Judgment Delegation as a Computational Primitive: Foundations for Autonomy Engineering

Computing has long executed decisions without accounting for how those decisions are made. As modern AI systems increasingly act under uncertainty—interpreting intent, framing problems, and advancing action without full specification—judgment delegation, not execution, has become the true organizing force of autonomy. This paper names judgment delegation as a foundational computational primitive and argues that many defining features of contemporary AI systems—including their perceived intelligence and their structural influence—emerge from the implicit delegation of judgment rather than from autonomous reasoning. By making judgment delegation explicit as an object of analysis and design, the paper establishes Autonomy Engineering as a new field concerned with how judgment authority is allocated, exercised, and governed in autonomous and semi-autonomous systems.