Theoretical analysis outlines governance requirements for transferable judgment across human-AI systems, highlighting the need to preserve underlying rationales alongside conclusions.
Consequential decisions increasingly move across boundaries: from one human to another, from organizations to automated systems, from artificial intelligence systems to human operators, and between interconnected computational systems. Existing approaches to provenance and accountability can establish where a decision originated and how it was produced, yet provenance alone does not guarantee that the recipient possesses sufficient information to independently evaluate whether the decision remains valid. This paper introduces decision durability as a governance property of transferable judgment. A durable decision is not one whose conclusion merely survives transmission; it is one whose underlying rationale, evidentiary basis, uncertainty, authority boundaries, assumptions, and conditions for reconsideration survive sufficiently for a subsequent actor to reconstruct, challenge, validate, reject, or reopen the decision. We propose a Decision Continuity Record for preserving these elements across human–AI andsystem-to-system handoffs and define five corresponding governance requirements: semantic intelligibility, rationale recoverability, challengeability, authority traceability, and reconsiderability. The framework distinguishes decision transfer from authority transfer and argues that executing inherited conclusions without recoverable judgment creates an accountability and governance gap. In high-stakes automated environments, preserving the answer is therefore insufficient: what must survive is the reason—and enough evidence about that reason for another actor to determine whether it still deserves authority. Central Thesis A decision is not governably transferable merely because its conclusion survives. It is transferable only when enough of its original judgment survives for the receiving actor to reconstruct, question, validate, reject, or reopen that conclusion.
No takes yet. Share an insight, caveat, or question.
Alexander Jorge Cisneros (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: