised methods poses major epistemic and ethical risks. I conclude by advocating for a pragmatic, error-statistical approach that embraces the opportunities and mitigates the challenges posed by this powerful class of algorithms. (shrink)

Artificial intelligence (AI) systems have demonstrated impressive performance across a variety of clinical tasks. However, notoriously, sometimes these systems are “black boxes.” The initial response in the literature was a demand for “explainable AI.” However, recently, several authors have suggested that making AI more explainable or “interpretable” is likely to be at the cost of the accuracy of these systems and that prioritizing interpretability in medical AI may constitute a “lethal prejudice.” In this paper, we defend the value of interpretability (...) in the context of the use of AI in medicine. Clinicians may prefer interpretable systems over more accurate black boxes, which in turn is sufficient to give designers of AI reason to prefer more interpretable systems in order to ensure that AI is adopted and its benefits realized. Moreover, clinicians may be justified in this preference. Achieving the downstream benefits from AI is critically dependent on how the outputs of these systems are interpreted by physicians and patients. A preference for the use of highly accurate black box AI systems, over less accurate but more interpretable systems, may itself constitute a form of lethal prejudice that may diminish the benefits of AI to—and perhaps even harm—patients. (shrink)

The debate about the ethical implications of Artificial Intelligence dates from the 1960s :741–742, 1960; Wiener in Cybernetics: or control and communication in the animal and the machine, MIT Press, New York, 1961). However, in recent years symbolic AI has been complemented and sometimes replaced by Neural Networks and Machine Learning techniques. This has vastly increased its potential utility and impact on society, with the consequence that the ethical debate has gone mainstream. Such a debate has primarily focused on principles—the (...) ‘what’ of AI ethics —rather than on practices, the ‘how.’ Awareness of the potential issues is increasing at a fast rate, but the AI community’s ability to take action to mitigate the associated risks is still at its infancy. Our intention in presenting this research is to contribute to closing the gap between principles and practices by constructing a typology that may help practically-minded developers apply ethics at each stage of the Machine Learning development pipeline, and to signal to researchers where further work is needed. The focus is exclusively on Machine Learning, but it is hoped that the results of this research may be easily applicable to other branches of AI. The article outlines the research method for creating this typology, the initial findings, and provides a summary of future research needs. (shrink)

This paper proposes a formal top-level model of explaining the outputs of machine-learning-based decision-making applications and evaluates it experimentally with three data sets. The model draws on AI & law research on argumentation with cases, which models how lawyers draw analogies to past cases and discuss their relevant similarities and differences in terms of relevant factors and dimensions in the problem domain. A case-based approach is natural since the input data of machine-learning applications can be seen as cases. While the (...) approach is motivated by legal decision making, it also applies to other kinds of decision making, such as commercial decisions about loan applications or employee hiring, as long as the outcome is binary and the input conforms to this paper’s factor- or dimension format. The model is top-level in that it can be extended with more refined accounts of similarities and differences between cases. It is shown to overcome several limitations of similar argumentation-based explanation models, which only have binary features and do not represent the tendency of features towards particular outcomes. The results of the experimental evaluation studies indicate that the model may be feasible in practice, but that further development and experimentation is needed to confirm its usefulness as an explanation model. Main challenges here are selecting from a large number of possible explanations, reducing the number of features in the explanations and adding more meaningful information to them. It also remains to be investigated how suitable our approach is for explaining non-linear models. (shrink)

In recent years, the design and production of information systems have seen significant growth. However, these information artefacts often exhibit characteristics that compromise their reliability. This issue appears to stem from the neglect or underestimation of certain crucial aspects in the application of Information Systems Design (ISD). For example, it is frequently difficult to prove when one of these products does not work properly or works incorrectly (falsifiability), their usage is often left to subjective experience and somewhat arbitrary choices (anecdotes), (...) and their functions are often obscure for users as well as designers (explainability). In this paper, we propose an approach that can be used to support the analysis and re-(design) of information systems grounded on a well-known theory of information, namely, teleosemantics. This approach emphasizes the importance of grounding the design and validation process on dependencies between four core components: the producer (or designer), the produced (or used) information system, the consumer (or user), and the design (or use) purpose. We analyze the ambiguities and problems of considering these components separately. We then present some possible ways in which they can be combined through the teleological approach. Also, we debate guidelines to prevent ISD from failing to address critical issues. Finally, we discuss perspectives on applications over real existing information technologies and some implications for explainable AI and ISD. (shrink)

Wang (_Philosophy & Technology_ 35, 2022) introduces a Foucauldian power account of algorithmic transparency. This short commentary explores when this power account is appropriate. It is first observed that the power account is a constructionist one, and that such accounts often come with both factual and evaluative claims. In an instance of Hume’s law, the evaluative claims do not follow from the factual claims, leaving open the question of how much constructionist commitment (Hacking, 1999) one should have. The concept of (...) acts in equilibrium (Nozick, 1981) is then used to explain how different individuals reading Wang can end up with different evaluative attitudes towards algorithmic transparency, despite factual agreement. The commentary concludes by situating constructionist commitment inside a larger question of how much to think of our actions, identifying conflicting arguments. (shrink)

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)

Artificial intelligence (AI) has long been recognised as a challenge to responsibility. Much of this discourse has been framed around robots, such as autonomous weapons or self-driving cars, where we arguably lack control over a machine’s behaviour and therefore struggle to identify an agent that can be held accountable. However, most of today’s AI is based on machine-learning technology that does not act on its own, but rather serves as a decision-support tool, automatically analysing data to help human agents make (...) better decisions. I argue that decision-support tools pose a challenge to responsibility that goes beyond the familiar problem of finding someone to blame or punish for the behaviour of agent-like systems. Namely, they pose a problem for what we might call “decision ownership”: they make it difficult to identify human agents to whom we can attribute value-judgements that are reflected in decisions. Drawing on recent philosophical literature on responsibility and its various facets, I argue that this is primarily a problem of attributability rather than of accountability. This particular responsibility problem comes in different forms and degrees, most obviously when an AI provides direct recommendations for actions, but also, less obviously, when it provides mere descriptive information on the basis of which a decision is made. (shrink)

Durán et al argue in ‘Who is afraid of black box algorithms? On the epistemological and ethical basis of trust in medical AI’1 that traditionally proposed solutions to make black box machine learning models in medicine less opaque and more transparent are, though necessary, ultimately not sufficient to establish their overall trustworthiness. This is because transparency procedures currently employed, such as the use of an interpretable predictor,2 cannot fully overcome the opacity of such models. Computational reliabilism, an alternate approach to (...) adjudicating trustworthiness that goes beyond transparency solutions, is argued to be a more promising approach. CR can bring the benefits of traditional process reliabilism in epistemology to bear on this problem of model trustworthiness. Durán et al ’s explicitly reliabilist epistemology to assess the trustworthiness of black box models is a timely addition to current transparency-focused approaches in the literature. Their delineation of the epistemic from the ethical also serves the debate by clarifying the nature of the different problems. However, their overall account underestimates the epistemic value of certain transparency-enabling approaches by conflating different types of opacity and also oversimplifies transparency-advocating arguments in the literature. First, it is unclear why Durán et al consider transparency approaches as insufficient to overcome epistemic opacity, if heiraccount of opacity is the traditional one from the machine learning literature: opacity stemming from the mismatch between mathematical optimisation in high dimensionality that is characteristic of machine learning and the demands of human-scale reasoning and styles of semantic interpretation.3 …. (shrink)

As artificial intelligence (AI) thrives and propagates through modern life, a key question to ask is how to include humans in future AI? Despite human involvement at every stage of the production process from conception and design through to implementation, modern AI is still often criticized for its “black box” characteristics. Sometimes, we do not know what really goes on inside or how and why certain conclusions are met. Future AI will face many dilemmas and ethical issues unforeseen by their (...) creators beyond those commonly discussed (e.g., trolley problems and variants of it) and to which solutions cannot be hard-coded and are often still up for debate. Given the sensitivity of such social and ethical dilemmas and the implications of these for human society at large, when and if our AI make the “wrong” choice we need to understand how they got there in order to make corrections and prevent recurrences. This is particularly true in situations where human livelihoods are at stake (e.g., health, well-being, finance, law) or when major individual or household decisions are taken. Doing so requires opening up the “black box” of AI; especially as they act, interact, and adapt in a human world and how they interact with other AI in this world. In this article, we argue for the application of cognitive architectures for ethical AI. In particular, for their potential contributions to AI transparency, explainability, and accountability. We need to understand how our AI get to the solutions they do, and we should seek to do this on a deeper level in terms of the machine-equivalents of motivations, attitudes, values, and so on. The path to future AI is long and winding but it could arrive faster than we think. In order to harness the positive potential outcomes of AI for humans and society (and avoid the negatives), we need to understand AI more fully in the first place and we expect this will simultaneously contribute towards greater understanding of their human counterparts also. (shrink)

Deep learning and other black-box models are becoming more and more popular today. Despite their high performance, they may not be accepted ethically or legally because of their lack of explainability. This paper presents the increasing number of legal requirements on machine learning model interpretability and explainability in the context of private and public decision making. It then explains how those legal requirements can be implemented into machine-learning models and concludes with a call for more inter-disciplinary research on explainability.

AI-driven journalism refers to various methods and tools for gathering, verifying, producing, and distributing news information. Their potential is to extend human capabilities and create new forms of augmented journalism. Although scholars agreed on the necessity to embed journalistic values in these systems to make AI systems accountable, less attention was paid to data quality, while the results’ accuracy and efficiency depend on high-quality data in any machine learning task. Assessing data quality in the context of AI-driven journalism requires a (...) broader and interdisciplinary approach, relying on the challenges of data quality in machine learning and the ethical challenges of using machine learning in journalism. To better identify these, we propose a data quality assessment framework to support the collection and pre-processing stages in machine learning. It relies on three of the core principles of ethical journalism—accuracy, fairness, and transparency—and participates in the shift from model-centric to data-centric AI, by focusing on data quality to reduce reliance on large datasets with errors, making data labelling consistent, and better integrating journalistic knowledge. (shrink)

As more decisions are made by automated algorithmic systems, the transparency of these systems has come under scrutiny. While such transparency is typically seen as beneficial, there is a also a critical, Foucauldian account of it. From this perspective, worries have recently been articulated that algorithmic transparency can be used for manipulation, as part of a disciplinary power structure. Klenk (Philosophy & Technology 36, 79, 2023) recently argued that such manipulation should not be understood as exploitation of vulnerable victims, but (...) rather as indifference to whether the information provided enhances decision-making by revealing reasons. This short commentary on Klenk uses Berlin’s (1958) two concepts of liberty to further illuminate the concept of transparency as manipulation, finding alignment between positive liberty and the critical account. (shrink)

Under the rubric of understanding the problem of explainability of AI in terms of abductive cognition, I propose to review the lessons from AlphaGo and her more powerful successors. As AI players in Baduk have arrived at superhuman level, there seems to be no hope for understanding the secret of their breathtakingly brilliant moves. Without making AI players explainable in some ways, both human and AI players would be less-than omniscient, if not ignorant, epistemic agents. Are we bound to have (...) less explainable AI Baduk players as they make further progress? I shall show that the resolution of this apparent paradox depends on how we understand the crucial distinction between abduction and inference to the best explanation. Some further philosophical issues arising from explainable AI will also be discussed in connection with this distinction. (shrink)

In the last decades, supervised machine learning has seen the widespread growth of highly complex, non-interpretable models, of which deep neural networks are the most typical representative. Due to their complexity, these models have showed an outstanding performance in a series of tasks, as in image recognition and machine translation. Recently, though, there has been an important discussion over whether those non-interpretable models are able to provide any sort of understanding whatsoever. For some scholars, only interpretable models can provide understanding. (...) More popular, however, is the idea that understanding can come from a careful analysis of the dataset or from the model’s theoretical basis. In this paper, I wish to examine the possible forms of obtaining understanding of such non-interpretable models. Two main strategies for providing understanding are analyzed. The first involves understanding without interpretability, either through external evidence for the model’s inner functioning or through analyzing the data. The second is based on the artificial production of interpretable structures, through three main forms: post hoc models, hybrid models, and quasi-interpretable structures. Finally, I consider some of the conceptual difficulties in the attempt to create explanations for these models, and their implications for understanding. (shrink)

Recent years have seen rapid advancements in selection assessments, shifting away from human and toward algorithmic judgments of candidates. Indeed, algorithmic recruitment tools have been created to screen candidates’ resumes, assess psychometric characteristics through game-based assessments, and judge asynchronous video interviews, among other applications. While research into candidate reactions to these technologies is still in its infancy, early research in this regard has explored user experiences and fairness perceptions. In this article, we review applicants’ perceptions of the procedural fairness of (...) algorithmic recruitment tools based on key findings from seven key studies, sampling over 1,300 participants between them. We focus on the sub-facets of behavioral control, the extent to which individuals feel their behavior can influence an outcome, and social presence, whether there is the perceived opportunity for a social connection and empathy. While perceptions of overall procedural fairness are mixed, we find that fairness perceptions concerning behavioral control and social presence are mostly negative. Participants feel less confident that they are able to influence the outcome of algorithmic assessments compared to human assessments because they are more objective and less susceptible to manipulation. Participants also feel that the human element is lost when these tools are used since there is a lack of perceived empathy and interpersonal warmth. Since this field of research is relatively under-explored, we end by proposing a research agenda, recommending that future studies could examine the role of individual differences, demographics, and neurodiversity in influencing fairness perceptions of algorithmic recruitment. (shrink)

Recent work has demonstrated how data-driven AI methods can leverage consumer protection by supporting the automated analysis of legal documents. However, a shortcoming of data-driven approaches is poor explainability. We posit that in this domain useful explanations of classifier outcomes can be provided by resorting to legal rationales. We thus consider several configurations of memory-augmented neural networks where rationales are given a special role in the modeling of context knowledge. Our results show that rationales not only contribute to improve the (...) classification accuracy, but are also able to offer meaningful, natural language explanations of otherwise opaque classifier outcomes. (shrink)