Just as humans can draw conclusions responsibly or irresponsibly, so too can computers. Machine learning systems that have been trained on data sets that include irresponsible judgments are likely to yield irresponsible predictions as outputs. In this paper I focus on a particular kind of inference a computer system might make: identification of the intentions with which a person acted on the basis of photographic evidence. Such inferences are liable to be morally objectionable, because of a way in which (...) they are presumptuous. After elaborating this moral concern, I explore the possibility that carefully procuring the training data for image recognition systems could ensure that the systems avoid the problem. The lesson of this paper extends beyond just the particular case of image recognition systems and the challenge of responsibly identifying a person’s intentions. Reflection on this particular case demonstrates the importance (as well as the difficulty) of evaluating machine learning systems and their training data from the standpoint of moral considerations that are not encompassed by ordinary assessments of predictive accuracy. (shrink)

This article examines the interplay between contemporary algorithmic security technology and the political theory of the state of exception. I argue that the exception, as both a political and a technological concept, provides a crucial way to understand the power operating through machine learning technologies used in the security apparatuses of the modern state. I highlight how algorithmic security technology, through its inherent technical properties, carries exceptions throughout its political and technological architecture. This leads me to engage with Theodor (...) Adorno’s negative dialectics to interrogate the question of ‘ground truth’ in machine learning. I conclude that most machine learning technology asserts identity between itself and bourgeois reality – and thus inherently reinforces and reproduces the relations of domination entailed in that image of the world. However, space still exists for machine learning to operate within spaces of political non-identity, or exceptions to the bourgeois totality, and aid in liberatory politics. (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 past few years, machine learning (ML) tools have been implemented with success in the medical context. However, several practitioners have raised concerns about the lack of transparency—at the algorithmic level—of many of these tools; and solutions from the field of explainable AI (XAI) have been seen as a way to open the ‘black box’ and make the tools more trustworthy. Recently, Alex London has argued that in the medical context we do not need machine learning tools (...) to be interpretable at the algorithmic level to make them trustworthy, as long as they meet some strict empirical desiderata. In this paper, we analyse and develop London’s position. In particular, we make two claims. First, we claim that London’s solution to the problem of trust can potentially address another problem, which is how to evaluate the reliability of ML tools in medicine for regulatory purposes. Second, we claim that to deal with this problem, we need to develop London’s views by shifting the focus from the opacity of algorithmic details to the opacity of the way in which ML tools are trained and built. We claim that to regulate AI tools and evaluate their reliability, agencies need an explanation of how ML tools have been built, which requires documenting and justifying the technical choices that practitioners have made in designing such tools. This is because different algorithmic designs may lead to different outcomes, and to the realization of different purposes. However, given that technical choices underlying algorithmic design are shaped by value-laden considerations, opening the black box of the design process means also making transparent and motivating (technical and ethical) values and preferences behind such choices. Using tools from philosophy of technology and philosophy of science, we elaborate a framework showing how an explanation of the training processes of ML tools in medicine should look like. (shrink)

This article asks the question, ``what is reliable machine learning?'' As I intend to answer it, this is a question about epistemic justification. Reliable machine learning gives justification for believing its output. Current approaches to reliability (e.g., transparency) involve showing the inner workings of an algorithm (functions, variables, etc.) and how they render outputs. We then have justification for believing the output because we know how it was computed. Thus, justification is contingent on what can be shown about (...) the algorithm, its properties, and its behavior. In this paper, I defend computational reliabilism (CR). CR is a computationally-inspired off-shoot of process reliabilism that does not require showing the inner workings of an algorithm. CR credits reliability to machine learning by identifying reliability indicators external to the algorithm (validation methods, knowledge-based integration, etc.). Thus, we have justification for believing the output of machine learning when we have identified the appropriate reliability indicators. CR is advanced as a more suitable epistemology for machine learning. The main goal of this article is to lay the groundwork for CR, how it works, and what we can expect as a justificatory framework for reliable machine learning. (shrink)

Recent controversies about the level of replicability of behavioral research analyzed using statistical inference have cast interest in developing more efficient techniques for analyzing the results of psychological experiments. Here we claim that complementing the analytical workflow of psychological experiments with Machine Learning-based analysis will both maximize accuracy and minimize replicability issues. As compared to statistical inference, ML analysis of experimental data is model agnostic and primarily focused on prediction rather than inference. We also highlight some potential pitfalls resulting (...) from adoption of Machine Learning based experiment analysis. If not properly used it can lead to over-optimistic accuracy estimates similarly observed using statistical inference. Remedies to such pitfalls are also presented such and building model based on cross validation and the use of ensemble models. ML models are typically regarded as black boxes and we will discuss strategies aimed at rendering more transparent the predictions. (shrink)

Machine learning and essentialism have been connected in the past by various researchers, in order to state that the main paradigm in machine learning processes is equivalent to choosing the “essential” attributes for the machine to search for. Our goal in this paper is to show that there are connections between machine learning and essentialism, but only for some kinds of machine learning, and often not including deep learning methods. Similarity-based approaches, more connected to the (...) overall prototype theory, spanning from psychology and linguistics, seem more suited for pattern recognition and complex deep-learning issues, while for classification problems, mostly for unsupervised learning, essentialism seems like the best choice. In order to illustrate the difference better, we will connect both paths to their sources in other disciplines and see how human psychology influences our decision in machine-learning modeling as well. This leads to a philosophically very interesting consequence: even in the setting of supervised machine learning, essences are not present in data, but in targets, which in turn means that the categories which purport to be essences are in fact human-made, and hand-coded in the targets. The success of machine learning, therefore, does not give any substantial evidence for the independent existence of essential properties. Our stance here is to state that neither the existence nor the lack of “essential” properties in machine learning can lead to metaphysical, i.e., ontological claims. (shrink)

Current methods of operationalizing concepts of misinformation in machine learning are often problematic given idiosyncrasies in their success conditions compared to other models employed in the natural and social sciences. The intrinsic value-ladenness of misinformation and the dynamic relationship between citizens' and social scientists' concepts of misinformation jointly suggest that both the construct legitimacy and the construct validity of these models needs to be assessed via more democratic criteria than has previously been recognized.

Prediction-based decisions, which are often made by utilizing the tools of machine learning, influence nearly all facets of modern life. Ethical concerns about this widespread practice have given rise to the field of fair machine learning and a number of fairness measures, mathematically precise definitions of fairness that purport to determine whether a given prediction-based decision system is fair. Following Reuben Binns (2017), we take ‘fairness’ in this context to be a placeholder for a variety of normative egalitarian (...) considerations. We explore a few fairness measures to suss out their egalitarian roots and evaluate them, both as formalizations of egalitarian ideas and as assertions of what fairness demands of predictive systems. We pay special attention to a recent and popular fairness measure, counterfactual fairness, which holds that a prediction about an individual is fair if it is the same in the actual world and any counterfactual world where the individual belongs to a different demographic group (cf. Kusner et al. (2018)). (shrink)

Judgments concerning animals have arisen across a variety of established practice areas. There is, however, no publicly available repository of judgments concerning the emerging practice area of animal protection law. This has hindered the identification of individual animal protection law judgments and comprehension of the scale of animal protection law made by courts. Thus, we detail the creation of an initial animal protection law repository using natural language processing and machine learning techniques. This involved domain expert classification of 500 (...) judgments according to whether or not they were concerned with animal protection law. 400 of these judgments were used to train various models, each of which was used to predict the classification of the remaining 100 judgments. The predictions of each model were superior to a baseline measure intended to mimic current searching practice, with the best performing model being a support vector machine (SVM) approach that classified judgments according to term frequency—inverse document frequency (TF-IDF) values. Investigation of this model consisted of considering its most influential features and conducting an error analysis of all incorrectly predicted judgments. This showed the features indicative of animal protection law judgments to include terms such as ‘welfare’, ‘hunt’ and ‘cull’, and that incorrectly predicted judgments were often deemed marginal decisions by the domain expert. The TF-IDF SVM was then used to classify non-labelled judgments, resulting in an initial animal protection law repository. Inspection of this repository suggested that there were 175 animal protection judgments between January 2000 and December 2020 from the Privy Council, House of Lords, Supreme Court and upper England and Wales courts. (shrink)

This study investigates the performance of 21 monolingual and 56 bilingual children aged 5;6-9;0 on German-LITMUS-sentence-repetition (SRT; Hamann et al., 2013) and nonword-repetition-tasks (NWRT; Grimm et al., 2014), which were constructed according to the LITMUS-principles (Language Impairment Testing in Multilingual Settings; Armon-Lotem et al., 2015). Both tasks incorporate complex structures shown to be cross-linguistically challenging for children with Specific Language Impairment (SLI) and aim at minimizing bias against bilingual children while still being indicative of the presence of language impairment across (...) language combinations. Given the great variability in bilingual language exposure and the potential effect of language experience on language performance in bilingual children, we examined whether background variables related to bilingualism, particularly, the degree language dominance as measured by relative amount of use and exposure, could compromise the diagnostic accuracy of the German-LITMUS-SRT and NWRT. We further investigated whether a combination of the two tasks provides better diagnostic accurracy. To address this, we used an unsupervised machine learning algorithm, the Partitioning-Around-Medoids (PAM), for deriving a clinical category for the children as ±language-impaired based on performance scores on SRT and/or NWRT while withholding information about their clinical status based on standardized assessment in L1/L2. Subsequently, we calculated diagnostic accuracy and investigated which background variables (age of onset, length of exposure, degree of language-dominance, socio-economic-status, and risk factors for SLI) best explained clinical-group-membership yielded from the PAM-analysis based on NWRT and/or SRT scores. Results show that although language-dominance influences the performance of bilingual typically-developing children, especially in the SRT, the diagnostic accuracy of the tools is not compromised by language dominance: while risk factors for SLI were significant predictors for clinical group membership in all models, language dominance did not contribute at all to explaining clinical cluster membership as typically-developing or SLI based on any of the combinations of the SRT and NWRT variables. Additionally, results confirm that a combination of SRT scored by correct target structure and the structurally more complex language-dependent part of the NWRT yields better diagnostic accuracy than single measures and is only sensitive for risk factors for SLI and not to dominance levels or SES. (shrink)

Machine learning researchers distinguish between reinforcement learning and supervised learning and refer to reinforcement learning systems as “agents”. This paper vindicates the claim that systems trained by reinforcement learning are agents while those trained by supervised learning are not. Systems of both kinds satisfy Dretske’s criteria for agency, because they both learn to produce outputs selectively in response to inputs. However, reinforcement learning is sensitive to the instrumental value of outputs, giving rise to systems which exploit the effects of (...) outputs on subsequent inputs to achieve good performance over episodes of interaction with their environments. Supervised learning systems, in contrast, merely learn to produce better outputs in response to individual inputs. (shrink)

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

When courts started publishing judgements, big data analysis within the legal domain became possible. By taking data from the European Court of Human Rights as an example, we investigate how natural language processing tools can be used to analyse texts of the court proceedings in order to automatically predict judicial decisions. With an average accuracy of 75% in predicting the violation of 9 articles of the European Convention on Human Rights our approach highlights the potential of machine learning approaches (...) in the legal domain. We show, however, that predicting decisions for future cases based on the cases from the past negatively impacts performance. Furthermore, we demonstrate that we can achieve a relatively high classification performance when predicting outcomes based only on the surnames of the judges that try the case. (shrink)

I consider three aspects in which machine learning and philosophy of science can illuminate each other: methodology, inductive simplicity and theoretical terms. I examine the relations between the two subjects and conclude by claiming these relations to be very close.

We argue why interpretability should have primacy alongside empiricism for several reasons: first, if machine learning models are beginning to render some of the high-risk healthcare decisions instead of clinicians, these models pose a novel medicolegal and ethical frontier that is incompletely addressed by current methods of appraising medical interventions like pharmacological therapies; second, a number of judicial precedents underpinning medical liability and negligence are compromised when ‘autonomous’ ML recommendations are considered to be en par with human instruction in (...) specific contexts; third, explainable algorithms may be more amenable to the ascertainment and minimisation of biases, with repercussions for racial equity as well as scientific reproducibility and generalisability. We conclude with some reasons for the ineludible importance of interpretability, such as the establishment of trust, in overcoming perhaps the most difficult challenge ML will face in a high-stakes environment like healthcare: professional and public acceptance. (shrink)

When courts started publishing judgements, big data analysis within the legal domain became possible. By taking data from the European Court of Human Rights as an example, we investigate how natural language processing tools can be used to analyse texts of the court proceedings in order to automatically predict judicial decisions. With an average accuracy of 75% in predicting the violation of 9 articles of the European Convention on Human Rights our approach highlights the potential of machine learning approaches (...) in the legal domain. We show, however, that predicting decisions for future cases based on the cases from the past negatively impacts performance. Furthermore, we demonstrate that we can achieve a relatively high classification performance when predicting outcomes based only on the surnames of the judges that try the case. (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)

Parameterization and parameter tuning are central aspects of climate modeling, and there is widespread consensus that these procedures involve certain subjective elements. Even if the use of these subjective elements is not necessarily epistemically problematic, there is an intuitive appeal for replacing them with more objective (automated) methods, such as machine learning. Relying on several case studies, we argue that, while machine learning techniques may help to improve climate model parameterization in several ways, they still require expert judgment (...) that involves subjective elements not so different from the ones arising in standard parameterization and tuning. The use of machine learning in parameterizations is an art as well as a science and requires careful supervision. (shrink)

In this paper, we pursue three general aims: (I) We will define a notion of fundamental opacity and ask whether it can be found in High Energy Physics (HEP), given the involvement of machine learning (ML) and computer simulations (CS) therein. (II) We identify two kinds of non-fundamental, contingent opacity associated with CS and ML in HEP respectively, and ask whether, and if so how, they may be overcome. (III) We address the question of whether any kind of opacity, (...) contingent or fundamental, is unique to ML or CS, or whether they stand in continuity to kinds of opacity associated with other scientific research. (shrink)

Typically, fair machine learning research focuses on a single decision maker and assumes that the underlying population is stationary. However, many of the critical domains motivating this work are characterized by competitive marketplaces with many decision makers. Realistically, we might expect only a subset of them to adopt any non-compulsory fairness-conscious policy, a situation that political philosophers call partial compliance. This possibility raises important questions: how does partial compliance and the consequent strategic behavior of decision subjects affect the allocation (...) outcomes? If k% of employers were to voluntarily adopt a fairness-promoting intervention, should we expect k% progress (in aggregate) towards the benefits of universal adoption, or will the dynamics of partial compliance wash out the hoped-for benefits? How might adopting a global (versus local) perspective impact the conclusions of an auditor? In this paper, we propose a simple model of an employment market, leveraging simulation as a tool to explore the impact of both interaction effects and incentive effects on outcomes and auditing metrics. Our key findings are that at equilibrium: (1) partial compliance by k% of employers can result in far less than proportional (k%) progress towards the full compliance outcomes; (2) the gap is more severe when fair employers match global (vs local) statistics; (3) choices of local vs global statistics can paint dramatically different pictures of the performance vis-a-vis fairness desiderata of compliant versus non-compliant employers; (4) partial compliance based on local parity measures can induce extreme segregation. Finally, we discuss implications for auditors and insights concerning the design of regulatory frameworks. (shrink)

Aesthetic experience is original, dynamic and ever-changing. This article covers three research questions (RQs) concerning how immersive installation artworks can elicit emotions that may contribute to their popularity. Based on Yayoi Kusama’s and Peter Kogler’s kaleidoscopic rooms, this study aims to predict the emotions of visitors of immersive installation art based on their Twitter activity. As indicators, we employed the total number of likes, comments, retweets, followers, followings, the average of tweets per user, and emotional response. According to our evaluation (...) of emotions, panic obtained the highest scores. Furthermore, compared to traditional machine learning algorithms, Tree-based Pipeline Optimization Tool (TPOT) Automated Machine Learning used in this research yielded slightly lower performance. We forecast that our findings will stimulate future research in the fields of data analysis, cultural heritage management and marketing, aesthetics and cultural analytics. (shrink)

Objective: To examine the role of explainability in machine learning for healthcare (MLHC), and its necessity and significance with respect to effective and ethical MLHC application. Study Design and Setting: This commentary engages with the growing and dynamic corpus of literature on the use of MLHC and artificial intelligence (AI) in medicine, which provide the context for a focused narrative review of arguments presented in favour of and opposition to explainability in MLHC. Results: We find that concerns regarding explainability (...) are not limited to MLHC, but rather extend to numerous well-validated treatment interventions as well as to human clinical judgment itself. We examine the role of evidence-based medicine in evaluating unexplainable treatments and technologies, and highlight the analogy between the concept of explainability in MLHC and the related concept of mechanistic reasoning in evidence-based medicine. Conclusion: Ultimately, we conclude that the value of explainability in MLHC is not intrinsic, but is instead instrumental to achieving greater imperatives such as performance and trust. We caution against the uncompromising pursuit of explainability, and advocate instead for the development of robust empirical methods to successfully evaluate increasingly inexplicable algorithmic systems. (shrink)

Machine learning is the study of algorithms that enable computers to improve their performance and increase their knowledge base. Research in machine learning has taken place since the beginning of artificial intelligence in the mid‐1950s. The first notable success was Arthur Samuel's program that learned to play checkers well enough to beat skilled humans. The program estimated the best move in a situation by using a mathematical function whose sixteen parameters describe board positions, and it improved its performance (...) by adjusting weights on the parameters in order to generalize about what moves were successful. (shrink)

Decisions based on algorithmic, machine learning models can be unfair, reproducing biases in historical data used to train them. While computational techniques are emerging to address aspects of these concerns through communities such as discrimination-aware data mining and fairness, accountability and transparency machine learning, their practical implementation faces real-world challenges. For legal, institutional or commercial reasons, organisations might not hold the data on sensitive attributes such as gender, ethnicity, sexuality or disability needed to diagnose and mitigate emergent indirect (...) discrimination-by-proxy, such as redlining. Such organisations might also lack the knowledge and capacity to identify and manage fairness issues that are emergent properties of complex sociotechnical systems. This paper presents and discusses three potential approaches to deal with such knowledge and information deficits in the context of fairer machine learning. Trusted third parties could selectively store data necessary for performing discrimination discovery and incorporating fairness constraints into model-building in a privacy-preserving manner. Collaborative online platforms would allow diverse organisations to record, share and access contextual and experiential knowledge to promote fairness in machine learning systems. Finally, unsupervised learning and pedagogically interpretable algorithms might allow fairness hypotheses to be built for further selective testing and exploration. Real-world fairness challenges in machine learning are not abstract, constrained optimisation problems, but are institutionally and contextually grounded. Computational fairness tools are useful, but must be researched and developed in and with the messy contexts that will shape their deployment, rather than just for imagined situations. Not doing so risks real, near-term algorithmic harm. (shrink)

The workshop “Machine Learning: Prediction Without Explanation?” brought together philosophers of science and scholars from various fields who study and employ Machine Learning (ML) techniques, in order to discuss the changing face of science in the light of ML's constantly growing use. One major focus of the workshop was on the impact of ML on the concept and value of scientific explanation. One may speculate whether ML’s increased use in science exemplifies a paradigmatic turn towards mere pattern recognition (...) and prediction and away from science’s traditional aim of explanation. In contrast, certain conceptions of explanation, such as statistical explanation, could turn out to fit well with the achievements of present-day ML and concede an explanatory value to these achievements after all. It is an open question how to explain ML successes themselves, and this question was in the focus of several talks in the workshop. Based on the topics raised, we will discuss the talks’ contents in more detail as organized into (i) practitioners’ perspectives, (ii) explanations from ML, (iii) explanations of ML, (iv) societal implications, and (v) global and historical perspectives. (shrink)

As scientists start to adopt machine learning (ML) as one research tool, the security of ML and the knowledge generated become a concern. In this paper, I explain how supervised ML can be improved with better data ontology, or the way we make categories and turn information into data. More specifically, we should design data ontology in such a way that is consistent with the knowledge that we have about the target phenomenon so that such ontology can help us (...) make the inductive leap. I do so by thinking through a thought experiment, Goodman’s New Riddle of Induction (Fact, fiction, and forecast, Harvard University Press, 1955). Goodman’s riddle helps flesh out three problems of induction: (1) the problem of equal goodies, that there are often too many equally good inductive results given the same data; (2) the problem of diverging performance, that these equally good results can give opposite predictions in the future; and (3) the problem of mediocrity, that when averaged across all equally possible datasets and tasks, no inductive algorithm outperforms any other. I show that all these three problems are manifested as real obstacles in ML practice, namely, the Rashomon effect (Breiman in Stat Sci 16(3):199–231, 2001), the problem of underspecification (D’Amour et al. in J Mach Learn Res, 2020, https://doi.org/10.48550/arXiv.2011.03395 ), and the No Free Lunch theorem (Wolpert in Neural Comput 8(7):1341–90, 1996, https://doi.org/10.1162/neco.1996.8.7.1341 ). Lastly, I argue that proper data ontology can help mitigate these problems and I demonstrate how using concrete examples from climate science. This research highlights the links between philosophers’ discussions of induction and implications in ML practice. (shrink)

This paper develops an account of how the implementation of ML models into healthcare settings requires revising the methodological apparatus of philosophical bioethics. On this account, ML models are cognitive interventions that provide decision-support to physicians and patients. Due to reliability issues, opaque reasoning processes, and information asymmetries, ML models pose inferential problems for them. These inferential problems lay the grounds for many ethical problems that currently claim centre-stage in the bioethical debate. Accordingly, this paper argues that the best way (...) to make progress in remedying these ethical problems is to distil their epistemic core and to identify appropriate epistemic norms as guardrails. The viability of this approach will be highlighted based on four key issues: trust, responsibility, paternalism, and fairness. In that respect, the paper also contributes to a more pronounced view of the specific methodological challenges for ethical theorizing, when the point of reference are complex statistical models that make predictions, rather than individual agents, acting for reasons. (shrink)

There has been an increased focus within the AI ethics literature on questions of power, reflected in the ideal of accountability supported by many Responsible AI guidelines. While this recent debate points towards the power asymmetry between those who shape AI systems and those affected by them, the literature lacks normative grounding and misses conceptual clarity on how these power dynamics take shape. In this paper, I develop a workable conceptualization of said power dynamics according to Cristiano Castelfranchi’s conceptual framework (...) of power and argue that end-users depend on a system’s developers and users, because end-users rely on these systems to satisfy their goals, constituting a power asymmetry between developers, users and end-users. I ground my analysis in the neo-republican moral wrong of domination, drawing attention to legitimacy concerns of the power-dependence relation following from the current lack of accountability mechanisms. I illustrate my claims on the basis of a risk-prediction machine learning system, and propose institutional and project-specific solutions to mitigate domination. (shrink)

As we write this research paper, we notice an explosion in popularity of machine learning in numerous fields (ranging from governance, education, and management to criminal justice, fraud detection, and internet of things). In this contribution, rather than focusing on any of those fields, which have been well-reviewed already, we decided to concentrate on a series of more recent applications of deep learning models and technologies that have only recently gained significant track in the relevant literature. These applications are (...) concerned with artistic production (Sect. 2.1), the writing process (Sect. 2.2), music production (Sect. 2.3), text recognition and attribution (Sect. 2.4). After reviewing and analyzing the positive contributions as well as some of the major limitations of these technologies in each of those fields, we critically reflect (Sect. 3) on how their widespread implementation may affect humans and their creativity. In Sect. 4, we notice that deep learning models are here to stay; so, rather than embracing a negative or pessimistic stance with respect to their future applications in creative domains, we suggest a balanced approach for their assessment and beneficial usage. Finally (Sect. 5), we conclude by summarising what we have achieved and by pointing out possible future research directions. (shrink)

The use of AI algorithms in criminal trials has been the subject of very lively ethical and legal debates recently. While there are concerns over the lack of accuracy and the harmful biases that certain algorithms display, new algorithms seem more promising and might lead to more accurate legal decisions. Algorithms seem especially relevant for bail decisions, because such decisions involve statistical data to which human reasoners struggle to give adequate weight. While getting the right legal outcome is a strong (...) desideratum of criminal trials, advocates of the relational theory of procedural justice give us good reason to think that fairness and perceived fairness of legal procedures have a value that is independent from the outcome. According to this literature, one key aspect of fairness is trustworthiness. In this paper, I argue that using certain algorithms to assist bail decisions could increase three diferent aspects of judges’ trustworthiness: (1) actual trustworthiness, (2) rich trustworthiness, and (3) perceived trustworthiness. (shrink)

This article assesses how autonomy and machine learning impact the existential risk of nuclear war. It situates the problem of cyber security, which proceeds by stealth, within the larger context of nuclear deterrence, which is effective when it functions with transparency and credibility. Cyber vulnerabilities poses new weaknesses to the strategic stability provided by nuclear deterrence. This article offers best practices for the use of computer and information technologies integrated into nuclear weapons systems. Focusing on nuclear command and control, (...) avoiding autonomy and machine learning is recommended as one means to reduce the existential risk of unintended nuclear conflict. (shrink)

In this paper, we explore the possibility that machine learning approaches to naturallanguage processing being developed in engineering-oriented computational linguistics may be able to provide specific scientific insights into the nature of human language. We argue that, in principle, machine learning results could inform basic debates about language, in one area at least, and that in practice, existing results may offer initial tentative support for this prospect. Further, results from computational learning theory can inform arguments carried on within (...) linguistic theory as well. (shrink)

Machine learning (ML) systems play an increasingly relevant role in medicine and healthcare. As their applications move ever closer to patient care and cure in clinical settings, ethical concerns about the responsibility of their use come to the fore. I analyse an aspect of responsible ML use that bears not only an ethical but also a significant epistemic dimension. I focus on ML systems’ role in mediating patient–physician relations. I thereby consider how ML systems may silence patients’ voices and (...) relativise the credibility of their opinions, which undermines their overall credibility status without valid moral and epistemic justification. More specifically, I argue that withholding credibilitydue tohow ML systems operate can be particularly harmful to patients and, apart from adverse outcomes, qualifies as a form of testimonial injustice. I make my case for testimonial injustice in medical ML by considering ML systems currently used in the USA to predict patients’ risk of misusing opioids (automated Prediction Drug Monitoring Programmes, PDMPs for short). I argue that the locus of testimonial injustice in ML-mediated medical encounters is found in the fact that these systems are treated asmarkers of trustworthinesson which patients’ credibility is assessed. I further show how ML-based PDMPs exacerbate and further propagate social inequalities at the expense of vulnerable social groups. (shrink)

The present paper shows how statistical learning theory and machine learning models can be used to enhance understanding of AI-related epistemological issues regarding inductive reasoning and reliability of generalisations. Towards this aim, the paper proceeds as follows. First, it expounds Price’s dual image of representation in terms of the notions of e-representations and i-representations that constitute subject naturalism. For Price, this is not a strictly anti-representationalist position but rather a dualist one (e- and i-representations). Second, the paper links this (...) debate with machine learning in terms of statistical learning theory becoming more viable epistemological tool when it abandons the perspective of object naturalism. The paper then argues that machine learning grounds a form of knowing that can be understood in terms of e- and i-representation learning. Third, this synthesis shows a way of analysing inductive reasoning in terms of reliability of generalisations stemming from a structure of e- and i-representations. In the age of Artificial Intelligence, connecting Price’s dual view of representation with Deep Learning provides an epistemological way forward and even perhaps an approach to how knowing is possible. (shrink)

Machine learning and statistical methods have yielded impressive results in a wide variety of natural language processing tasks. These advances have generally been regarded as engineering achievements. In fact it is possible to argue that the success of machine learning methods is significant for our understanding of the cognitive basis of language acquisition and processing. Recent work in unsupervised grammar induction is particularly relevant to this issue. It suggests that knowledge of language can be achieved through general learning (...) procedures, and that a richly articulated language faculty is not required to explain its acquisition. (shrink)

Learning general concepts in imperfect environments is difficult since training instances often include noisy data, inconclusive data, incomplete data, unknown attributes, unknown attribute values and other barriers to effective learning. It is well known that people can learn effectively in imperfect environments, and can manage to process very large amounts of data. Imitating human learning behavior therefore provides a useful model for machine learning in real-world applications. This paper proposes a new, more effective way to represent imperfect training instances (...) and rules, and based on the new representation, a Human-Like Learning (HULL) algorithm for incrementally learning concepts well in imperfect training environments. Several examples are given to make the algorithm clearer. Finally, experimental results are presented that show the proposed learning algorithm works well in imperfect learning environments. (shrink)

The operations of deep networks are widely acknowledged to be inscrutable. The growing field of Explainable AI has emerged in direct response to this problem. However, owing to the nature of the opacity in question, XAI has been forced to prioritise interpretability at the expense of completeness, and even realism, so that its explanations are frequently interpretable without being underpinned by more comprehensive explanations faithful to the way a network computes its predictions. While this has been taken to be a (...) shortcoming of the field of XAI, I argue that it is broadly the right approach to the problem. (shrink)

In this paper, we explore the possibility that machine learning approaches to naturallanguage processing being developed in engineering-oriented computational linguistics may be able to provide specific scientific insights into the nature of human language. We argue that, in principle, machine learning results could inform basic debates about language, in one area at least, and that in practice, existing results may offer initial tentative support for this prospect. Further, results from computational learning theory can inform arguments carried on within (...) linguistic theory as well. (shrink)

Whether it would take one decade or several centuries, many agree that it is possible to create a *superintelligence*---an artificial intelligence with a godlike ability to achieve its goals. And many who have reflected carefully on this fact agree that our best hope for a "friendly" superintelligence is to design it to *learn* values like ours, since our values are too complex to program or hardwire explicitly. But the value learning approach to AI safety faces three particularly philosophical puzzles: first, (...) it is unclear how any intelligent system could learn its final values, since to judge one supposedly "final" value against another seems to require a further background standard for judging. Second, it is unclear how to determine the content of a system's values based on its physical or computational structure. Finally, there is the distinctly ethical question of which values we should best aim for the system to learn. I outline a potential answer to these interrelated puzzles, centering on a "miktotelic" proposal for blending a complex, learnable final value out of many simpler ones. (shrink)

Machine learning has become a key component of contemporary information systems. Unlike prior information systems explicitly programmed in formal languages, ML systems infer rules from data. This paper shows what this difference means for the critical analysis of socio-technical systems based on machine learning. To provide a foundation for future critical analysis of machine learning-based systems, we engage with how the term is framed and constructed in self-education resources. For this, we analyze machine learning tutorials, an (...) important information source for self-learners and a key tool for the formation of the practices of the machine learning community. Our analysis identifies canonical examples of machine learning as well as important misconceptions and problematic framings. Our results show that machine learning is presented as being universally applicable and that the application of machine learning without special expertise is actively encouraged. Explanations of machine learning algorithms are missing or strongly limited. Meanwhile, the importance of data is vastly understated. This has implications for the manifestation of social inequalities through machine learning-based systems. (shrink)

Rapid advancements in machine learning techniques allow mass surveillance to be applied on larger scales and utilize more and more personal data. These developments demand reconsideration of the privacy-security dilemma, which describes the tradeoffs between national security interests and individual privacy concerns. By investigating mass surveillance techniques that use bulk data collection and machine learning algorithms, we show why these methods are unlikely to pinpoint terrorists in order to prevent attacks. The diverse characteristics of terrorist attacks—especially when considering (...) lone-wolf terrorism—lead to irregular and isolated footprints. The irregularity of data affects the accuracy of machine learning algorithms and the mass surveillance that depends on them which can be explained by three kinds of known problems encountered in machine learning theory: class imbalance, the curse of dimensionality, and spurious correlations. Proponents of mass surveillance often invoke the distinction between collecting data and metadata, in which the latter is understood as a lesser breach of privacy. Their arguments commonly overlook the ambiguity in the definitions of data and metadata and ignore the ability of machine learning techniques to infer the former from the latter. Given the sparsity of datasets used for machine learning in counterterrorism and the privacy risks attendant with bulk data collection, policymakers and other relevant stakeholders should critically re-evaluate the likelihood of success of the algorithms and the collection of data on which they depend. (shrink)

We apply an online optimization process based on machine learning to the production of Bose-Einstein condensates. BEC is typically created with an exponential evaporation ramp that is optimal for ergodic dynamics with two-body s-wave interactions and no other loss rates, but likely sub-optimal for real experiments. Through repeated machine-controlled scientific experimentation and observations our ’learner’ discovers an optimal evaporation ramp for BEC production. In contrast to previous work, our learner uses a Gaussian process to develop a statistical model (...) of the relationship between the parameters it controls and the quality of the BEC produced. We demonstrate that the Gaussian process machine learner is able to discover a ramp that produces high quality BECs in 10 times fewer iterations than a previously used online optimization technique. Furthermore, we show the internal model developed can be used to determine which parameters are essential in BEC creation and which are unimportant, providing insight into the optimization process of the system. (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)

As machine learning informs increasingly consequential decisions, different metrics have been proposed for measuring algorithmic bias or unfairness. Two popular “fairness measures” are calibration and equality of false positive rate. Each measure seems intuitively important, but notably, it is usually impossible to satisfy both measures. For this reason, a large literature in machine learning speaks of a “fairness tradeoff” between these two measures. This framing assumes that both measures are, in fact, capturing something important. To date, philosophers have (...) seldom examined this crucial assumption, and examined to what extent each measure actually tracks a normatively important property. This makes this inevitable statistical conflict – between calibration and false positive rate equality – an important topic for ethics. In this paper, I give an ethical framework for thinking about these measures and argue that, contrary to initial appearances, false positive rate equality is in fact morally irrelevant and does not measure fairness. (shrink)

Informed consent is at the core of the clinical relationship. With the introduction of machine learning in healthcare, the role of informed consent is challenged. This paper addresses the issue of whether patients must be informed about medical ML applications and asked for consent. It aims to expose the discrepancy between ethical and practical considerations, while arguing that this polarization is a false dichotomy: in reality, ethics is applied to specific contexts and situations. Bridging this gap and considering the (...) whole picture is essential for advancing the debate. In the light of the possible future developments of the situation and the technologies, as well as the benefits that informed consent for ML can bring to shared decision-making, the present analysis concludes that it is necessary to prepare the ground for a possible future requirement of informed consent for medical ML. (shrink)

As more objections have been raised against grant peer-review for being costly and time-consuming, the legitimate question arises whether machine learning algorithms could help assess the epistemic efficiency of the proposed projects. As a case study, we investigated whether project efficiency in high energy physics can be algorithmically predicted based on the data from the proposal. To analyze the potential of algorithmic prediction in HEP, we conducted a study on data about the structure and outcomes of HEP experiments with (...) the goal of predicting their efficiency. In the first step, we assessed the project efficiency using Data Envelopment Analysis of 67 experiments conducted in the HEP laboratory Fermilab. In the second step, we employed predictive algorithms to detect which team structures maximize the epistemic performance of an expert group. For this purpose, we used the efficiency scores obtained by DEA and applied predictive algorithms – lasso and ridge linear regression, neural network, and gradient boosted trees – on them. The results of the predictive analyses show moderately high accuracy, indicating that they can be beneficial as one of the steps in grant review. Still, their applicability in practice should be approached with caution. Some of the limitations of the algorithmic approach are the unreliability of citation patterns, unobservable variables that influence scientific success, and the potential predictability of the model. (shrink)

Nowadays, online product reviews have been at the heart of the product assessment process for a company and its customers. They give feedback to a company on improving product quality, planning, and monitoring its business schemes in order to increase sale and gain more profit. They are also helpful for customers to select the right products in less effort and time. Most companies make spam reviews of products in order to increase the products sales and gain more profit. Detecting spam (...) product reviews is a challenging issue in NLP. Numerous machine learning approaches have attempted to detect and classify the product reviews as spam or nonspam. However, in order to improve the classification accuracy, this study has introduced an ensemble machine learning model that combines predictions from multilayer perceptron, K-Nearest Neighbour, and Random Forest and predicts the outcome of the review as spam or real, based on the majority vote of the contributing models. In order to accomplish the task of spam review classification, the proposed ensemble and other benchmark boosting approaches are tested with 25 statistical features extracted from mobile application reviews of Yelp Dataset. Then, three different selection techniques are exploited to diminish the feature space and filter out the top 10 optimal features. The effectiveness of the proposed ensemble, the individual models, and other benchmark boosting approaches is again evaluated with 10 optimal features in terms of classification accuracy. Experimental outcomes illustrate that the proposed ensemble model outperformed the individual classifiers and state-of-the-art boosting approaches like Generalized Boost Regression Model, Extreme Gradient Boost, and AdaBoost Regression Model in terms of classification accuracy. (shrink)

The increasing precision of observations of the large-scale structure of the universe has created a problem for simulators: running the simulations necessary to interpret these observations has become impractical. Simulators have thus turned to machine learning (ML) algorithms instead. Though ML decreases computational expense, one might be worried about the use of ML for scientific investigations: How can algorithms that have repeatedly been described as black-boxes deliver scientific understanding? In this paper, I investigate how cosmologists employ ML, arguing that (...) in this context, ML algorithms should not be considered black-boxes and can deliver genuine scientific understanding. Accordingly, understanding the methodological role of ML algorithms is crucial to understanding the types of questions they are capable of, and ought to be responsible for, answering. (shrink)