Ethics

Ethics is the study of how we ought to act and what we owe one another. Its traditional questions (what makes acts right, what makes lives go well, what we may demand of each other) remain the questions. What has changed is how often the answer now has to be given about an institution acting through a system rather than a person acting alone.

I do not think ethics can be done from outside such systems. A concept that works for a clinician and a competent adult patient may not survive contact with a biometric database that decides, at the point of transaction, whether a person counts. Much of my work here is an argument that the choice of normative category matters as much as the verdict. The data-protection regimes that govern algorithmic systems, and the remedial debate around their failures, make consent the load-bearing concept of legitimacy. For a class of systems that produce the standing of those they govern, I argue that consent is not merely strained but conceptually misplaced.

What takes its place is obligation. Onora O’Neill has long argued that rights-talk is recipient-oriented and leaves unspecified who must do what, while obligations are agent-oriented. I follow her in shifting the question from whether the subject has consented to what the institution owes her, and the answer I defend is the Constitutive Obligation Principle: non-waivable obligations of legibility, non-abandonment and remediation.

Ethics also has an epistemic side. What we owe one another as knowers (honesty, care with evidence, a willingness to give grounds) is part of why testimony works at all, and part of why it matters whether a machine’s sentence is the kind of thing anyone stands behind. Often no one does. Responsibility for putting a system into the world is not authorship of its sentences.