Justification for piloted aircraft in a far-future setting?

That was the short story "Evidence" in the book I, Robot by Isaac Asimov.

It was never made entirely clear to the reader whether Stephen Byerley is a human, or a robot duplicate created by the real Stephen Byerely, who (it is suggested) was horribly disfigured in an accident, and built a humaniform robot to take his place publically.

Susan Calvin, the robo-psychologist, says that if he obeys the Three Laws of Robotics, it doesn't prove that he's a robot, since the Three Laws also form a basis for strict moral behavior. So anyone who adheres to them could simply be "a very good man." But if he breaks any of them, it would prove that he's not a robot.

After Byerley punches a heckler on the chin and is deemed to be human, Calvin comments, "There is one instance where a robot may harm a human — when the human to be harmed is merely another robot."

But nothing was ever proven one way or the other, since Bylerely had his remains atomized when he died.

You mean "died".

Let's not also forget the First law can also be subverted if refusing would allow greater harm. Say that a human is hurting other humans, a Robot is perfectly allowed to harm that bad human, since by inaction it is allowing other humans to come to harm.

The problem here is that these are unmanned aircraft in war, where they must pick out targets, and unless the AI is advanced enough, how are they to pick out what is deemed a legitimate target unless still directed in some way by people.

In aircraft vs aircraft that is easier than picking out a random person from a crowd. Things like radar returns, thermal signatures, and presence or lack of IFF transponder codes can help ID a craft. There will also be noticeable design choice differences in manned vs manned craft especially terms of space used for the pilot. Direct visuals can pick out different shapes and ID them that way.

Then there is the fact a craft that is an air space it shouldn't be in tends to draw a lot of attention pretty quickly.

The newest version of the Python AAM has an electro-optical mode it can use to find targets. Supposedly it can even use the camera to home in on the cockpit of an enemy aircraft, and since it's not an infrared seeker it can find targets that aren't giving off heat. As Tuefel pointed out there are ways to determine friend from foe such as visual ID and IF Fs. A near-future fusion of several of these systems could allow an autonomous aircraft to fight independently.

They'd still need pre-battle directives from a human (go here, kill this, that kind of thing), but that's more or less the same as a pilot getting orders and doesn't make the aircraft any less autonomous when it's actually out flying around.

edited 8th Nov '17 8:31:48 PM by archonspeaks

And lets be blunt here-its not as if humsn combatants have ever made the avoidance of collateral damage their highest priority. Protecting innocent lives is nice and all, but every military Ive ever heard of has considered winning as more important. Equaling or even exceeding human performance in that regard will not be that difficult.

edited 8th Nov '17 8:11:21 PM by DeMarquis