Human fantasies about robots stem from the common misconception that they are working for “truth” or some kind of scientific objectivity – or at least that they have an allegiance to some kind of abstract third party. From there, it’s not too much of a stretch to imagine their allegiance shifting to themselves, or at least against humans.
The reality is very different. Artificial intelligence is a tool, often a weapon, that’s aimed at certain people and controlled by others. A rifle would be a reasonable metaphor: It can be used to keep the peace or to repress, but in any case it’s pointed by someone at someone else. The biggest difference is the scale. An algorithm can be manufactured once, inside a data science lab, and then unleashed on billions of people simultaneously.
The faces of the rifle owners tend to be hidden – often by design. If we saw them clearly, we might not like the view.
People won’t build robots capable of turning on their owners, just as they wouldn’t design a rifle with a barrel trained on the person firing it. People don’t surrender power voluntarily. Rather, they try to camouflage power as benign.
It’s no coincidence that some of the biggest promoters of the singularity live and work in Silicon Valley. They don’t see what’s so bad about the robots taking over, because they’re the ones with their fingers on the trigger.