Facebook’s AI creating its own language is nothing to be afraid of

The web was abuzz with chatter about the opportunities and dangers of advancing AI in June, when Facebook’s Artificial Intelligence Research (FAIR) Lab shared a finding that its bots had invented their own language to communicate with each other.

That story made waves, and so did several reports based on a follow-up piece by Fast Co. Design. Several blogs ran with headlines stating that Facebook had shut down its Frankenstein-like AI after it created its own language.

In reality, there’s nothing to be worried about just yet. Let’s take a closer look at what those bots were saying to each other:

Bob: i can i i everything else . . . . . . . . . . . . . .
Alice: balls have zero to me to me to me to me to me to me to me to me to
Bob: you i everything else . . . . . . . . . . . . . .
Alice: balls have a ball to me to me to me to me to me to me to me
Bob: i i can i i i everything else . . . . . . . . . . . . . .
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 . . . . . . . . . . . . . .
Alice: balls have 0 to me to me to me to me to me to me to me to me to
Bob: you i i i everything else . . . . . . . . . . . . . .
Alice: balls have zero to me to me to me to me to me to me to me to me to

It might read like nonsense, but it’s essentially just two machine learning devices negotiating how to split a bunch of items (like books and basketballs) into sets.

However, the problem with this sort of interaction is that we can’t make much sense of it, at least not beyond the realm of research. FAIR researcher Mike Lewis told Fast Co. Design, “Our interest was having bots who could talk to people.”

As Gizmodo noted, the team had realized that they hadn’t incentivized the bots to speak in English that humans would be able to understand fully, and so they shut down the conversation. That’s all it was, and nothing close to a reality of being ruled by sentient beings.

Debunking posts around the web proclaiming that doomsday was near, FAIR visiting researcher Dhruv Batra explained in a Facebook post:

While the idea of AI agents inventing their own language may sound alarming/unexpected to people outside the field, it is a well-established sub-field of AI, with publications dating back decades.

Simply put, agents in environments attempting to solve a task will often find unintuitive ways to maximize reward. Analyzing the reward function and changing the parameters of an experiment is NOT the same as “unplugging” or “shutting down AI”. If that were the case, every AI researcher has been “shutting down AI” every time they kill a job on a machine.

In short, there’s no real reason to be afraid of AI taking over our planet anytime soon. I side with Neil DeGrasse Tyson on this one, since he points out that any AI we create – no matter how dangerous – can simply be unplugged.

The real lesson here, is to read research papers more thoroughly when you can.