Say someone built a machine, that was perfectly human in mind. It got angry and killed someone. Is the person who created the machine, nothing, or somehow the machine at fault in U.S. law?
That would have to be decided in a lawsuit. Or several.
- Criminal responsibility generally requires either criminal intent or criminal negligence. If there was no intent, the question becomes if the operator of the gadget or the inventor of the gadget should have prevented it somehow.
- Civil responsibility has a lower standard of proof. There could be a wrongful death suit.
- Then there is the probable product liability suit.
As a layman, if the gadget is sentient enough to take the responsibility, it should have civil rights.
While the wording is slightly different, this is the big argument in car manufacturing right now.
Who is responsible if a self driving car has an accident?
The car itself is not a valid entity in law, currently the argument is leaning towards the manufacturer being responsible to the point of Volvo making a statement to that effect.
There's no definitive answer yet. There will have to be one before self driving cars are allowed loose on the roads so hold on for an answer soon.
There's also a secondary issue here, teaching the cars to handle the trolley problem.
Disclaimer: I'm no citizen of the US, nor am I a lawyer or anyone who comes in direct contact with correctional law.
That said, I think the biggest issues are proving that:
- The machine is 100% sentient in all legal terms, i.e. capable of making decisions independently from outside factors or human instructions.
- The murder was an intentional action of the machine, not a (by)product of it's original AI coding.
- There's no criminal intent (or neglect) by the manufacturer, i.e., there's no trace of coding that would cause the machine to kill under certain circumstances.
If these terms are met the machine would be held responsible. We wouldn't sue parents for their murderous child (unless they are proven accomplices).
Self-driving cars (at least the type we have now) would not be a legal entity, as they are nothing but the sum of human instructions, so you can't sue them (no point anyway even if you would). If there's no fault to the infrastructure, I would expect the manufacturer to be held responsible. However, unless there is again no prove of any criminal intent/neglect, that would be as an accident, not a murder. And this would end up an insurance issue, not a correctional one.
US law does not recognize machines as "persons". As a result, any blame (if blame were determined) would rest with the builder.
Of course, the question would arise, "Did the builder have reason to think that normal operation (or even abnormal operation) might be reasonably considered dangerous?". So, how, exactly, did the machine kill its victim? If the machine has a built-in machine gun or chain saw, then it's pretty clear that the builder is in big trouble. If the machine used a built-in internet connection to modify the operation of another machine, as long as the modification were not "reasonably forseeable" there is no reason lay blame.
EDIT - Martine Votvik has pointed out that corporations are recognized as "persons". While this is true, it is a very limited version of personhood. US corporations cannot:
Vote Marry or divorce Be ordained Practice medicine Practice law Be assaulted or murdered or raped Be arrested or imprisoned (note that a warrant might be issued for officers of the corporation - but not the corporation itself)