Toward a Science of Intent: Closure Gaps and Delegation Envelopes for Open-World AI Agents

TL;DR

This paper introduces a formal framework for understanding and improving the deployment of open-world AI agents by analyzing intent, closure gaps, and delegation regions to enhance verification and execution.

Contribution

It formalizes the concepts of closure gaps and delegation envelopes, providing metrics and distinctions to better deploy and verify open-world AI systems.

Findings

Introduces the concept of closure-gap vectors for open-world AI.

Defines delegation envelopes as pre-authorized action regions.

Proposes benchmark metrics for closure interventions versus inference search.

Abstract

Recent work has framed intelligence in verifiable tasks as reducing time-to-solution through learned structure and test-time search, while systems work has explored learned runtimes in which computation, memory and I/O migrate into model state. These perspectives do not explain why capable models remain difficult to deploy in open institutions. We propose intent compilation: the transformation of partially specified human purpose into inspectable artifacts that bind execution. The relevant deployment distinction is closed-world solver versus open-world agent. In closed worlds, a checker is largely given; in open worlds, verification is distributed across semantic, evidentiary, procedural and institutional dimensions. Weformalize this residual openness as a closure-gap vector, define delegation envelopes as pre-authorized regions of action space, distinguish misclosure from undersearch, and outline benchmark metrics for testing when closure interventions outperform additional inference-time search.