Is Artificial Intelligence Safe?

It seems hard these days to go a week without hearing something new regarding Artificial Intelligence (AI). Whether it's the government wanting to put regulations on it, or Amazon selling facial recognition software to law enforcement. With all of this coverage, the question begs to be asked, "is AI safe?"

Most researchers agree that there are really only two ways that AI can go bad. One, it is programmed to do something bad. Two, it is programmed to do something good and comes up with a bad way to accomplish its task. 

Keeping it from being programmed to do something bad sounds easy on paper, but in reality, it will probably be pretty hard. Government regulations are most likely the way this will be accomplished. With that said there is a fine line between enough regulations to keep people with bad intentions from getting their hands on this technology and so many regulations that nothing can be built with AI. 

Keeping AI from coming up with a bad way of completing a good task will most likely involve human approval of the plan the computer has come up with. While this sounds very basic, it might just be the only sure fire way. 

So, to dramatically oversimplify the question if there are good regulations in the system and the technology has many trip wires and off switches it can most likely be kept safe. 


Popular Posts