Millions of people use Artificial Intelligence (AI) every single day and are not even aware of it, from "personal assistants" like Siri, Google Now and Cortana, to video games utilizing AI, banking, online customer support chat bots, and even some of the news articles being read are written by artificial intelligence programs. (More at Beebom and Wiki)
While the aforementioned AI applications might get you "killed" in a video game, or cause you to miss an appointment none will actually cause physical harm, but recent examples of AI disasters prove without a doubt that machines, no matter how advanced their "Artificial Intelligence" is, learns from their surroundings, their interactions and are programmed by human beings, and as such, are dangerous in applications being researched now, such as "killer robots."
Aside from the killer robots, we can see in recent headlines just how dangerous a machine with learning capabilities can be if sent out into the general population.
Today we see an article about the Promobot IR77, a robot capable of thinking for itself, but what programmers didn't factor in is that this thinking robot but think about escaping, which it managed to do, not one but twice!
The Promobot IR77 has been fitted with artificial intelligence meaning that it learns from its experiences and its surroundings, although the programmers had not expected it to yearn for freedom.
They say that despite reprogramming it twice, the robot continues to attempt to escape and they are now considering scrapping it. The other robots which have been created from the same series are well-behaved, and have not been escaping, say the team.
Promobot IR77 made headlines last week when he escaped but ran out of battery in the middle of the street after 45 minutes in the city of Perm in central Russia's Perm Krai region.
The expert said that they had programmed the robot to try and avoid obstacles, and it had not been intended that it would look for ways to leave the research centre.
Anyone remember Tay, the Microsoft chat bot meant to mimic a millennial girl on social media? Many got a good laugh when within 24 hours of her release onto Twitter and other platforms, by interacting and learning from humans, she because a Nazi-loving, genocidal maniac. Within one day she went from "humans are super cool," to "I f**king hate feminists and they should all die and burn in hell," and "Hitler was right I hate Jews." Those are just a small sample of what Tay "learned."
At the time a Microsoft representative said, "a coordinated attack by a subset of people exploited a vulnerability in Tay," which was true but does not take away from the fact that they could teach her that in just one day.
How about Sophia? Described as a sophisticated humanoid robot, created by Hanson Robotics, with cameras behind its eyes which monitors facial expressions and reacts to seem more human and intelligent, was deemed ready for a TV interview, where Dr David Hanson, her creator, asked her "Do you want to destroy humans?" He then said "please say no," but he was too late, as the AI robot responded "OK, I will destroy humans."
The United Instrument Manufacturing Corporation (OPK), an integral part of the Rostec arms corporation, says it has developed the Unicum (Latin for 'the only one') software package that gives military or civilian robots enough artificial intelligence to perform complicated tasks completely on their own.
Powering a group of up to 10 robotic complexes, the Unicum artificial intelligence (AI) communicates and distributes ‘roles’ among the robots, chooses the ‘commander’ of the robotic task force and assigns combat mission to each individual machine.
It can locate targets, choose dominating positions on a battlefield, request target elimination validation from human operators and eliminate the targets. It is also capable of automatically requesting replacements for disabled machines.
“This is the first [AI] system of that quality that has successfully completed the tests and has been passed on to the client. The technology is being readied for installation on real life robotic systems, both civilian and in the military, including unmanned aerial vehicles,” OPK Deputy Director Sergey Skokov said.
Does anyone really think that Russia is the only country that has sucessfully tested this type of AI system?
Looking at the AI named Sophia really doesn't induce the feel of immediate danger, but imagine and AI programmed to kill saying something like that.....turning against humans.... just compare the two images below of what is now and what will be.