The notion of “responsibility gap” with artificial intelligence (AI) was originally introduced in the philosophical debate to indicate the concern that “learning automata” may make more difficult or impossible to attribute moral culpability to persons for untoward events. Building on literature in moral and legal philosophy, and ethics of technology, the paper proposes a broader and more comprehensive analysis of the responsibility gap. The responsibility gap, it is argued, is not one problem but a set of at least four interconnected (...) problems – gaps in culpability, moral and public accountability, active responsibility—caused by different sources, some technical, other organisational, legal, ethical, and societal. Responsibility gaps may also happen with non-learning systems. The paper clarifies which aspect of AI may cause which gap in which form of responsibility, and why each of these gaps matter. It proposes a critical review of partial and non-satisfactory attempts to address the responsibility gap: those which present it as a new and intractable problem (“fatalism”), those which dismiss it as a false problem (“deflationism”), and those which reduce it to only one of its dimensions or sources and/or present it as a problem that can be solved by simply introducing new technical and/or legal tools (“solutionism”). The paper also outlines a more comprehensive approach to address the responsibility gaps with AI in their entirety, based on the idea of designing socio-technical systems for “meaningful human control", that is systems aligned with the relevant human reasons and capacities. (shrink)

Is there a distinctively epistemic kind of blame? It has become commonplace for epistemologists to talk about epistemic blame, and to rely on this notion for theoretical purposes. But not everyone is convinced. Some of the most compelling reasons for skepticism about epistemic blame focus on disanologies, or asymmetries, between the moral and epistemic domains. In this paper, I defend the idea that there is a distinctively epistemic kind of blame. I do so primarily by developing an account of the (...) nature of epistemic blame. My account draws on a prominent line of theorizing in moral philosophy that ties blame to our relationships with one another. I argue that with my account of epistemic blame on hand, the most compelling worries about epistemic blame can be deflated. There is a distinctively epistemic kind of blame. (shrink)

Promising is clearly a social practice or convention. By uttering the formula, “I hereby promise to do X,” we can raise in others the expectation that we will in fact do X. But this succeeds only because there is a social practice that consists (inter alia) in a disposition on the part of promisers to do what they promise, and an expectation on the part of promisees that promisers will so behave. It is equally clear that, barring special circumstances of (...) some kind, it is morally wrong for promisers to fail to do what they have promised to do. What is perhaps less clear is how the moral wrongness that is involved when promises are broken is related to the social practice that makes promising possible in the first place. (shrink)

Some of the systems used in natural language generation (NLG), a branch of applied computational linguistics, have the capacity to create or assemble somewhat original messages adapted to new contexts. In this paper, taking Bernard Williams’ account of assertion by machines as a starting point, I argue that NLG systems meet the criteria for being speech actants to a substantial degree. They are capable of authoring original messages, and can even simulate illocutionary force and speaker meaning. Background intelligence embedded in (...) their datasets enhances these speech capacities. Although there is an open question about who is ultimately responsible for their speech, if anybody, we can settle this question by using the notion of proxy speech, in which responsibility for artificial speech acts is assigned legally or conventionally to an entity separate from the speech actant. (shrink)

Assertions are praised as true, informative, relevant, sincere, warranted, well-phrased, or polite. They are criticized as false, uninformative, irrelevant, insincere, unwarranted, ill-phrased, or rude. Sometimes they deserve such praise or criticism. If any respect in which performances of an act can deserve praise or criticism is a norm for that act, then the speech act of assertion has many norms. So has almost any act; jumps can deserve praise as long or brave, criticism as short or cowardly. But it is (...) natural to suppose that some norms are more intimately connected to the nature of asserting than any norm is to the nature of jumping. One might suppose, for example, that someone who knowingly asserts a falsehood has thereby broken a rule of assertion, much as if he had broken a rule of a game; he has cheated. On this view, the speech act, like a game and unlike the act of jumping, is constituted by rules. Thus, not all norms for assertion are on a par. Norms such as relevance, good phrasing, and politeness are just applications of more general cognitive or social norms to the specific act of assertion. Perhaps the norm of informativeness results from a more complex interaction between a general norm of cooperativeness and the nature of assertion as a source of information. But on this view, not all norms for assertion derive from more general norms; otherwise nothing would differentiate it from other speech acts. (shrink)

The power to promise is morally fundamental and does not, at its foundation, derive from moral principles that govern our use of conventions. Of course, many features of promising have conventional components—including which words, gestures, or conditions of silence create commitments. What is really at issue between conventionalists and nonconventionalists is whether the basic moral relation of promissory commitment derives from the moral principles that govern our use of social conventions. Other nonconventionalist accounts make problematic concessions to the conventionalist's core (...) instincts, including embracing: the view that binding promises must involve the promisee's belief that performance will occur; the view that through the promise, the promisee and promisor create a shared end; and the tendency to take promises between strangers, rather than intimates, as the prototypes to which a satisfactory account must answer. I argue against these positions and then pursue an account that finds its motivation in their rejection. My main claim is: the power to make promises, and other related forms of commitment, is an integral part of the ability to engage in special relationships in a morally good way. The argument proceeds by examining what would be missing, morally, from intimate relationships if we lacked this power. (shrink)

This paper aims to identify the constitutive rule of assertion, conceived by analogy with the rules of a game. That assertion has such rules is by no means obvious; perhaps it is more like a natural phenomenon than it seems. One way to find out is by supposing that it has such rules, in order to see where the hypothesis leads and what it explains. That will be done here. The hypothesis is not perfectly clear, of course, but we have (...) at least a crude conception of constitutive rules, which we may refine as we elaborate the hypothesis. Although no attempt will be made here to define ‘rule’, some remarks on constitutive rules will focus the discussion. (shrink)

The doyen of living English philosophers, by these reflections, took hold of and changed the outlook of a good many other philosophers, if not quite enough. He did so, essentially, by assuming that talk of freedom and responsibility is talk not of facts or truths, in a certain sense, but of our attitudes. His more explicit concern was to look again at the question of whether determinism and freedom are consistent with one another -- by shifting attention to certain personal (...) rather than moral attitudes, first of all gratitude and resentment. In the end, he arrived at a kind of Compatibilist or, as he says, Optimist conclusion. That is no doubt a recommendation but not the largest recommendation of this splendidly rich piece of philosophy. (shrink)

Intuitively, proper referential extensions of psychological and moral terms exclude artifacts. Yet ordinary speakers commonly treat AI robots as moral patients and use psychological terms to explain their behavior. This paper examines whether this referential shift from the human domain to the AI domain entails semantic changes: do ordinary speakers literally consider AI robots to be psychological or moral beings? Three non-literalist accounts for semantic changes concerning psychological and moral terms used in the AI domain will be discussed: the technical (...) view (ordinary speakers express technical senses), the habit view (ordinary speakers subconsciously express ingrained social habits), and the emotion view (ordinary speakers express their own affective empathetic emotional states). I discuss whether these non-literalist accounts accommodate the results of relevant empirical experiments. The non-literalist accounts are shown to be implausible with respect to the ordinary use of agency-terms (e.g., “believe,” “know,” “decide,” etc.), and therefore I conclude that the concepts ordinary speakers express by agency-terms in reference to AI robots are similar to the concepts they express when applying the same terms to humans. When ordinary speakers extend emotion-terms and/or moral-patiency-terms to AI robots, however, I argue that semantic changes have taken place because ordinary speakers are in fact referring to their own affective empathetic emotional states rather than AI robots. This argument suggests that the judgments made by ordinary speakers regarding the proper referential extensions of emotion-terms and moral-patiency-terms are fallacious. (shrink)

The advent of intelligent artificial systems has sparked a dispute about the question of who is responsible when such a system causes a harmful outcome. This paper champions the idea that this dispute should be approached as a conceptual engineering problem. Towards this claim, the paper first argues that the dispute about the responsibility gap problem is in part a conceptual dispute about the content of responsibility and related concepts. The paper then argues that the way forward is to evaluate (...) the conceptual choices we have, in the light of a systematic understanding of why the concept is important in the first place—in short, the way forward is to engage in conceptual engineering. The paper then illustrates what approaching the responsibility gap problem as a conceptual engineering problem looks like. It outlines argumentative pathways out of the responsibility gap problem and relates these to existing contributions to the dispute. (shrink)

Algorithms silently structure our lives. Algorithms can determine whether someone is hired, promoted, offered a loan, or provided housing as well as determine which political ads and news articles consumers see. Yet, the responsibility for algorithms in these important decisions is not clear. This article identifies whether developers have a responsibility for their algorithms later in use, what those firms are responsible for, and the normative grounding for that responsibility. I conceptualize algorithms as value-laden, rather than neutral, in that algorithms (...) create moral consequences, reinforce or undercut ethical principles, and enable or diminish stakeholder rights and dignity. In addition, algorithms are an important actor in ethical decisions and influence the delegation of roles and responsibilities within these decisions. As such, firms should be responsible not only for the value-laden-ness of an algorithm but also for designing who-does-what within the algorithmic decision. As such, firms developing algorithms are accountable for designing how large a role individual will be permitted to take in the subsequent algorithmic decision. Counter to current arguments, I find that if an algorithm is designed to preclude individuals from taking responsibility within a decision, then the designer of the algorithm should be held accountable for the ethical implications of the algorithm in use. (shrink)

This book is the culmination of almost forty years of writing and thinking about speech acts and the use theory of meaning. Chapter 1 sets out and defends a version of the Austin-Searle trichotomy of a sentential act, i.e., uttering a sentence or surrogate, an illocutionary act, i.e., uttering a sentence with a certain "content" as reported by indirect speech, and a perlocutionary act, i.e., producing an effect on an audience by an utterance. Chapter 2 poses the question: what condition (...) C will turn an SA into an IA: F? Perhaps adding a perlocutionary intention to the effect that "U asserts that p only if U utters p with the intention of getting H to believe that p" will meet condition C. Alston argues, contra Schiffer, that it will not. Chapter 3 objects to Searle's analysis of sincere and non-defective promising, on the one hand, because there are too many ways a promise can be defective and, on the other hand, because some of Searle's conditions are not necessary for promising simpliciter. Alston proposes. (shrink)