学进去

The Google Brain team has been working on computing systems called “neural networks”—systems that were designed based on how neurons work in the human brain.

This time, they gave a mission to three of the neural networks, which they named Alice, Bob and Eve. Each of the networks had its own job—Alice sent messages to Bob, Eve tried to “eavesdrop”(偷听) and find out the messages, and Alice and Bob had to figure out a way to hide the messages from Eve. All the help that Alice and Bob got from the researchers before the mission began was made up of a set of numbers, which Eve didn’t have access                    to.

At first, Alice was not very good at sending secret messages. But slowly it worked on a way to encrypt(加密) them—putting information into a special code so that others could not understand it if they got the information—using he numbers given by researchers. And after practice, Bob also came to be able to decrypt(解密) Alice’s messages. Without the numbers or keys, Eve failed to understand Alice’s “speech” most of the time.

This test is considered a big step in the development of computers’ learning skills. “Computing with neural networks on this scale has only become possible in the last few years, so we truly are at the beginning of what’s possible, ” Joe Sturonas of US encryption company PKWARE told New Scientist magazine.

Unfortunately, the test happened only one week after UK physicist Stephen Hawking, While speaking at Cambridge University, warned how AI(artificial intelligence)could develop a will of its own. This could be “either the best or the worst thing ever to happen to human being”, he said.

But just as Sturonas pointed out, no matter what the possibilities of computers are in the future, they are just starting out. We still have plenty of time to work out a solution before they get anywhere near becoming a threat to humanity.