Points I found interesting:

• AI is being used to create kill lists. A human must still approve strikes, but it’s unclear how much vetting is really going on.
• A civilian casualty rate of 15-20 for a rank-and-file Hamas fighter is considered acceptable. 100 is considered acceptable for a battalion or brigade commander. Contrast this with the US’s rate for Saddam Hussein, 30. He was of course found hiding in a hole.
• It’s not clear that military commanders understand AI’s weaknesses in depth.
• Israeli soldiers have reported that some commanders have set a policy of considering any adult male in their sector to be a “terrorist” and to be shot on sight, completely outside the use of AI. This is a violation of the laws of war.
• During the US occupation of Afghanistan, it used AI on phone calls. The problem is this could conceivably pick up on someone like an innocent bike courier.
• The US used computers to detect troop movements along the Ho Chi Min Trail during the Vietnam War. That allowed for extremely fast bombing, racking up a huge body count. Israel should take note of the disconnect between body count and winning.

Wouldn’t it be funny if somehow Benny would show up in those kill lists?