Isaac Asimov proposed the Three Laws of Robotics, but the technology proposed back in the 1950’s was not what we have now. Researchers are saying, logically, that if an AI is beyond human intelligence, then we really have no way of restricting it, because it is too complex for us to fully understand.
If you know about neural networks, which are the basis for modern artificial intelligence, you know that is is a series of coefficients trained on data. When you have a dozen layers of dozens of coefficients it is really beyond the the interpretation of a human mind.
From the article: “The alternative to teaching AI some ethics and telling it not to destroy the world – something which no algorithm can be absolutely certain of doing, the researchers said – is to limit the capabilities of the super-intelligence. It could be cut off from parts of the internet or from certain networks, for example.”
What did you expect? Think you can control something that is smarter than you? We are in deep shit.