ARTIFICIAL INTELLIGENCE AND URBAN RESILIENCE: FROM DISASTER PREDICTION TO ETHICAL AND SUSTAINABLE TERRITORIAL MANAGEMENT

Abstract

This article analyzes the incorporation of artificial intelligence into urban planning as a structural inflection point in the administrative function, marked by the transition from a reactive model to an anticipatory, data-driven governance. Based on the resilient city paradigm, it investigates how deep learning systems and georeferenced monitoring have come to influence the formation of state will, shifting the traditional axis of administrative discretion to a regime of technical discretion mediated by algorithms. It argues that this reconfiguration necessitates a revision of the classic categories of Administrative Law, especially regarding the motivation, transparency, control, and accountability of administrative acts. The study demonstrates that algorithmic opacity compromises due process and demands the institutionalization of explainability and auditability as requirements for the validity of automated public decisions. Furthermore, it examines the impact of data governance and dependence on private infrastructures on the digital sovereignty of the State, advocating for public control of informational assets as a prerequisite for decisional autonomy. In the field of civil liability, it is evident that the causal fragmentation inherent in intelligent systems does not preclude the objective regime of state liability, imposing the adoption of expanded models of imputation based on the risk of the activity and the socialization of damages. In light of the precautionary principle, it is argued that algorithmic urbanism should incorporate ethical safeguards from the design of the systems, ensuring effective human oversight in critical decisions. Finally, the article highlights that the legitimacy of intelligent governance depends on citizen participation and overcoming the digital divide, as well as the construction of transnational normative standards capable of subjecting technology to the values ​​of the rule of law. It concludes that artificial intelligence is only compatible with the constitutional order when structured on four pillars: centrality of human decision-making, transparent and auditable motivation, objective responsibility, and prudential governance guided by the dignity of the human person.