• Langue : Français
• Identifiant : ULIL_MSPL_2025_009
• Faculté/Ecole : Sciences Po Lille
• Date de soutenance : 28/05/2025
• Type de mémoire : Mémoire de Master
• Discipline : SPL - Stratégie, Intelligence économique et Gestion des Risques (SIGR)

Résumé

In less than a decade, AI has become a widely applied technology in numerous everyday contexts. Specifically, numerous states started investing and researching its application within the military field, with the aim of transforming wars in conflicts developing within a quicker time frame, reducing the number of civilian and military victims. Soldiers are being less and less retrieved from the battlefield, leaving the space to new technologies able to preserve their lives. Nevertheless, such AI systems are still affected by a still quite elevated error rate, biases within training data, vulnerabilities in the cyber protection of their systems, which affect enormously the safety with which they can be applied within military contexts, where specifically in battlefield situations, human lives are at stake. New issues about responsibility for mistakes or unforeseen actions are arising, and the international legislation is still completely unprepared in welcoming such novelty. Historically speaking, military technologies have been always researched, developed, used and then finally regulated or banned when it was noticed that the amount of humanitarian and environmental damage they caused. Will it be the case of AI integrated to military technologies? However it should not be forgotten that AI does not solely help with target acquisition and striking as it is generally believed by civil society organizations, but also helps in lowering the costs of transportation and similar within the logistic sector, alongside supporting the military decision-making and personalized training. The aim of this thesis is to unravel this change starting from a historical point of view, passin through the analysis of some technical aspects, which will help in understanding benefits and risks of military-AI, in addition to the analysis of a specific case (the application within nuclear field and weapons), ending with an examination of the level of unpreparedness of international legislation on the topic.