THE GOVERNED MACHINE: PRACTICAL WISDOM AND THE BOUNDARIES OF DELEGATION TO ARTIFICIAL INTELLIGENCE

When work passes from people to intelligent systems, what principle decides where the transfer must stop? I answer with a division internal to deliberation itself. One kind of reasoning operates under conditions it treats as fixed and looks for the strongest available move; another kind puts those very conditions on trial, asking whether a different arrangement of restrictions and opportunities would serve a life better. Machines belong in the first kind and are excluded from the second, since trying the conditions of one’s existence calls on an interlocking set of humanly grounded powers: the perception of what is at stake, a story one actually lives, an identity exposed in the choosing, and emotion working as insight. A modest formalism of frames and reachability carries the analysis, and its ethical yield is constructive: delegation, once it is properly ruled, opens lives that overburdened agents could not otherwise reach. Along the way the argument recovers, strictly for insensate instruments, the defensible skeleton inside an Aristotelian doctrine whose historical use against persons was inexcusable, and it turns the strongest findings of technology studies, that artefacts carry values and that technique is steeped in judgement, from objections into premises. A luthier, a family carer, and a university examiner supply the tests, and the conclusion issues in four demands on system design concerning authority over objectives, economy of attention, the openness of embedded values, and deference to human judgement.