I just came across a report about the large-scale application of AI in real combat, which is indeed quite shocking. During this Iran conflict, the U.S. and Israeli militaries for the first time extensively used AI technology in actual warfare, potentially marking the first "AI war" in history.

The most interesting part is in intelligence processing. Under traditional methods, U.S. analysts could only review about 4% of intelligence data because the volume was too large to handle. But now, with AI, the situation is completely different. Yishai Kohn, an official responsible for planning, economics, and IT at the Israeli Ministry of Defense, mentioned, "AI has the greatest direct impact in the intelligence field," and many critical pieces of intelligence that were previously unassessable due to manpower shortages can now be processed.

According to reports, Israeli intelligence agencies have already infiltrated traffic cameras in Tehran, intercepted high-ranking officials' communications, and are increasingly relying on AI to sift useful clues from the massive intercepted data. AI visual recognition can quickly identify targets from large amounts of video and images, even distinguish specific models of aircraft or vehicles, and can also capture and summarize relevant conversations from recordings. The CEO of Israeli software company Conntour said it well: intelligence agencies already have a large amount of video data, and now AI allows them to precisely find what they need within this ocean of data.

The speed of mission planning has also improved significantly. Traditional military operation plans require coordination among intelligence officers, combat commanders, weapons experts, logistics managers, and others, taking several weeks to finalize. With AI involvement, this process can be compressed to just a few days. Because any change in the target location triggers a chain reaction—affecting pilot scheduling, flight routes, fuel consumption, and more. In the past, these updates were slow and subjective; now, AI can instantly process complex interactions and calculate the impact of each change on overall deployment. The U.S. Department of Defense now frequently uses AI to run digital models and simulations, optimizing target priorities and action plans.

But there are significant risks involved. War is one of the most chaotic and complex human activities. Jack Shanahan, the first AI chief at the U.S. Department of Defense and a retired Air Force Major General, pointed out that a major challenge for military AI is that training data itself is often outdated or unclear. Even more seriously, errors made by AI systems on the battlefield could lead to catastrophic consequences. There are reports that the U.S. military may have caused dozens of children to die on the first day of an operation at a girls' elementary school in Iran due to intelligence errors.

What is most concerning is over-reliance on AI decision-making. Emelia Probasco, a senior researcher at Georgetown University’s Center for Emerging Security and Technology, warned that handing decision-making over to AI "is a serious issue." She emphasized the need for appropriate safeguards to limit risks, but current infrastructure investments in this area are still far from sufficient. In war, human judgment will always be irreplaceable.