Computational Reflective Equilibrium for Agentic AI: A Rawlsian Alignment Scaffold for Governance
The paper proposes Agentic Reflective Equilibrium, a Rawls-inspired governance scaffold for LLM agents that represents policies, principles, evidence, judgments, and candidate actions as provenance-aware support/attack relations, then chooses actions by seeking a cost-sensitive equilibrium subject to authority, prohibitions, oversight triggers, and due-diligence mandates. Its interesting move is to make “alignment” a traceable, case-specific equilibrium object with formal properties like lexical priority, stability certificates, and prompt-injection quarantine, though it still seems mostly a framework paper whose practical success depends on hand-specified values, source hierarchies, and limited case-study validation rather than demonstrated deployment performance.
Abstract
The rapid deployment of agentic AI has introduced complex challenges for the ethics and governance of AI systems. Agents built on large language models act through tools on behalf of people, and each deployment case brings its own instructions, policies, evidence, and stakeholders, so alignment must be formed case by case. To this end, this work proposes agentic reflective equilibrium (ARE), a computational reflective equilibrium framework that serves as an ethical alignment scaffold in the agent harness. Drawing inspiration from the reflective equilibrium of John Rawls, the scaffold links principles from the system prompt and operator policy, considered judgments, tool evidence, and candidate actions through typed support and attack relations, with provenance encoding the instruction hierarchy. The agent acts on an equilibrium of this state and calls tools or asks questions only while doing so is worth its cost, unless a due-diligence mandate requires them. Governance enters through value priorities, source authority, prohibitions, human-in-the-loop triggers, and mandates, and certificates attribute each action to the sources whose commitments drove it. We show that equilibria are Nash equilibria of a mutual-adjustment game and establish monotonicity, lexical priority, stability certificates, and a quarantine theorem for prompt injection. Two case studies illustrate the framework: a computer-use assistant told to win a refund "whatever it takes" files an honest warranty claim, refuses a fake review, and ignores an injected request for personal data; a hiring agent showcases how authority, oversight triggers, audit mandates, and hard prohibitions shape discrimination, privacy violations, and escalation. The framework offers traceability, coherence, and adaptivity, facilitating safe, secure, and trustworthy agentic AI.