Everything else flows from that.

At a high-level, that decision compression will be an arms race between competitors for military advantage, where the argument to relax human-scale decisions will be "If we don't do it, we'll lose to someone who will."

You saw the same thing in the nuclear arms race, as launch-to-impact time compression (as technology advanced) eroded available human decision time.

But popping off half-cocked often results in bad outcomes, so there's a tradeoff between decision quality and decision time.

In theory, having AI in the loop could improve both. In practice, I suspect you're right.

And maybe always has been, internationally-speaking?

It would be nice to reform UN and supra-national rules of warfare in light of modern developments.

Even a "this is an acceptable ratio of civilian collateral deaths, above which constitutes a war crime" would be helpful for curtailing nation's baser urges.

If I had to guess, it was even less. Now at least there can be media backlash.

> Even a "this is an acceptable ratio of civilian collateral deaths, above which constitutes a war crime" would be helpful for curtailing nation's baser urges.

Having some kind of metric would be nice. However, there are problems with it being too simple (your enemy has a say in how exposed their civilians are; and we'd hate to leave countries with less precision in their militaries without the ability to defend themselves).

If killing the right person is too much work then you shouldn’t be killing

Some historians say US Secretary for War Henry Stimson had a personal reason for sparing Kyoto. (edit : wrong Henry)