Iinteligência artificial e tomada de decisão: o caso dos sistemas Lavender e Gospel no conflito entre Israel e Hamas (2023–2024)

Abstract

The integration of artificial intelligence systems into military functions, stemming from their essence as a dual-use technology, shifts the axis of transformation of contemporary warfare from the weapons platform to the decision-making process that precedes the use of force. This research project aims to analyse to what extent, and under which conditions, the employment of the Lavender and Gospel systems by the Israel Defense Forces (IDF), in the conflict against Hamas (2023–2024), affected human scrutiny over the decision on the use of lethal force. The object of the research is therefore to examine the degree of human deliberation exercised in that decision, in its relation to algorithmic systems of target-selection support. Anchored in the perspective of Strategic Studies — and deliberately detached from normative approaches — the research departs from the hypothesis that the employment of these systems would have promoted a partial transfer of decision-making authority from the human operator to the algorithmic system, reducing the degree of effective human deliberation over target selection. The study is designed as qualitative single-case research, based on secondary documentation from open sources submitted to cross-verification, and unfolds procedurally in stages ranging from the constitution of the documentary corpus to the assessment of the hypothesis according to explicit decision criteria. The analysis confronts the evidence of the case with a conceptual framework structured around the notions of meaningful human control, automation bias, kill chain, selector and algorithmic decision-making, examining human control in its temporal, informational and effective-authority dimensions. It is expected to contribute to the debate on the limits of human control in contemporary military decision-support systems, offering empirical evidence on the reconfiguration of decision-making authority in the algorithmically mediated employment of lethal force.