We argue that two long-term goals of AI research stand in tension with one another. The first involves creating AI that is safe, where this is understood as solving the problem of value alignment. The second involves creating artificial general intelligence, meaning AI that operates at or beyond human capacity across all or many intellectual domains. Our argument focuses on the human capacity to make what we call “existential choices”, choices that transform who we are as persons, including transforming what (...) we most deeply value or desire. It is a capacity for a kind of value misalignment, in that the values held prior to making such choices can be significantly different from (misaligned with) the values held after making them. Because of the connection to existentialist philosophers who highlight these choices, we call the resulting form of risk “existentialist risk.” It is, roughly, the risk that results from AI taking an active role in authoring its own values rather than passively going along with the values given to it. On our view, human-like intelligence requires a human-like capacity for value misalignment, which is in tension with the possibility of guaranteeing value alignment between AI and humans. (shrink)

Can the question "Why do what morality requires?" be answered in such a way that anyone regardless of their desires or interests has reason to be moral? One strategy for answering this question appeals to constitutive arguments. In general, constitutive arguments attempt to establish the normativity of rational requirements by pointing out that we are already committed to them insofar as we are believers or agents. This study is concerned with the general prospects for such arguments. It starts by explaining (...) the general constitutive argument strategy, followed by an examination of constitutive arguments that have been given regarding theoretical reason and the instrumental principle in practical reason, and concluding with a discussion of some challenges to constitutive arguments in moral philosophy and some possible responses to these challenges. (shrink)

Human beings are internally inconsistent in various ways. One way to develop this thought involves using the language of value alignment: the values we hold are not always aligned with our behavior, and are not always aligned with each other. Because of this self-misalignment, there is room for potential projects of human enhancement that involve achieving a greater degree of value alignment than we presently have. Relatedly, discussions of AI ethics sometimes focus on what is known as the value alignment (...) problem, the challenge of how to build AI that acts in accordance with our human values. We argue that there is an especially close connection between solving the value alignment problem in AI ethics and using AI to pursue certain forms of human enhancement. But in addition, we also argue that there are important limits to what kinds of human enhancement can be pursued in this way, because some forms of human enhancement—namely moral revolutions—involve a kind of value misalignment rather than alignment. (shrink)

This paper develops an account of existential choices and their role in practical reasoning. In contrast to other views that attempt to make sense of existential choices as a type of rational choice, the proposed account takes them to be choices among the normative outlooks that determine the reasons we have, and as such are nonrational. According to the argument in the paper, existential choices bring to light a feature of all choices, that they are made against the backdrop of (...) a normative outlook, which grounds the rationality of the choice but is not itself rationally determined. (shrink)

Reasons internalism holds that reasons for action contain an essential connection with motivation. I defend an account of reasons internalism based on the advisor model. The advisor model provides an account of reasons for action in terms of the advice of a more rational version of the agent. Contrary to Pettit and Smith's proposal and responding to Sobel's and Johnson's objections, I argue that the advisor model can provide an account of internal reasons and that it is too caught up (...) in the psychology of the actual agent to be able to account for anything other than internal reasons. (shrink)

This paper argues for interpreting Nietzsche along the lines of a self-constitution view. According to the self-constitution view, a person is a kind of creation: we constitute our selves throughout our lives. The self-constitution view may take more than one form: on the narrative version, the self is like a story, while on the Kantian version, the self is a set of principles or commitments. Taking Marya Schechtman’s and Christine Korsgaard’s accounts as paradigmatic, I take the self-constitution view to emphasize (...) practical considerations and the first person point of view and to conceive of the self as active in self-creation. The interpretation I offer can make sense of Nietzsche’s remarks about self-creation and of many of Nietzsche’s remarks about the self that would otherwise seem contradictory. In particular, Nietzsche’s anti-metaphysical remarks about the self fit well with the self-constitution view as long as they are understood as theoretical claims that do not undermine the importance of the practical point of view. (shrink)

Often in romantic relationships, people want partners whose moral, political, and religious values align with their own. This article connects this point to the project of value alignment in AI research, in which researchers aim to design artificial intelligence systems that behave in ways that are in accordance with human values. There are good reasons to pursue the project of value alignment in connection with romantic chatbots—artificial intelligence systems designed to allow human users to simulate or replicate elements of romantic (...) relationships that ordinarily exist only between human beings. However, this paper argues that certain types of value alignment in such systems would preclude them from entering into authentic loving relationships. Some important features of authentic love require a capacity for value misalignment between romantic partners. A romantic chatbot whose values were locked in to match those of its human user only allows for a kind of inauthentic love that does not treat the artificial partner as a person in their own right but at most makes them into a kind of tool. The arguments presented draw on the existentialist feminist views of Simone de Beauvoir and her understanding of authentic romantic love. De Beauvoir held that inauthentic loving relationships entail that certain interactions between partners lack the significance or meaning they would have in cases of authentic love, that authentic love promotes self-knowledge and reflection, and that authentic love allows individuals to accomplish projects they could not accomplish on their own, including projects of self-transformation. The article explores ways in which the project of value alignment involving romantic chatbots interferes with such things. (shrink)

Constitutive arguments for the principles of practical reason attempt to justify normative requirements by claiming that we already accept them in so far as we are believers or agents. In two constitutive arguments for the requirement that we must will universally, Korsgaard attempts first to arrive at the requirement that we will universally from observations about the causality of the will, and secondly to establish that willing universally is constitutive of having a self. Some rational requirements may be established by (...) some version of this second argument, but the strategy does not seem promising when it comes to establishing the requirement that we will universally. I draw on the discussion of Korsgaard to highlight a challenge facing constitutive arguments in general. (shrink)

ABSTRACT Following Robert C. Solomon's Living with Nietzsche, I defend an interpretation of Nietzsche's views about freedom that are in line with the existentialist notion of self-creation. Given Nietzsche's emphasis on the limitations on human freedom, his critique of the notion of causa sui, and his critique of morality for relying on the assumption that we have free will, it may be surprising that he could be taken seriously as an existentialist—existentialism characteristically takes freedom and self-creation to be central to (...) the human condition. Nietzsche does not endorse a radical notion of freedom; he is rather emphatically critical of any such notion. However, I argue that he does have room for a certain kind of freedom and self-creation that supports Solomon's characterization of him as an existentialist. (shrink)

This paper develops a philosophical account of environmental racism and explains why having such an account is worthwhile. After reviewing some data points and common uses of the term linking environmental racism to the distribution of environmental burdens by race, I argue that environmental racism should be understood as referring to an unequal distribution caused by a history of racism. Environmental racism is thus analyzed in terms of two conditions: first, that environmental burdens and benefits be distributed according to race, (...) and second, that this distribution be partly caused by a history of racism, where this causal claim is understood counterfactually. The account entails that environmental racism is a derivative form of racism – it depends on other kinds of racism to exist – while being compatible with a variety of views about the more basic forms of racism. Given the proposed framework, I consider the kind of evidence needed to establish whether a given case is one of environmental racism and make the case for environmental racism being a concept worth focusing on as it can highlight a type of injustice that may otherwise go unnoticed. (shrink)

This Article explores the possibility of having ethical artificial intelligence. It argues that we face a dilemma in trying to develop artificial intelligence that is ethical: either we have to be able to codify ethics as a set of rules or we have to value a machine’s ability to make ethical mistakes so that it can learn ethics like children do. Neither path seems very promising, though perhaps by thinking about the difficulties with each we may come to a better (...) understanding of artificial intelligence and ourselves. (shrink)