이란 전쟁은 최초의 AI 전쟁이다…“전장의 분석관 참모”

The latest war in the Middle East, following the American-Israeli strikes against Iran, marks the first Artificial Intelligence (AI) war in history. Up until now, the US military had utilized artificial intelligence in limited and supportive roles, such as for drone autonomous flight software, target recognition algorithms, and cyber defense systems.

During the Iran strike, artificial intelligence led the entire operation process, from intelligence collection to target identification. The US company Anthropic introduced its AI model, Claude, for military operations. South Korean Defense Research Agency research fellow, Seo Young-bo, evaluated the American-Israeli strike against Iran as a historic turning point where commercial artificial intelligence entered military operations as a key decision support tool for military decision-making.

This was the first time in history that artificial intelligence was entrusted with the role of a commander on the battlefield, taking over from human generals. Traditionally, generals used six key factors of military operations (METT-TC) – mission, enemy, terrain and weather, available forces, time available, and civilian considerations – to establish and evaluate their strategies.

Humans take time to make judgments. US analysts spent several days reviewing satellite imagery, communication intercepts, social media, open-source information, and intelligence reports. Claude dramatically reduced the time for information processing and analysis. Claude supported all aspects of the operation, from information evaluation and target identification to battlefield simulations.

It quickly analyzed vast amounts of satellite imagery and intercepted data to extract strategically significant information, identify targets, evaluate their value, and predict their chain reaction effects. The final decision on target strikes was still left to human generals, but the era of artificial intelligence taking over the "battalion-level analyst role" in the battlefield had begun, according to Seo.

The commercial artificial intelligence model Claude was the only one approved to be deployed on the US military's classified network, SIPRNET. No other commercial artificial intelligence models like Google's Gemini could be used on the internet network because they could only process textual data. The Korean military finds it challenging to adopt the AI warfare execution method used by the US military.

This is because the use of foreign company platforms for satellite imagery, signal information, and target evaluation could lead to "military information sovereignty issues." The US military runs Claude on an isolated classified network, connected with Amazon Web Services and Palantir.