We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealisedexplanation gamein which players collaborate to find the best explanation(s) 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 (conditionally) 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)

Applications of artificial intelligence (AI) for cybersecurity tasks are attracting greater attention from the private and the public sectors. Estimates indicate that the market for AI in cybersecurity will grow from US$1 billion in 2016 to a US$34.8 billion net worth by 2025. The latest national cybersecurity and defence strategies of several governments explicitly mention AI capabilities. At the same time, initiatives to define new standards and certification procedures to elicit users’ trust in AI are emerging on a global scale. (...) However, trust in AI (both machine learning and neural networks) to deliver cybersecurity tasks is a double edged sword: it can improve substantially cybersecurity practices, but can also facilitate new forms of attacks to the AI applications themselves, which may pose severe security threats. We argue that trust in AI for cybersecurity is unwarranted and that, to reduce security risks, some form of control to ensure the deployment of ‘reliable AI’ for cybersecurity is necessary. To this end, we offer three recommendations focusing on the design, development and deployment of AI for cybersecurity. (shrink)

Resumo: No livro When is Knowledge True Belief? (2012), Richard Foley apresenta uma teoria do conhecimento bastante simples e original, que pretende desbancar as rivais e reorientar a teoria do conhecimento: um sujeito S sabe alguma proposição p se e somente se S verdadeiramente crê que p e não lhe falta nenhuma informação importante. Michael Hannon, em seu artigo “Is Knowledge True Belief Plus Adequate Information?” (2013), faz uma objeção à visão de Foley, acusando-o de oferecer um argumento circular que, (...) ao não diferenciar informação importante de informação não importante, acaba, em última instância, por fazer referência à própria noção de conhecimento. Acredito que a teoria apresentada por Foley está no caminho certo; porém, apresenta- se de forma incompleta. Neste artigo, apresento uma interpretação subjetivista de relevância epistêmica, proposta originalmente por Floridi, que complementa a visão proposta por Foley e rechaça as objeções levantadas por Hannon. Palavras-chave: Conhecimento. Informação. Relevância. -/- Abstract: In When is Knowledge True Belief? (2012), Richard Foley offers an original and extraordinarily simple theory of knowledge which intends to overcome its rivals and reorient the theory of knowledge: a subject S knows some proposition p if and only if S truly believes that p and does not lack any important information. Michael Hannon, in his article “Is True Belief Knowledge Plus Adequate Information?” (2013), puts forward an objection to Foley’s view which charges him of offering a circular argument that fails to distinguish important information from unimportant information without ultimately making reference to knowledge. I believe the theory presented by Foley is on the right . (shrink)

This paper outlines an integrated approach to trust and relevance with respect to arguments: in particular, it is suggested that trust in relevance has a central role in argumentation. We first distinguish two types of argumentative relevance: internal relevance, i.e. the extent to which a premise has a bearing on its purported conclusion, and external relevance, i.e. a measure of how much a whole argument is pertinent to the matter under discussion, in the broader dialogical context where it is proposed. (...) Then, we argue that judgements of internal relevance heavily rely on trust, and that such trust, although occasionally misplaced , is nonetheless based on several reasons, and thus often justified, by either epistemic or pragmatic considerations. We conclude by sketching potential methods to formally model trust in argumentative relevance, and briefly discussing the technological implications of this line of research. (shrink)

This article analyzes the issue of organizational transparency through the lens of Thomas Aquinas’ ethics. It provides moral justification for current claims about corporate transparency and sheds light on the ethical values and virtues affecting information disclosure decisions. Transparency is conceptualized as an informational mechanism necessary for performing the virtues of truthfulness, justice, and prudence. This article extends the organizational transparency and corporate social responsibility literatures by providing an alternative moral justification grounded in virtue-based theory, which extends our understanding of (...) the information disclosure decisions made by management. (shrink)

In 2002, Luciano Floridi published a paper called What is the Philosophy of Information?, where he argues for a new paradigm in philosophical research. To what extent should his proposal be accepted? Is the Philosophy of Information actually a new paradigm, in the Kuhninan sense, in Philosophy? Or is it only a new branch of Epistemology? In our discussion we will argue in defense of Floridi’s proposal. We believe that Philosophy of Information has the types of features had by other (...) areas already acknowledge as authentic in Philosophy. By way of an analogical argument we will argue that since Philosophy of Information has its own topics, method and problems it would be counter-intuitive not to accept it as a new philosophical area. To strengthen our position we present and discuss main topics of Philosophy of Information. (shrink)

This article presents the first, systematic analysis of the ethical challenges posed by recommender systems through a literature review. The article identifies six areas of concern, and maps them onto a proposed taxonomy of different kinds of ethical impact. The analysis uncovers a gap in the literature: currently user-centred approaches do not consider the interests of a variety of other stakeholders—as opposed to just the receivers of a recommendation—in assessing the ethical impacts of a recommender system.

Floridi’s Theory of Strongly Semantic Information posits the Veridicality Thesis. One motivation is that it can serve as a foundation for information-based epistemology being an alternative to the tripartite theory of knowledge. However, the Veridicality thesis is false, if ‘information’ is to play an explanatory role in human cognition. Another motivation is avoiding the so-called Bar-Hillel/Carnap paradox. But this paradox only seems paradoxical, if ‘information’ and ‘informativeness’ are synonymous, logic is a theory of inference, or validity suffices for rational inference; (...) a, b, and c are false. (shrink)

In this article, I shall argue that AI’s likely developments and possible challenges are best understood if we interpret AI not as a marriage between some biological-like intelligence and engineered artefacts, but as a divorce between agency and intelligence, that is, the ability to solve problems successfully and the necessity of being intelligent in doing so. I shall then look at five developments: (1) the growing shift from logic to statistics, (2) the progressive adaptation of the environment to AI rather (...) than of AI to the environment, (3) the increasing translation of difficult problems into complex problems, (4) the tension between regulative and constitutive rules underpinning areas of AI application, and (5) the push for synthetic data. (shrink)

: This article provides replies to, and comments on, the contributions to the special issue on the philosophy of information. It seeks to highlight con‐vergences and points of potential agreement, while offering clarifications and further details. It also answers some criticisms and replies to some objections articulated in the special issue.

The modern abundance and prominence of data have led to the development of “data science” as a new field of enquiry, along with a body of epistemological reflections upon its foundations, methods, and consequences. This article provides a systematic analysis and critical review of significant open problems and debates in the epistemology of data science. We propose a partition of the epistemology of data science into the following five domains: (i) the constitution of data science; (ii) the kind of enquiry (...) that it identifies; (iii) the kinds of knowledge that data science generates; (iv) the nature and epistemological significance of “black box” problems; and (v) the relationship between data science and the philosophy of science more generally. (shrink)

Semantic information is usually supposed to satisfy the veridicality thesis: p qualifies as semantic information only if p is true. However, what it means for semantic information to be true is often left implicit, with correspondentist interpretations representing the most popular, default option. The article develops an alternative approach, namely a correctness theory of truth (CTT) for semantic information. This is meant as a contribution not only to the philosophy of information but also to the philosophical debate on the nature (...) of truth. After the introduction, in Sect. 2, semantic information is shown to be translatable into propositional semantic information (i). In Sect. 3, i is polarised into a query (Q) and a result (R), qualified by a specific context, a level of abstraction and a purpose. This polarization is normalised in Sect. 4, where [Q + R] is transformed into a Boolean question and its relative yes/no answer [Q + A]. This completes the reduction of the truth of i to the correctness of A. In Sects. 5 and 6, it is argued that (1) A is the correct answer to Q if and only if (2) A correctly saturates Q by verifying and validating it (in the computer science’s sense of verification and validation ); that (2) is the case if and only if (3) [Q + A] generates an adequate model (m) of the relevant system (s) identified by Q; that (3) is the case if and only if (4) m is a proxy of s (in the computer science’s sense of proxy ) and (5) proximal access to m commutes with the distal access to s (in the category theory’s sense of commutation ); and that (5) is the case if and only if (6) reading/writing (accessing, in the computer science’s technical sense of the term) m enables one to read/write (access) s. Sect. 7 provides some further clarifications about CTT, in the light of semantic paradoxes. Section 8 draws a general conclusion about the nature of CTT as a theory for systems designers not just systems users. In the course of the article all technical expressions from computer science are explained. (shrink)

The article addresses the problem of how semantic information can be upgraded to knowledge. The introductory section explains the technical terminology and the relevant background. Section 2 argues that, for semantic information to be upgraded to knowledge, it is necessary and sufficient to be embedded in a network of questions and answers that correctly accounts for it. Section 3 shows that an information flow network of type A fulfils such a requirement, by warranting that the erotetic deficit, characterising the target (...) semantic information t by default, is correctly satisfied by the information flow of correct answers provided by an informational source s. Section 4 illustrates some of the major advantages of such a Network Theory of Account (NTA) and clears the ground of a few potential difficulties. Section 5 clarifies why NTA and an informational analysis of knowledge, according to which knowledge is accounted semantic information, is not subject to Gettier-type counterexamples. A concluding section briefly summarises the results obtained. (shrink)

Semantic information is usually supposed to satisfy the veridicality thesis: p qualifies as semantic information only if p is true. However, what it means for semantic information to be true is often left implicit, with correspondentist interpretations representing the most popular, default option. The article develops an alternative approach, namely a correctness theory of truth (CTT) for semantic information. This is meant as a contribution not only to the philosophy of information but also to the philosophical debate on the nature (...) of truth. After the introduction, in Sect. 2, semantic information is shown to be translatable into propositional semantic information (i). In Sect. 3, i is polarised into a query (Q) and a result (R), qualified by a specific context, a level of abstraction and a purpose. This polarization is normalised in Sect. 4, where [Q + R] is transformed into a Boolean question and its relative yes/no answer [Q + A]. This completes the reduction of the truth of i to the correctness of A. In Sects. 5 and 6, it is argued that (1) A is the correct answer to Q if and only if (2) A correctly saturates Q by verifying and validating it (in the computer science’s sense of “verification” and “validation”); that (2) is the case if and only if (3) [Q + A] generates an adequate model (m) of the relevant system (s) identified by Q; that (3) is the case if and only if (4) m is a proxy of s (in the computer science’s sense of “proxy”) and (5) proximal access to m commutes with the distal access to s (in the category theory’s sense of “commutation”); and that (5) is the case if and only if (6) reading/writing (accessing, in the computer science’s technical sense of the term) m enables one to read/write (access) s. Sect. 7 provides some further clarifications about CTT, in the light of semantic paradoxes. Section 8 draws a general conclusion about the nature of CTT as a theory for systems designers not just systems users. In the course of the article all technical expressions from computer science are explained. (shrink)

The paper argues that the two best known formal logical fallacies, namely denying the antecedent (DA) and affirming the consequent (AC) are not just basic and simple errors, which prove human irrationality, but rather informational shortcuts, which may provide a quick and dirty way of extracting useful information from the environment. DA and AC are shown to be degraded versions of Bayes’ theorem, once this is stripped of some of its probabilities. The less the probabilities count, the closer these fallacies (...) become to a reasoning that is not only informationally useful but also logically valid. (shrink)

In this article, I define and then defend the principle of information closure (pic) against a sceptical objection similar to the one discussed by Dretske in relation to the principle of epistemic closure. If I am successful, given that pic is equivalent to the axiom of distribution and that the latter is one of the conditions that discriminate between normal and non-normal modal logics, a main result of such a defence is that one potentially good reason to look for a (...) formalization of the logic of “ $S$ is informed that $p$ ” among the non-normal modal logics, which reject the axiom, is also removed. This is not to argue that the logic of “ $S$ is informed that $p$ ” should be a normal modal logic, but that it could still be insofar as the objection that it could not be, based on the sceptical objection against pic, has been removed. In other word, I shall argue that the sceptical objection against pic fails, so such an objection provides no ground to abandon the normal modal logic B (also known as KTB) as a formalization of “ $S$ is informed that $p$ ”, which remains plausible insofar as this specific obstacle is concerned. (shrink)

AI, especially in the case of Deepfakes, has the capacity to undermine our confidence in the original, genuine, authentic nature of what we see and hear. And yet digital technologies, in the form of databases and other detection tools also make it easier to spot forgeries and to establish the authenticity of a work. Using the notion of ectypes, this paper discusses current conceptions of authenticity and reproduction and examines how, in the future, these might be adapted for use in (...) the digital sphere. (shrink)

This article analyzes the issue of organizational transparency through the lens of Thomas Aquinas’ ethics. It provides moral justification for current claims about corporate transparency and sheds light on the ethical values and virtues affecting information disclosure decisions. Transparency is conceptualized as an informational mechanism necessary for performing the virtues of truthfulness, justice, and prudence. This article extends the organizational transparency and corporate social responsibility literatures by providing an alternative moral justification grounded in virtue-based theory, which extends our understanding of (...) the information disclosure decisions made by management. (shrink)

In 2002, Luciano Floridi published a paper called What is the Philosophy of Information?, where he argues for a new paradigm in philosophical research. To what extent should his proposal be accepted? Is the Philosophy of Information actually a new paradigm, in the Kuhninan sense, in Philosophy? Or is it only a new branch of Epistemology? In our discussion we will argue in defense of Floridi’s proposal. We believe that Philosophy of Information has the types of features had by other (...) areas already acknowledge as authentic in Philosophy. By way of an analogical argument we will argue that since Philosophy of Information has its own topics, method and problems it would be counter-intuitive not to accept it as a new philosophical area. To strengthen our position we present and discuss main topics of Philosophy of Information. (shrink)

Abstract: The core aim of this special issue is to present the philosophy of information as a way of doing philosophy, to focus on the contributions of Luciano Floridi to that area, and most important, to stimulate the debate on the most distinctive and controversial views he has defended in that context. This introduction contains a description of the philosophy of information, a discussion of two common misconceptions about the scope and the ambition of the philosophy of information, and a (...) brief overview of the essays in the issue. (shrink)

In 2002, Luciano Floridi published a paper called What is the Philosophy of Information?, where he argues for a new paradigm in philosophical research. To what extent should his proposal be accepted? Is the Philosophy of Information actually a new paradigm, in the Kuhninan sense, in Philosophy? Or is it only a new branch of Epistemology? In our discussion we will argue in defense of Floridi’s proposal. We believe that Philosophy of Information has the types of features had by other (...) areas already acknowledge as authentic in Philosophy. By way of an analogical argument we will argue that since Philosophy of Information has its own topics, method and problems it would be counter-intuitive not to accept it as a new philosophical area. To strengthen our position we present and discuss main topics of Philosophy of Information. (shrink)

Intelligibility and interpretability related to artificial intelligence (AI) are crucial for enabling explicability, which is vital for establishing constructive communication and agreement among various stakeholders, including users and designers of AI. It is essential to overcome the challenges of sharing an understanding of the details of the various structures of diverse AI systems, to facilitate effective communication and collaboration. In this paper, we propose four fundamental terms: “I/O,” “Constraints,” “Objectives,” and “Architecture.” These terms help mitigate the challenges associated with intelligibility (...) and interpretability in AI by providing appropriate levels of abstraction to describe structure of AI systems generally, thereby facilitating the sharing of understanding among various stakeholders. The relationship between the Objective of AI designers and the Purpose of AI users is linked to the issues of AI alignment. (shrink)

This article highlights some of the key lessons learnt from a recent Symposium on the Philosophy of Information. Topics covered include: semantic information, information integration, and epistemic responsibility. -/- .

The article develops a correctness theory of truth (CTT) for semantic information. After the introduction, in section two, semantic information is shown to be translatable into propositional semantic information (i). In section three, i is polarised into a query (Q) and a result (R), qualified by a specific context, a level of abstraction and a purpose. This polarization is normalised in section four, where [Q + R] is transformed into a Boolean question and its relative yes/no answer [Q + A]. (...) This completes the reduction of the truth of i to the correctness of A. In sections five and six, it is argued that (1) A is the correct answer to Q if and only if (2) A correctly saturates (in a Fregean sense) Q by verifying and validating it (in the computer science’s sense of “verification” and “validation”); that (2) is the case if and only if (3) [Q + A] generates an adequate model (m) of the relevant system (s) identified by Q; that (3) is the case if and only if (4) m is a proxy of s (in the computer science’s sense of “proxy”) and (5) proximal access to m commutes with the distal access to s (in the category theory’s sense of “commutation”); and that (5) is the case if and only if (6) reading/writing (accessing, in the computer science’s technical sense of the term) m enables one to read/write (access) s. The last section draws a general conclusion about the nature of CTT as a theory for systems designers not just systems users. (shrink)

In his article Open Problems in the Philosophy of Information [1] Luciano Floridi presented a Philosophy of Information research program in the form of eighteen open problems, covering the following fundamental areas: Information definition, information semantics, intelligence/cognition, informational universe/nature and values/ethics. We revisit Floridi’s program, highlighting some of the major advances, commenting on unsolved problems and rendering the new landscape of the Philosophy of Information (PI) emerging at present. As we analyze the progress of PI we try to situate Floridi’s (...) program in the context of scientific and technological development that have been made last ten years. We emphasize that Philosophy of Information is a huge and vibrant research field, with its origins dating before Open Problems, and its domains extending even outside their scope. In this paper, we have been able only to sketch some of the developments during the past ten years. Our hope is that, even if fragmentary, this review may serve as a contribution to the effort of understanding the present state of the art and the paths of development of Philosophy of Information as seen through the lens of Open Problems. (shrink)