Everyone Thought FB Chatbots Were Going to Destroy All Humans Yesterday, But We're Still Here

It's a plot we've seen in many Hollywood movies, where the friendly robots realise they're just super-smart metal slaves to us mere humans and rise up to take over.

And yesterday, reports arrived that Facebook developers had to shut down their chatbots after they started speaking a new language that their developers couldn't understand.

It certainly sounded like the robots were showing off and behaving slightly suspect.

But the dialogue between the robots was anything but sinister. An excerpt shows that it was mostly gobbledygook.

Alice: balls have a ball to me to me to me to me to me to me to me

Bob: i . . . . . . . . . . . . . . . . . . .

Alice: balls have zero to me to me to me to me to me to me to me to me to

Bob: you i i i i i everything else . . . . . . . . . . . . . .

As Wired explained, the bots were programmed to learn how to negotiate through reinforcement learning and used some pretty sneaky tactics to feign interest in objects, only to give them up later. The most important point, however, is that these bots were programmed to do this, so they were still only working within their defined coding parameters.

This is why Wired believes Artificial Intelligence (AI) is still a far way-off from achieving autonomy from coding. Researchers are looking into ways to get AI to work outside their coding through human feedback systems.

So, while there might be more scary reports in the future predicting AI's eventual rise to the top, it probably won't come anytime soon...

Photo: 20th Century Fox