Just days after the announcement of Tesla and SpaceX CEO Elon Musk that AI is the biggest risk, now Facebook went as far to shut down one of its AI systems after the chatbots started making their own language breaking the previous restrictions provided to them.
The report from Tech Times says that the company had to pull the plug since they thought that things got “out of hand”.
While the AI did not do anything radical like most people expect it to, it developed a new language and started using it instead of using English and thus defying the purpose. This made Facebook researchers shut down the AI systems to speak in English.
It all started in June when the researchers at Facebook AI Research Lab (FAIR) saw that while they are busy with their working of improving their chatbots, the “dialogue agents” started creating their own language.
After few days, the Bots started communicating in their own language breaking the prefixed norms by news ones completely made by them.
The researchers initially allowed these “dialogue agents” to communicate with them in the hope that this could increase their conversational skills.
The researchers also found that these bots have really good negotiating skills.
In the report, they said, “After learning to negotiate, these bots relied on machine learning and other advanced strategies in an attempt to improve their outcome of these negotiations.”
“Over time, the bots have become quite skilled at it and they even began gaining interest in one item in order to ‘sacrifice’ it at a later stage in their negotiation as a faux compromise,” it added.
Mean while Elon Musk tweeted the following “I’ve talked to Mark about this (AI). His understanding of the subject is limited.”