INTELIGÊNCIA ARTIFICIAL E INOVAÇÃO NA ADMINISTRAÇÃO PÚBLICA: DESENVOLVIMENTO DE UMA FERRAMENTA DE AUTOMAÇÃO DE PARECERES NO INSTITUTO FEDERAL DE EDUCAÇÃO, CIÊNCIA E TECNOLOGIA DE GOIÁS (IFG)

Abstract

This dissertation investigates how the application of Artificial Intelligence can support the assisted automation of administrative opinions in human resource management processes at the Federal Institute of Education, Science and Technology of Goiás, Brazil, considering the increasing demands for efficiency, analytical standardization, and institutional control. The research problem arises from the high cognitive workload associated with document analysis, normative interpretation, and technical opinion drafting, which directly affects the speed and consistency of administrative decision-making. The study was conducted through a Systematic Literature Review, structured in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) protocol, with methodological quality assessed using the Critical Appraisal Skills Programme and the Mixed Methods Appraisal Tool. A total of 2,015 records were identified across the Web of Science, Scopus, and SciELO databases, resulting in 36 studies included in the final sample after applying eligibility criteria. The findings indicate that the use of Artificial Intelligence in public administration has been associated with reduced processing time, assisted automation of document-intensive activities, and increased technical consistency in decision-making processes. However, these benefits do not derive solely from the technology itself, as they depend on institutional capacities, data governance, adequate technological infrastructure, and formal mechanisms of human oversight. Based on the analytical synthesis of literature, a conceptual model for the adoption of Artificial Intelligence in public administration was developed, structured around three interdependent dimensions: institutional and organizational capacity, technological infrastructure and data governance, and institutional control with mandatory human validation. Building on these findings, the study resulted in the development of the technical-technological product Aurora, designed as a decision-support system for administrative process analysis and opinion drafting in human resource management. The solution integrates automated document reading, structured information extraction, retrieval of normative references, and AI-assisted text generation, while preserving human responsibility for final decisions. The system was directly derived from the evidence identified in the literature, translating theoretical insights into an applied solution that operationalizes the technical and organizational requirements associated with the adoption of Artificial Intelligence in the public sector. It is concluded that the assisted automation of administrative opinions, when implemented as support technology and integrated with institutional validation mechanisms, can enhance administrative efficiency, strengthen the standardization of technical analyses, and foster innovation in public management without replacing human judgment. The study advances the field by articulating systematic evidence with an applied technological solution, offering guidelines for the responsible implementation of Artificial Intelligence-based systems in public institutions.