Search

In a recent paper [1], Brian Hedden has argued that most of the group fairness constraints discussed in the machine learning literature are not necessary conditions for the fairness of predictions, and hence that there are no genuine fairness metrics. This is proven by discussing a special case of a fair prediction. In our paper, we show that Hedden's argument does not hold for the most common kind of predictions used in data science, which are about people and based on data from similar people; we call these “human-group-based practices.” We argue that there is a morally salient distinction between human-group-based practices and those that are based on data of only one person, which we call “human-individual-based practices.” Thus, what may be a necessary condition for the fairness of human-group-based practices may not be a necessary condition for the fairness of human-individual-based practices, on which Hedden's argument is based. Accordingly, the group fairness metrics discussed in the machine learning literature may still be relevant for most applications of prediction-based decision making.

Many literature reviews on artificial intelligence (AI) in higher education or in education in general have focused on the different applications of AI in this domain, the AI techniques used, and the benefits/risks of the use of AI. One of the greatest potentials of AI is to personalise higher education to the needs of students and offer timely feedback. This could benefit students with disabilities tremendously if their needs are also considered in the development of new AI educational technologies (Edtech). However, current reviews fail to address the perspective of students with disabilities. This perspective is essential because AI is likely to bring several ethical concerns for people with disabilities. For instance, AI can treat people with disabilities as outliers in the data and end up discriminating against them. For that reason, two questions were raised: To what extent are ethical concerns relevant for students with disabilities considered in articles presenting AI applications assessing students in higher education? What are the potential risks of using AI that assess students with disabilities in higher education? This extended abstract presents summarised results of a scoping review that will be published in a journal. The goal of this article is to start a discussion within the AI ethics community to raise awareness about the issues that students with disabilities may face and to collaboratively explore solutions. Results suggest that there is a lack of ethical reflection on AI technologies and an absence of discussion and inclusion of people with disabilities. Moreover, risks associated with utilising AI for students with disabilities relate to the choice of data, reliance on simplistic classification, face monitoring, and the low involvement of students.

In this paper, we provide a moral analysis of two criteria of statistical fairness debated in the machine learning literature: 1) calibration between groups and 2) equality of false positive and false negative rates between groups. In our paper, we focus on moral arguments in support of either measure. The conflict between group calibration vs. false positive and false negative rate equality is one of the core issues in the debate about group fairness definitions among practitioners. For any thorough moral analysis, the meaning of the term “fairness” has to be made explicit and defined properly. For our paper, we equate fairness with (non-)discrimination, which is a legitimate understanding in the discussion about group fairness. More specifically, we equate it with “prima facie wrongful discrimination” in the sense this is used in Prof. Lippert-Rasmussen's treatment of this definition. In this paper, we argue that a violation of group calibration may be unfair in some cases, but not unfair in others. Our argument analyzes in great detail two specific hypothetical examples of usage of predictions in decision making. The most important practical implication is that between-group calibration is defensible as a bias standard in some cases but not others; we show this by referring to examples in which the violation of between-group calibration is discriminatory, and others in which it is not. This is in line with claims already advanced in the literature, that algorithmic fairness should be defined in a way that is sensitive to context. The most important practical implication is that arguments based on examples in which fairness requires between-group calibration, or equality in the false-positive/false-negative rates, do no generalize. For it may be that group calibration is a fairness requirement in one case, but not in another.

This paper supports the personal data platform cooperative as a means of bringing about John Rawls’s favoured institutional realisation of a just society, the property-owning democracy. It describes personal data platform cooperatives and applies Rawls’s political philosophy to analyse the institutional forms of a just society in relation to the economic power deriving from aggregating personal data. It argues that a society involving a significant number of personal data platform cooperatives will be more suitable to realising Rawls’s principle of fair equality of opportunity.

Trust and monitoring are traditionally antithetical concepts. Describing trust as a property of a relationship of reliance, we introduce a theory of trust and monitoring, which uses mathematical models based on two classes of functions, including q-exponentials, and relates the levels of trust to the costs of monitoring. As opposed to several accounts of trust that attempt to identify the special ingredient of reliance and trust relationships, our theory characterizes trust as a quantitative property of certain relations of reliance that can be quantified and expressed as a scalar quantity. Our theory is applicable to both human–human and human–artificial agent interactions, as it is agnostic with respect to the concrete realization of trustworthiness properties, and is compatible with many views differing on which properties contribute to trust and trustworthiness. Finally, as our mathematical models make the quantitative features of trust measurable, they provide empirical studies on trust with a rigorous methodology for its measurement.

The widespread use of algorithms for prediction-based decisions urges us to consider the question of what it means for a given act or practice to be discriminatory. Building upon work by Kusner and colleagues in the field of machine learning, we propose a counterfactual condition as a necessary requirement on discrimination. To demonstrate the philosophical relevance of the proposed condition, we consider two prominent accounts of discrimination in the recent literature, by Lippert-Rasmussen and Hellman respectively, that do not logically imply our condition and show that they face important objections. Specifically, Lippert-Rasmussen’s definition proves to be over-inclusive, as it classifies some acts or practices as discriminatory when they are not, whereas Hellman’s account turns out to lack explanatory power precisely insofar as it does not countenance a counterfactual condition on discrimination. By defending the necessity of our counterfactual condition, we set the conceptual limits for justified claims about the occurrence of discriminatory acts or practices in society, with immediate applications to the ethics of algorithmic decision-making.

The Code is available in German, English, French, and Italian., The “Code of Ethics for Data-Based Value Creation” is aimed at companies and organizations that offer services or products based on data. Its purpose is to systematically address the ethical issues that arise in the creation or use of such products and services. To this end, concrete recommendations are made, based on three ethical and three procedural values and structured by the four main steps of the data life cycle. The Code consists of the following documents: 1) Overview (as poster in formats A2 and A3) 2) Basics 3) Recommendations 4) Implementation 5) Context

Up to date, more than 80 codes exist for handling ethical risks of artificial intelligence and big data. In this paper, we analyse where those codes converge and where they differ. Based on an in-depth analysis of 20 guidelines, we identify three procedural action types (1. control and document, 2. inform, 3. assign responsibility) as well as four clusters of ethical values whose promotion or protection is supported by the procedural activities. We achieve a synthesis of previous approaches with a framework of seven principles, combining the four principles of biomedical ethics with three distinct procedural principles: control, transparency and accountability.

Clients may feel trapped into sharing their private digital data with insurance companies to get a desired insurance product or premium. However, private insurance must collect some data to offer products and premiums appropriate to the client’s level of risk. This situation creates tension between the value of privacy and common insurance business practice. We argue for three main claims: first, coercion to share private data with insurers is pro tanto wrong because it violates the autonomous choice of a privacy-valuing client. Second, we maintain that irrespective of being coerced, the choice of accepting digital surveillance by insurers makes it harder for the client to protect his or her autonomy (and to act spontaneously and authentically). The violation of autonomy also makes coercing customers into digital surveillance pro tanto morally wrong. Third, having identified an economically plausible process involving no direct coercion by insurers, leading to the adoption of digital surveillance, we argue that such an outcome generates further threats against autonomy. This threat provides individuals with a pro tanto reason to prevent this process. We highlight the freedom dilemma faced by regulators who aim to prevent this outcome by constraining market freedoms and argue for the need for further moral and empirical research on this question.

In a recent paper, Brian Hedden has argued that most of the group fairness constraints discussed in the machine learning literature are not necessary conditions for the fairness of predictions, and hence that there are no genuine fairness metrics. This is proven by discussing a special case of a fair prediction. In our paper, we show that Hedden's argument does not hold for the most common kind of predictions used in data science, which are about people and based on data from similar people; we call these “human-group-based practices.” We argue that there is a morally salient distinction between human-group-based practices and those that are based on data of only one person, which we call “human-individual-based practices.” Thus, what may be a necessary condition for the fairness of human-group-based practices may not be a necessary condition for the fairness of human-individual-based practices, on which Hedden's argument is based. Accordingly, the group fairness metrics discussed in the machine learning literature may still be relevant for most applications of prediction-based decision making.

The ELSI White Paper is the final achievement of the ELSI Task Force for the National Research Programme “Big Data” (NRP 75). It is an informational document that provides an overview of the key ethical, legal, and social challenges of big data and provides guidance for the collection, use, and sharing of big data. The document aims to bring together the expertise of the ELSI Task Force members rather than exhaustively covering all topics in big data relating to ethical, legal, and social issues (ELSI). The white paper comprises two parts: main articles and commentaries on it. The main articles give an overview of the major concerns associated with the use of big data, based on the assessment of the participating researchers. The commentary articles either examine in depth one or more of the issues that are presented in the main articles or highlight other issues that are considered relevant by their authors but are not covered in the main articles. The main articles are divided into three sections corresponding to the three ELSI levels of analysis. In the section on ethics, Marcello Ienca explores the threat of big data to ethics commissions, privacy rights, personal autonomy, and equality in the healthcare sector and biomedical research. Bernice Elger focuses on the need to address informed consent differently and complement it with additional mechanisms in the big data context. In the legal section, Christophe Schneble explores whether current Swiss data protection laws adequately regulate and protect individuals’ data. Eleonora Viganò analyses the threat of big data to state sovereignty and explore the two contrasting acceptations of the term “digital sovereignty” in the context of big data. In the section on social issues, Markus Christen addresses the big data divide, namely, the uneven distribution of benefits and harms from big data and the connected issue of the transparency asymmetry between data givers and data owners. Michele Loi delves into the debate o

This chapter provides a political and philosophical analysis of the values at stake in ensuring cybersecurity for critical infrastructures. It presents a review of the boundaries of cybersecurity in national security, with a focus on the ethics of surveillance for protecting critical infrastructures and the use of AI. A bibliographic analysis of the literature is applied until 2016 to identify and discuss the cybersecu- rity value conflicts and ethical issues in national security. This is integrated with an analysis of the most recent literature on cyber-threats to national infrastructure and the role of AI. This chapter demonstrates that the increased connectedness of digital and non-digital infrastructure enhances the trade-offs between values identified in the literature of the past years, and supports this thesis with the analysis of four case studies.