AI-Driven Naval Encounters: 'Haegum' Vessel's Rapid Response Raises Human Control Questions

In a simulated encounter off the Northern Limit Line (NLL), an artificial intelligence-powered unmanned surface vessel named 'Haegum' demonstrated its rapid response capabilities to an enemy incursion. The demonstration showcased an intelligent command and control system that fused real-time data from maritime, air, and satellite sources to generate a high-precision operational map. An 'AI tactical staff' then proposed multiple tactical plans, including reconnaissance and combat options, for human commanders to select and execute.

The entire process, from enemy detection to engagement, was completed in over 90% less time compared to traditional human decision-making. LIG Defense & Aerospace (LIG D&A), the company behind the system, announced on June 1st that it had successfully verified its AI-based multi-purpose unmanned system technology, emphasizing "speed" and "integration" as key competitive advantages for future battlefields. They claim the system can reduce the time from detection to engagement to one-tenth of the current duration, enabling 24-hour unmanned operations without personnel rotation.

The AI solution supporting human commanders' decision-making was developed in collaboration with Palantir, an AI company serving clients like the U.S. Department of Defense. The system played a crucial role in simulated engagements, involving five stages against enemy vessels and seven stages against a simulated enemy submarine. Palantir's 'Maven Smart System' has been instrumental in analyzing vast battlefield data, identifying target candidates and coordinates, prioritizing attacks, and evaluating outcomes, transforming tasks that previously took weeks or months into real-time operations.

“

This system can significantly reduce the time required from enemy detection to commander's decision and engagement to one-tenth of the current level, providing an overwhelming advantage in modern warfare. Furthermore, unmanned operations are possible 24 hours a day without personnel rotation.

However, the rapid acceleration of decision-making in AI-driven warfare raises concerns about the reduction of human oversight and the potential for errors, such as unintended friendly fire incidents. The NLL, a sensitive area prone to accidental clashes between North and South Korea, highlights the risks associated with escalating conflicts. Despite simplified rules of engagement following past naval skirmishes, the principle of using force as a last resort and to the minimum extent necessary remains critical.

The UN Human Rights Council's advisory committee has expressed concerns that autonomous weapon systems weaken human control and accountability, creating a "transparency and accountability vacuum" in AI-based decision-making. LIG D&A stated that the AI only makes recommendations, and the final decision and execution of actions, including engagement, remain with human commanders at every stage, emphasizing that the system's development is focused on human intervention.

“

AI only recommends, and the final selection and execution of actions, including engagement, are made through human commanders at every stage. The system's development focus is also on human intervention.