Israeli study reveals: AI worsened casualties and destruction in Gaza

An Israeli study by the Open University reveals that Israel's use of artificial intelligence in the Gaza war worsened casualties and destruction, accelerating target production and expanding the target bank, leading to more misidentifications and civilian deaths, and noting the absence of human oversight and ethical review.

An Israeli study by the Open University of Israel reveals that Israel's use of artificial intelligence in the Gaza war worsened casualties and destruction. The study was authored by Professor Yagil Levy and Professor Ori Shurtz, and published as a position paper by the Open University's Institute for the Study of Civil-Military Relations. The study indicates that Israel used AI extensively to produce intelligence targets without deliberations on ethical, political, and social consequences. It identifies four groups of risks: changing the form of knowledge production, accelerating and reducing the cost of target production, errors in accuracy, and encouraging thinking in technological solutions. It clarified that Israel relied on indirect behavioral characteristics to criminalize targets, which includes many civilians such as journalists and aid workers. It affirmed the ethical and legal difficulty in relying on statistical logic to determine human targets. It noted that AI does not meet the principles of the rule of law and cannot be held responsible. Former Chief of Staff Aviv Kochavi announced that the use of AI increased target production from dozens per year to hundreds per week. According to the study, this causes mass killing of civilians; even if the technology reduced errors by 90%, the 300-fold increase in targets raises misidentifications 30 times. It mentioned that Israel described 37,000 people as Hamas operatives with a probability exceeding 90%, including police and civilian employees. It noted the absence of human oversight, leading to a decline in information quality and a preference for quantity of attacks over quality.