Believe it or not, Google patented a way to turn off all artificial intelligence if threat to humans
Science fiction has raised several times: what would happen if computers and robots rebelled against humans?
Perhaps the war between machines and humans showing James Cameron in his film “Terminator” (1984) has been the inspiration for Google, which decided to patent a technology that tries to protect the human race if technology threatens its security and integrity.
Or maybe it was the most recent film “Ex Machina,” which explored the challenges of the interaction between artificial intelligence and human.
the solution Google is a “red emergency button “that would be able to” turn off “all artificial intelligence , according to a recently published by the Institute for research on intelligence Machine (MIRI, for its acronym in English), University of Berkeley research in California, USA.
Although, according to experts who study the possibility that the machines from becoming dangerous for humans, it is not certain that all the robots can be “deactivated easily” with this system. We explain what the device Google.
Out of control?
Project leaders are Laurent Orseau, scientist intelligence company artificial Google, Deep Mind, and Stuart Armstrong of the University of Oxford in the UK. Both argue that is “unlikely” robots “act optimally all the time”
And some of the recent technological inventions are increasingly questioning this possibility. Or at least put machines on the ropes, trying their reaction if they can attack humans.
The robot that created the artist and robotics specialist Alexander Reben University Berkeley, is a good example. It is a machine able to decide whether to inflict pain or not people , sticking a needle in your finger when so determines.
The aim of Reben is to deepen the debate on the possible dangers of artificial intelligence to humans.
“the real concern about artificial intelligence is that is out of control ” explains Reben the BBC . “The technology giants say that much remains to that, but it is better to think now, before it is too late”.
“I’m showing that (harmful robberies) are now a reality. We must confront switch off “Reben, who believes that explains” “being developed by Google could be useful, sooner or later.
” It is interesting to know what is proposed the off button, “says Reben.
“How can a robot would not be able to disable that button, if so smart?” , the specialist asks that defines its robot “harmful” as a “philosophical experiment”.
“the robot makes a decision that I, as its creator, I can not predict . I do not know whether or not it will hurt me, “explains
Artificial Intelligence.” Safe “
According Orseau and Amstrong, the goal of” big button red “shutdown would” prevent the machine follow a damaging sequence of actions to humans or the environment “.
the danger potential of intelligent machines for humans is a growing concern of scientists and roboticists . The founder of Tesla, Elon Musk, has said more than once.
But artificial intelligence is already part of our daily life, ever greater extent. And often their “intelligence” may improve ourselves.
The creators of the “red button” mentioned examples such as a robot that learned to stop the game Tetris to avoid losing . And this could be extrapolated to other situations more harmful to humans.
A Orseau seems understandable that people would worry about the future of artificial intelligence.
” it is healthy to be worried. But today, the level of our knowledge does not require that we “ says Orseau. “It is important to start working on safe artificial intelligence before there is any problem.”
A major challenge
However, according to the scientist, ” no system is foolproof ; it is a matter of doing it as good as possible. And this is one of the first steps “.
Noel Sharkey, professor of artificial intelligence at the University of Sheffield, UK, agrees. “Awareness of security is essential to (the development of) almost all computer systems, algorithms and robots,” he tells the BBC.
According to Sharkey, is equally important, “the ability to disconnect the system in an instant, as is always possible for a system of reinforcement learning to find shortcuts to remove “operator.
“What would be even better is that an artificial intelligence program could detect when something goes wrong and autodetenerse” says Sharkey. “That would have been very useful when Tay, the bot of Microsoft, began broadcasting racist and sexist comments on Twitter ” says the expert.
“But it is a huge scientific challenge ” he said.
No comments:
Post a Comment