Human Authority in Non-Voluntary Settings: Government Use of Artificial Intelligence

This exploratory brief examines distinctive questions that arise when governments deploy artificial intelligence or analogous advanced technologies. The paper does not address whether artificial intelligence is, as a technology, benign, benevolent, or dangerous. It is concerned with government use of such capabilities under conditions of non-voluntary authority. The central claim is that government acts and determinations that bind natural persons through non-voluntary authority necessarily involve value judgments about ends, trade-offs, acceptable risks, and prioritization. Those are not mere technical problems to be optimized; they are the very substance of governing. The legitimacy of public authority rests on binding acts and determinations remaining with natural persons who can stand behind, revise, defend, and answer for them. Government may use artificial intelligence to inform that work and to perform additive functions, but binding acts and determinations must not be left with the tool. When the binding act is left with the tool, two institutional failures follow: weakened or absent accountability, and the functional circumvention of rights. Because new technical capabilities and misuses can be subtle, unforeseeable, or invisible, a provisional negative-rights orientation (categories of illegitimate power rather than approved uses) is offered. How government uses artificial intelligence is a different question from how private firms develop technology in voluntary markets. Only the first is taken up here.