INTELIGÊNCIA ARTIFICIAL EM SISTEMAS AUTÔNOMOS DE DEFESA: SEGURANÇA, CONTROLE HUMANO, EDUCAÇÃO, ENGENHARIAS E CONFIABILIDADE PARA O EMPREGO DEFENSIVO
INTELIGÊNCIA ARTIFICIAL EM SISTEMAS AUTÔNOMOS DE DEFESA: SEGURANÇA, CONTROLE HUMANO, EDUCAÇÃO, ENGENHARIAS E CONFIABILIDADE PARA O EMPREGO DEFENSIVO
-
DOI: https://doi.org/10.22533/at.ed.575122603071
-
Palavras-chave: inteligência artificial; sistemas autônomos de defesa; controle humano; educação em Engenharia; confiabilidade
-
Keywords: artificial intelligence; autonomous defence systems; human control; engineering education; reliability
-
Abstract: The integration of artificial intelligence (AI) into autonomous defence systems expands perception, forecasting, decision-support, infrastructure protection and threat-response capabilities, while introducing risks related to unpredictability, classification failures, cyber vulnerabilities, automation bias and diluted human responsibility. This article examines the technical, human, educational, ethical and legal-institutional requirements for the responsible defensive use of such systems, emphasizing safety, meaningful human control, engineering education and reliability. A theoretical-conceptual and documentary review was conducted, covering standards, resolutions, governmental strategies, documents from international organizations and scientific literature, prioritizing publications from 2018 to June 2026 together with indispensable classic studies. The results show that the formal presence of an operator does not ensure effective control; situational understanding, real authority, sufficient intervention time, competence, traceability and the ability to suspend action are required. The SC²D-AI Model is proposed with six decision gates: defensive legitimacy, operational-domain delimitation, technical safety and reliability, human control, testing-evaluation-verification-validation, and monitoring with recertification. The model also supports engineering education through simulated projects, risk analysis, human factors and interdisciplinary assessment. AI may support surveillance, logistics, search and rescue, cyber defence and the neutralization of material threats, provided that verifiable limits, effective human supervision and evidence-based governance are maintained.
- Marcio Mendonca
- Cintya Wedderhoff Machado
- Vitor Blanc Milani
- André Luiz Przybysz
- Andressa Haiduk
- Augusto Florencio Dias Neto
- João Luiz Dallamuta Lopes
- Fabio Rodrigo Milanez
- Vlademir Aparecido Freire Junior
- Paulo Alexandre Lourenço Jesus
- Ana Clara Augusto Jesus
- Adriano da Silva Moreira
- Francisco de Assis Scannavino Junior
- Marcio Takahashi Kawamura
- Norwin Porfirio Carrasquel Poturo
- Ana Cláudia Araújo da Silva
- Luiz Francisco Sanches Buzzacchero
- Emerson Ravazzi Pires da Silva