Algorithms can never be ethical – attempts fail to account for the relations in which they first emerged
Amoore 20 (Louise Amoore, Professor of Political Geography at Durham University, “Cloud Ethics: Algorithms and the Attributes of Ourselves and Others”, Duke University Press, 8/30/20, https://doi.org/10.1080/1369118X.2020.1800782) // EL
A cloud ethics acknowledges that algorithms contain, within their spatial arrangements, multiple potentials for cruelties, surprises, violences, joys, distillations of racism and prejudice, injustices, probabilities, discrimination, and chance. Indeed, many of the features that some would like to excise from the algorithm—bias, assumptions, weights—are routes into opening up their politics. Algorithms come to act in the world precisely in and through the relations of selves to selves, and selves to others, as these relations are manifest in the clusters and attributes of data. To learn from relations of selves and others, the algorithm must already be replete with values, thresholds, assumptions, probability weightings, and bias. In a real sense, an algorithm must necessarily discriminate to have any traction in the world. The very essence of algorithms is that they afford greater degrees of recognition and value to some features of a scene than they do to others. In so doing, algorithms generate themselves as ethicopolitical beings in the world. If to have ethics is not merely to have a code prohibiting, for example, bias or assumptions, but to work on oneself via relations, then the ethicopolitics of algorithms involves investigations of how they learn to recognize and to act, how they extract assumptions from data relations, and how they learn what ought to be from relations with other humans and algorithms. To be clear, the cloud ethics I propose here does not belong to an episteme of accountability, transparency, and legibility, but on the contrary begins with the opacity, partiality, and illegibility of all forms of giving an account, human and algorithmic. To advance a cloud ethics is to engage the ungrounded politics of all forms of ethical relations. The significant new ethical challenges that algorithms seem to present to society actually manifest novel features of some profoundly old problems of the grounds for ethical action. As Judith Butler explains in her Spinoza lectures, the demand to give an account of oneself will always fall short, for “I cannot give an account of myself without accounting for the conditions under which I emerge.”19 If one assumes that the determination of an unequivocal I who acts is a necessary precondition of ethics, as Butler cautions, then this identifiable self is “dispossessed” by the condition of its emergence in relation to others. For Butler, this persistent failure to give a clear-sighted account does not mark the limit point of ethics. On the contrary, the opaque and unknowable nature of making all kinds of acting subjects is the condition of possibility of having an ethicopolitical life.20 In short, and in contrast to the equation of ethics with transparency and disclosure, ethical responsibility is sustained by conditions of partiality and opacity. My notion of a cloud ethics extends the opacity of the human subject, envisaging a plurality of venues for ethical responsibility in which all selves—human and algorithmic—proceed from their illegibility. The apparent opacity and illegibility of the algorithm should not pose an entirely new problem for human ethics, for the difficulty of locating clear-sighted action was already present. The I who forms the ethical relation was always in question and is now, with algorithms, in question in new ways. Though the mathematical propositions of algorithms cannot be made fully legible, or rendered accountable, they can be called to give accounts of the conditions of their emergence. These conditions include some relations that are identifiably between humans and algorithms—such as the selection and labeling of training data, the setting of target outputs, or the editing of code “in the wild,” for example—but others still are relations of algorithms to another algorithm, such as a classifier supplying the training data from which a neural network will learn. In all such instances of iterative learning, the significant point is that the conditions of an algorithm’s emergence—a composite of human-algorithm relations—are venues for ethicopolitics.