If you were to watch a reality cooking show like Hell’s Kitchen for the first time, you might be a bit confused by the cooking terms and shorthand that chefs use to communicate with each other. To you it might sound like gibberish, but to them it makes perfect sense, and that’s exactly the kind of situation that Facebook accidentally created with its AI efforts.
So much so that Facebook was forced to shut it down after their AI started inventing its own language to communicate with each other. It might sound like the plot to sci-fi movie where robots take over the world, but it’s precisely what happened. According to Dhruv Batra, visiting research scientist from Georgia Tech at Facebook AI Research, this happened after the researchers made a mistake during the programming phrase, where they did not include rewards for the AI to stick to English.
According to Batra, “Agents will drift off understandable language and invent codewords for themselves. Like if I say ‘the’ five times, you interpret that to mean I want five copies of this item. This isn’t so different from the way communities of humans create shorthands.” That being said, it is more dramatic than what it seems.
The reality is that the program was shut down because Facebook was developing the AI that were capable of talking to people, and obviously creating its own language means it failed on that front. However it does have traces of what some prominent figures in the tech industry fear, like Tesla and SpaceX’s Elon Musk, who has expressed concern over AI leading to a Terminator-like future, and who recently criticized Facebook’s CEO of having a “limited understanding” of the dangers of AI.