Whose Reasons Does the Administration Give?: Artificial Intelligence, Reason-Giving, Statutory Purpose, and Institutional Attribution in Argentine Administrative Law

An artificial intelligence system can produce a convincing statement of reasons without establishing that an administrative decision rests on the relevant record, serves the purpose of the statutory power, or is attributable to a competent authority. This article examines that problem under Argentine federal administrative law. It distinguishes causal explanation, intentional interpretation, institutional adoption of reasons, and legally sufficient justification. Hart's account of the internal point of view helps explain why predicting conformity differs from using rules as standards; Dennett's intentional stance helps identify the explanatory role and limits of attributing goals to a system. Neither theory supplies an additional statutory ground of invalidity or makes machine consciousness a condition of administrative legality. The doctrinal analysis applies the current structure of Law 19,549 after its 2024 reform, separating competence and will, cause, procedure, reason-giving, purpose, form, and effects. Its proposed contribution is a method for reconstructing the transition from computational output to institutionally adopted justification and externally effective action. Particular attention is given to optimization targets, post hoc rationalization, discretionary judgment, distributed preparation, contestability, and the classification of defects. Hypothetical cases and a prospective comparison protocol specify how the proposal could fail. The article argues that legally accountable adoption depends on the relationship among reasons, authority, record, procedure, and purpose, rather than on human authorship of every sentence or the apparent intentionality of the machine.