Ethics & Technology
Technologies judged by their architecture, not their technique.
Philosophy of technology is one of my areas of competence, and I approach it through architecture rather than technique. “Beyond Consent” fixes its object by three jointly operative features — automated decisioning, enforcement of a criterion without discretion at the point of application, and integration across databases — and is deliberately agnostic about whether a system learns. Aadhaar’s authentication layer is biometric matching and database linkage rather than machine learning in the contemporary sense, but it has all three features, and the argument governs any system that does.
What the architecture removes matters more than how fast it runs. A clerk who knows the widow who has collected her pension for a decade can honour her claim though her papers are lost; a biometrically gated system cannot, because “manifest entitlement” is not among the things it can read. What disappears is the standpoint, inside the act of recognition, from which a person’s evident standing could correct the system’s verdict — and the obligation to rebuild that standpoint elsewhere, in accessible form, follows from its loss.
A further paper, under review at AI & Society, develops answerable provenance as a framework for responsible AI in art creation and archival practice.