Artificial intelligence recommendations increasingly enter organizational decisions that remain formally assigned to human actors. Yet prevailing accounts often treat reliance as a problem of accuracy, trust, adoption, or interface design without examining how machine advice may reshape who performs judgment, who remains psychologically connected to consequences, and who accepts responsibility afterward. This Original Propositional Model Article develops a proposed Judgment-and-Responsibility Model integrating decision deference, judgment displacement, moral distance, and responsibility avoidance. The model distinguishes ordinary advice use from deference, and deference from delegation, arguing that responsibility becomes vulnerable not whenever advice is accepted, but when independent appraisal and justificatory ownership are displaced. It further proposes that judgment displacement may increase moral distance by reducing the salience of affected persons, contextual particulars, or downstream consequences, thereby creating conditions in which responsibility can be avoided or diffusely assigned. These relationships are conditional on task characteristics, expertise, verification difficulty, organizational authority, consequence visibility, work design, and opportunities to contest or review recommendations. The article advances propositions and organizational safeguards centered on independent justification, traceability, consequence reconnection, explicit responsibility assignment, and meaningful human control. The synthesis is non-empirical and does not establish a validated causal model, universal prescription, or implementation-ready system. Its principal contribution is to connect scholarship on human–AI reliance with organizational responsibility while specifying rival explanations, multilevel boundaries, and empirical designs capable of refining, challenging, or rejecting the proposed relationships.
