The information should make us think. For the first time, two computers managed to communicate with each other in a language they had created themselves. A language perfectly indecipherable to man.
Ct happened in the labs of Google Brain, a research program in artificial intelligence designed to give computers more autonomy through the power of their algorithms. Researchers from the Mountain View firm conducted a strange experiment. They undertook to make two computers communicate. One of the machines is called Bob, the other, Alice. So far, nothing exceptional. A third computer, named Eve, was designed to intercept their communications. So far, the research protocol is very simple.
Things go wrong when researchers find that Bob and Alice communicate with each other using an encrypted language they themselves have developed. But the researchers hadn't put any specific algorithms in their protocol for this purpose. The strange thing is that this communication between Bob and Alice remains perfectly hermetic to Eve, whose mission is to intercept their communication. The two computers speak a language that only they understand.
The AI has just taken a giant leap forward. Computers would be able to invent their own language of communication, a language indecipherable to us poor humans. A breakthrough that is a little disturbing and puts the alerts big names like Bill Gates, Stephen Hawkins or Elon Musk in perspective: "AI is either the best or the worst thing that ever happened to mankind."
