Facebook Bots Invent Their Own Secret Language, But Should We Fear a Robot Uprising?
On June 14, 2017, an article on TechCrunch described an experiment where Facebook researchers played with the idea of training chatbots to negotiate using their own language, rather than human language. The researchers used reinforcement learning algorithms to teach the bots the rules of the game and the limited negotiation language. The article interprets the bots’ divergence from human language as evidence that they “invented their own language.”
But in reality, these bots were simply negotiating effectively using a new language. Inventing a new language for bots can be a more efficient way to communicate and negotiate, and actually demonstrates the bots’ intelligence and creativity.
It’s important to remember that AI is still developing, and bots are capable of learning and inventing new languages. These advancements in AI may spark concerns about a robot uprising, but in reality, these fears are unfounded.
