Nearly 200 years later, artificial intelligence floods command centers with such volumes of data that it often leads to situations where decisions are made under great uncertainty.<\/p>
AI technology specialists have aimed to make modern weapons, enhanced by advanced technologies, strike accurately and effectively \u2014 and at the same time, spare civilians.<\/p>
What isn’t it working?<\/p>
In 2021, an Israeli brigadier general, commander of an elite intelligence unit, wrote a book under the pseudonym Y. S.<\/mark><\/a> about a partnership model between humans and artificial intelligence. According to the author, the outcome should be mutual enhancement of both parties, leading to increased efficiency of actions, improved security, and overall well-being.<\/p> However, the Hamas attack on October 7, 2023 erased that scenario.<\/p> In preparation for a response, the Israeli Defense Forces (IDF), together with Unit 8200 \u2014 an elite Israeli intelligence unit \u2014 developed advanced AI systems to support military operations in the Gaza Strip. Three innocuously named application teams were created: Lavender<\/strong>, Habsora<\/strong> (Eng. Gospel) and Where’s Daddy<\/strong>.<\/p> Lavender<\/strong> is a personal target identification system. Using machine learning algorithms, it analyzes different formats of intelligence data \u2014 from visual information and mobile phone data, through social media connections and contact networks, to battlefield information. It was designed to identify all suspected Hamas and Islamic Jihad fighters. It classified about 37,000 Palestinians as potential targets. Israeli sources<\/mark><\/a> assured that this system is 90% accurate. The “permissible error in wartime” included individuals with the same surnames, sharing phones with Hamas members, those who had contact with someone from Hamas, those who frequently changed phone numbers, and also those who were just misclassified by the system.<\/p> Habsora<\/strong> is a target identification system with a broad range of operations. It searches for enemy buildings, equipment and personnel. Based on statistical data, it also calculates the probable number of civilian casualties that would occur in an attack on a given target. It generates about 100 targets daily. For comparison, analysts used to be able to identify up to 50 targets… Annually.<\/p> Where’s Daddy<\/strong>is a system for tracking individuals previously identified by the Lavender system and informing the IDF of their return to their family homes. One intelligence officer explained the strategy: “The IDF bombed them at home without hesitation \u2014 it was a more effective and cheaper option<\/a>“<\/mark> compared to attempting to strike them in a bunker or tunnel. When asked about such types of attacks, the Israel Defense Forces (IDF) stated: “The IDF categorically denies allegations of employing any strategy involving killing people in their homes.”<\/p> The systems described above form a decision-making chain: locate\u2013track\u2013follow\u2013target. Unfortunately, the role of humans has been reduced to the “approving” component. As a result, officers often had merely 20 seconds to approve a target. They could usually only ensure that the target was male. In the first 35 days of the conflict, the IDF attacked over 15,000 targets. The Israeli side accepted significant civilian casualties: to eliminate a low-ranking Hamas fighter, the death of 15-20 innocents was anticipated, and for higher-ranking commanders, over 100 civilians per target. Families and neighbors were considered “collateral damage”.<\/p> The use of these systems has sparked a debate on the responsibility for AI system errors and compliance with international law.<\/p> UN Secretary-General Antonio Guterres and then White House spokesman John Kirby expressed their concern. Yes, “concern”.<\/p> In the face of Russian aggression, Ukraine has focused on developing the Drone Silicon Valley<\/mark><\/a>. Thanks to the capital and technology of foreign companies and the potential of local high-tech experts, the Ukrainian side was able to launch Brave1<\/mark><\/a>, a government program supporting technological startups. Among other things, they are trying \u2014 through the development of unmanned systems \u2014 to solve the chronic problem of soldier shortages. Ukrainian authorities emphasize that autonomous weapons are “a logical and inevitable next step” in arms modification.<\/p> The Ukrainian company Swarmer<\/mark><\/a> is building drone swarms that automatically identify targets, call for reinforcements, and verify the outcomes of actions taken by other units. The human role is minimal \u2014 it comes down to pressing a few buttons to launch the system and approve the attack plan. The conflict in Ukraine has accelerated the implementation of battlefield security technologies. Some examples are the Maven Program<\/mark><\/a> and the products of companies such as Palantir<\/mark><\/a> and Clearview<\/mark><\/a>. These technologies automate the creation of the war “backend”, gathering data from all available sources, including the social media of Russian soldiers. A real-time updated map of operations is being created, which radically changes the speed and effectiveness of military actions. Simultaneously, documentation is being compiled that will serve as the basis for potential future reparations claims and war crime charges.<\/p> Taiwan, in collaboration with the USA, is developing a deterrence strategy \u2014 the “Hellscape<\/a>“<\/mark> plan, which involves launching swarms of thousands of drones, autonomous boats and torpedoes to strike at key enemy units and systems. This is designed to confuse the defense mechanisms of a potential aggressor. Everything is intended to operate with minimal or no human involvement.<\/p>Ukraine War \u2014 Drone and AI Laboratory<\/strong><\/h4>
Potential scenarios and global implications \u2014 Taiwan, China and the USA<\/strong><\/h4>