The IA Arandu tool of the Amazonas State Court of Justice

Abstract

Technological advances and the drive for procedural efficiency within the judiciary are driving profound changes in the structure of the justice system. These needs are particularly pressing in the Amazon region, where geographical and logistical barriers hinder access to justice. In light of this, the question arises: how does the implementation and use of the generative artificial intelligence tool Arandu by the Amazonas Court of Justice impact the efficiency and management of cases? This study aims to analyse the Arandu tool and the Archeion database, identifying their potential and limitations for the management of cases in the State of Amazonas. To this end, an exploratory and descriptive methodology of a qualitative nature is adopted, with occasional recourse to third-party quantitative data, supported by bibliographic and documentary research from the analytical perspective of inductive reasoning. The theoretical framework is based on the waves of renewal in access to justice by Mauro Cappelletti and Bryant Garth, aligned with the contemporary debate on algorithmic governance. It is concluded that, although the tool presents itself as a promising means of mitigating delays and optimising bureaucratic procedures, its actual effectiveness remains contingent upon strict human supervision and addressing the risks of algorithmic bias and opacity, in order to safeguard the legal security of vulnerable groups.