They're not. Mad worked because any single use was substantial and unforgivable. For better or for worse, this does not apply to "AI war drones". We've already

They're not. Mad worked because any single use was substantial and unforgivable. For better or for worse, this does not apply to "AI war drones". We've already seen them drop bombs on caravans of civilians and miss their targets in asymmetric war zones, and they can only get more precise. Any outrage we wanted to see hasn't just been there and can only decrease.