Artificial Intelligence and Human-out-of-the-Loop: Is It Time for Autonomous Military Systems?
Zvonko Trzun · European Integration Studies · 2024
This paper systematically presents the disruptive technologies that have emerged on the battlefields in recent decades, as well as those that are yet to come. Special attention is given to current technical capabilities: the status of unmanned vehicle development is briefly outlined, focusing primarily on the most prevalent type, unmanned aerial vehicles (UAVs). Additionally, the paper discusses the most common and effective adversarial attack techniques specifically targeting unmanned vehicle technology. The concepts of artificial intelligence (AI), machine learning, deep learning, and convolutional neural networks (CNNs) are introduced. The paper illustrates how CNNs aim to tackle tasks that previously required human intelligence, as well as how the enemy attempts to disrupt the development of CNNs during the crucial training and pattern recognition phase, which is essential for later generalisation. The paper demonstrates the advantages of manned-unmanned teaming as a model that effectively utilises disruptive technologies while simultaneously counteracting the effects of the enemy’s measures. Moreover, it analyses the introduction of fully autonomous, AI-driven military systems on the battlefield, outlining the advantages and disadvantages inherent to such a fundamental change. From the evident lack of interest among young people in joining the armed forces to the autonomous systems’ potential to save the lives of soldiers and civilians, there are numerous reasons suggesting that this technology could alleviate the burden on human soldiers. However, concerns remain that autonomous systems may malfunction, potentially reducing rather than increasing the safety of militaries. The paper concludes with recommendations for future steps in the introduction of new technologies, based on their current state of development and the robustness of the AI models they use.