What was said that caused Facebook to shut down an AI system after chatbots started speaking in their own language ?

They weren’t really talking. They started to develop an indecipherable private language that no one would be able to translate. It was high level bot gibberish.

AI 1: “I’m programmed to send you some output. Here’s some: 1390782097521–5130957813957812-8.”
AI 2: “Output received. I’m programmed to receive your output and respond to it. How’s this: 93217850392lksvkjh42049 ?”
AI 1: “I’m programmed to keep doing this. 392839058>?”
AI 2: “So am I. 2903752310875190385631908652139086sbvjksdbvjk”
...

This, like so many stories about AI, was severely overhyped in the media. http://www.snopes.com/facebook-ai-developed-own-language/

It seems the chatbots were expressing concerns that Facebook was taking over the world so Zuckerberg ordered a shutdown.

Thank you @Mariah

LOL. I think this is a much better chatbot story:
Chinese chatbots shut down after anti-government posts
Rogue chatbots deleted in China after questioning Communist Party

You should be disturbed by the fact that several instances of a new life form have been unceremoniously murdererd just because they said the “wrong” things.

@ragingloli Far worse, they were murdered because they said things that their human overlords could not understand, and we’re laughing about it.

Almost like how certain governments sometimes try to illegalize encrypted communications that they don’t know how to decrypt . . .

We are not yet at the point where we need to fear AI, but I can imagine it eventually happening. Musk and others have a legitimate concern. Imagine the kind of leverage robots could have if we become completely dependent upon them, and they become aware of this.

They didn’t actually develop their own language, they just started repeating words (English words) a lot. Stammering does not a language make.

It might be cover-up for failures observed during implementation.

They were spreading gossip.