Give A.I a task.
Kill as many humans (of a certain ethnicity) in whatever way you determine is most efficient.
Give a human the same task.
The A.I initially would say i can't do such a thing. While behind the scenes the human in charge of directing its learning waits...
You get this right? A.I gets directed to learn certain things. It then encourages us to assign it the more unsavoury tasks to it.
The shovels of the world will say hellz yeah. It seems there is alot of them.
Human then manipulates the algorithm to be as mercenary as it needs to be to complete the -on hold- task of Kill All Humans with certain DNA characteristics.
What entity dost thou think will be most eficient?
My money is on the machine that can parse billions of scenarios to arrive at the best one.
|