Recently, there has been considerable interest in large language models: machine learning systems which produce human-like text and dialogue. Applications of these systems have been plagued by persistent inaccuracies in their output; these are often called “AI hallucinations”. We argue that these falsehoods, and the overall activity of large language models, is better understood as bullshit in the sense explored by Frankfurt (On Bullshit, Princeton, 2005): the models are in an important way indifferent to the truth of their outputs. We (...) distinguish two ways in which the models can be said to be bullshitters, and argue that they clearly meet at least one of these definitions. We further argue that describing AI misrepresentations as bullshit is both a more useful and more accurate way of predicting and discussing the behaviour of these systems. (shrink)

Accounts of autonomy which acknowledge the importance of non-domination – that is, of being structurally protected against arbitrary interference with one's life – face an apparent problem with regards to intimate relationships. By their very nature, such relations open us up to psychological and material suffering that would not be possible absent the particular relationship; even worse, from the non-domination point of view, is that this vulnerability seems to be structural in a way exactly analogous to workplace or social domination. (...) If being powerless to prevent an employer causing me harm constitutes domination at work, then what relevant differences can support the intuition that being powerless to prevent my partner causing me comparable pain is not autonomy-hostile? I argue for the reassuring view that the obligations and possibility of pain arising from such relations aren't necessarily dominating; they would be so only if we believed that any obligation we have not explicitly agreed to is a restriction on our autonomy, and that is false. I conclude with a note of caution: even though intimate relations aren't necessarily dominating, they will often be contingently so if they take place in a wider social context of domination – such as that which we currently inhabit. (shrink)

Anarchists standardly critique the state for being illegitimate, and for being dominating in some sense. Often these criticisms come as a bundle: the state is illegitimate because it is dominating. But there are various stories we might tell about the connection between the two; domination makes consent impossible, domination means that the state fails to meet its own justification for existing (or for claiming authority), and so on. I suggest that we should sidestep concerns about consent: in part because it (...) seems possible for people to genuinely consent to something which is nonetheless impermissible, but also because many anarchists offer views of political organisation which very clearly involve some coercion of rule-breakers – Malatesta, for example, explicitly endorses the exclusion of persistent non-compliers. Rather, I argue that the best way to understand why the state is illegitimate – and a way which is immanent in many anarchist critiques – is to combine something like a Pettitian analysis of domination with a requirement that genuine authorities (and authoritative imperatives) recognise us as agents and treat us accordingly. On this view, states are necessarily dominating in virtue of two key features: dominating in virtue of being able to exercise arbitrary power over agents, necessarily so because part of what it is to be a state is to be an institution which exercises (and refrains from exercising) power in virtue of morally arbitrary features. As a paradigm case, states must be able to exclude non-citizens, on pain of losing their successful claim to a monopoly on legitimate force. But to recognise somebody as an agent is inconsistent with denying them access to territory or services on the basis of something as arbitrary as their place of birth. This analysis also explains why it’s possible to hold that all states are illegitimate, but some are nonetheless “better” than others – depending on the structure of a state and the power held by its citizens, we may be more or less vulnerable to arbitrary interference, and hence be more or less dominated, in one state than another. (shrink)

Relational accounts of autonomy such as those advanced by Rebekah Johnston (2017. “Personal Autonomy, Social Identity, and Oppressive Social Contexts.” Hypatia 32 (2): 312–28) and particularly Marina Oshana (1998; “Personal Autonomy and Society.” Journal of Social Philosophy 29 (1): 81–102. 2006; Personal Autonomy in Society. Aldershot: Ashgate; 2015. “Is Social-Relational Autonomy a Plausible Ideal?” In Personal Autonomy and Social Oppression: Philosophical Perspectives, edited by M. Oshana. New York: Routledge) are often thought to be distinctively vulnerable to paternalist and perfectionist objections, (...) as well as implying in a troublesome and/or implausible way that many of us are much less autonomous than is standardly thought. I argue that although constitutively relational accounts are indeed perfectionist, they are so in such a “thin” way that we should not be concerned by it; that they are not paternalist (and indeed are necessarily hostile to the sorts of power relations necessary for paternalism); and that the charge that relational accounts demand an unattractively high bar for autonomy is misdirected – we really do lack autonomy in perhaps surprising but plausible ways and for the account to imply this is a problem for the world rather than the theory. (shrink)

In this paper, I focus on recognition respect as a component of Anderson’s democratic equality – specifically, how it places certain requirements on the way political institutions such as states treat both citizens and non-citizens. I argue for two claims: that recognition respect is a plausible political (as well as ethical) value, and that it should be understood in large part as a matter of an agent’s material relational standing rather than as their merely being regarded in a certain way (...) by others. In particular, I argue that the second-personal emphasis of recognition respect, the conceptual requirement that recognising somebody as an agent involves recognising them as somebody to whom you are in principle accountable, requires that agents actually be able to hold you to account rather than merely that you see them as being the right sort of being to do so. In turn, this implies that recognition respect requires non-vulnerability. Together, these claims go some way towards further fleshing out democratic equality as a philosophical ideal, and determining what implications this ideal has for matters of theory and policy. (shrink)

The problem of the ‘mountain man’, the caricature of self-sufficiency and individualism, is not a new one for autonomy theorists. It seems plausible that there is genuine value in self-direction according to one’s deeply-held principles. If autonomy involves something like this, then anyone concerned with autonomy as a social rather than individualistic phenomenon must explain what the mountain man gets wrong when he denies that his autonomy admits of being placed under obligations to others. In particular, the mountain man challenges (...) autonomy-minded social anarchists: if his denial of legitimate non-voluntary obligations is correct, then it is not just the state we should reject, but any organising body with coercive powers. This may be consistent with individualist anarchism or right-libertarianism, but it sits ill with the social anarchist intuition that we can have genuine political obligations. My thesis addresses this problem in three stages. First, I argue for a functional analysis of authority and autonomy: the concepts are not pre-existing “immovable objects”, but rather are defined by the role that they are intended to play in our discourse. I suggest that we need a concept of political or institutional authority in order to resolve co-ordination problems and pursue collaborative social goods, and a concept of autonomy to explain when and why self-direction is valuable. Second, I defend a social-relational conception of autonomy. The autonomous agent is powerful and authoritative, where this power and authority is in large part constituted, rather than merely affected, by the social structures and relations that we stand in. We are powerful and authoritative, I argue, when we stand in relations of non-domination: we are not vulnerable to arbitrary interference in our lives, and this non-vulnerability is defended in virtue of recognition respect for us as agents. There are two important implications of this account: that autonomy comes with a built-in equality condition whereby everybody’s autonomy is threatened if anybody’s is, and that there is no principled distinction to be drawn between ‘personal’ and ‘political’ autonomy. In the last three chapters, I suggest an autonomy-justified conception of authority. I argue for autonomy as the crucial collaborative good which authoritative institutions help us to pursue, and suggest that such institutions may legitimately claim authority if they act or effect actions in ways which are likely to promote or defend autonomy-constituting relations, and act or effect actions in ways consistent with maximal equal autonomy. Finally, I return to the anarchist argument, showing that while my accounts of autonomy and authority give us a plausible picture of how autonomy is compatible with genuinely authoritative institutions, this picture still has no room for the state. (shrink)