In this case I think they're broken down immediately because they won't be used.

He did also do some stories about 'fixing' the laws but in every case that resulted in either limiting the robots capabilities in a restricting way or requiring them to have more power over people in ways that could be harmful to humanites self autonomy in the longrun.

Actually autonomous? Or guided by computer vision? Technically CV is AI but not what concerns most people.

Which this article is not about.

It's seen as an unexpected loophole in the books. Similar to how a surgeon won't kill one healthy person to save two with their organs.

At least on I,Robot

Edit: also, on I,Robot they harmed very specific people with very calculated results. Not like going to war, even on the defensive side.

Where’s this from?

You just know that Trump is salivating to deploy something like this against ANTIFA. It'll be allowed to go into fully autonomous mode as it is modified to shoot "non-lethal" (yes I know thats not the right term but thats what they will say) rounds as it deems fit by their specially trained AI to basically shoot anything moving that isn't ICE.

The primary use of unmanned ground vehicles are for logistics especially logistics with explodey things it is very dangerous to ask a human to carry through a battlefield. If a UGV carrying a bunch of mortar rounds is hit by an FPV drone that sucks for the mortar crew that doesn't get ammunition but it is unimaginably better than a human being driving a truck full of mortar rounds being hit by one.

The next most useful use of unmanned ground vehicles are for casualty evacuation under fire and under conditions too risky for humans to try.

The third most useful is as a weapon.

So yes this is scary but honestly a human with a gun is still far more terrifying. The broader future of these big derpy rolling ground robots are disaster response and lifesaving applications, they are useful for war too but under very specific human coordinated contexts.