Now this could work quite well, that is IF they can successfully program said drone to
attack targets based on the current state of geo-political affairs, based on the location
of the mission.
Sometimes, the battlefield isn't always as well laid out, with distinct targets,
as commanders would like. Remember what happened with Vietnam: American pilots would fly close to
Vietnamese ports, and could actually SEE THE SURFACE-TO-AIR MISSLES that would eventually
be flung at them, and under orders from Washington, were NOT allowed to attack those ports.
Why? On-site intel suggested that the Russians were actively funding and working with the
North Vietnamese Army in an 'advisory' capacity, so any attack by US forces would have been seen
as a direct attack on Russia (if any Russians were killed). You can imagine the pilots' frustration at
that scenario, knowing full well that those same SAMs would eventually kill US airmen.
A drone might be able to decide which target to attack from a purely tactical standpoint, but how will
it distinguish between 'ally' and 'foe' when those two entities use the same hardware?
I think there are too many variables for this to be deployed effectively. That drone could only be deployed
in distinct, controlled mission environments, where the target packages have been clearly demarcated
and identified, or else it might just end up blowing up the 'wrong enemy'.
...and yes, what happened in Stealth is a definite possibility.
Humans should always retain that ability and responsibility to choose what targets to attack, and then let the drone
decide the most efficient way to attack those targets. Leaving that decision up to a machine is a recipe for disaster.
