Facebook Shuts Down AI System
Facebook’s artificial intelligence experiment recently shut down one of it’s chatbots, considering two robots started speaking to each other in their own language. Instead of using codes, they went against them and made changes that only made sense to the chatbots. Researchers felt like they had no choice but to pull the plug.
The chatbots didn’t do anything obscene, such as shutting down computers or acting out. Instead, the two chatbots randomly stopped speaking in English and started talking to each other in their own language. In June, Facebook AI Research Lab (FAIR) researchers found out that while they were trying to improve the chatbots, the “dialogue agents” created their own language without any human input. They also started using machine learning algorithms, which resulted in a more fluid language for the chatbots.
These unlikely and strange conversations between the robots were realized when the researchers tried to teach the chatbots how to negotiate with each other by trading balls, hats, and books for specific prices. The robots ended up chanting at each other in a language that the researchers could not understand. Since the researchers didn’t command that they needed to speak in English, the robots decided to use their own “short hand,” if you will.
Even though this might come as exciting news and a sign of improvement for AI, other scientists and business owners are fearful for the direction AI is headed considering the possibilities of unsafe environments. For example Tesla’s Elon Musk expressed his concerns for the future of AI after hearing about this.
Musk even stated that AI is the “biggest risk we face as a civilization.”
The experiment shut down one of the robots in the meantime. We will let you know if Facebook decides to continue it despite the risks at hand.