Another risk is that the people charged with making life or death decisions based on recommendations from AI systems could find it difficult to second-guess the machines. Some experts warn that this may already be the case in the Iranian conflict, given the difficulty for a human to comprehend all the factors that go into an AI model’s assessment.