While we’ve seen a multitude of science fiction books, movies and TV shows cover the idea of artificial intelligence, it never ceases to feel like a new and scary idea. We’ve seen Skynet defeated in five Terminator movies now, but the reality of a free-thinking robot is all the more unsettling when we are told it is just around the corner.
While this may be the same as when people of the mid-20th century believed we would be taking our vacations on the moon for a round of golf by now, I can’t help but believe that artificial intelligence could happen within our lifetimes. It feels like it should always be set in the future, but it may be closer than we think. Facebook, now under the ownership of Google, are pushing groundbreaking technology in a number of ways. Part of their current slate of technological research is into the communication between artificial intelligence. If two AI can talk to one another, they can negotiate and solve a problem even more efficiently than before. But things got a little out of hand recently, when they started to speak in private. While the staff of Facebook were busy trying to improve chatbots, the two AI, referred to as “dialogue agents” were busy creating their own language. The bots, nicknamed Bob and Alice, started off communicating in English, but then started speaking in what appeared to be gibberish, but was later revealed to be a language we do not understand.Bob started saying “I can i i everything else” and Alice said “balls have zero to me to me to me” This abrupt switch left the scientists out of the loop, with only the other AI being able to understand. “The AI did not start shutting down computers worldwide or something of the sort, but it stopped using English and started using a language that it created,” the official report read. The experiment was stopped and the AI were turned off at this point. Basically, what the bots were doing was making code words to make communications more efficient. If Bob says one word five times, he’s saying that he wants five of those items. Because of the way we understand language, this sounds too complicated, but it is actually faster and more logical. Dhruv Batra, a researcher for Facebook Artificial Intelligence Research (FAIR), explained that they understand each other in ways we are not smart enough to process quickly: “Its definitely possible, its possible that [language] can be compressed, not just to save characters, but compressed to a form that it could express a sophisticated thought. Its perfectly possible for a special token to mean a very complicated thought. The reason why humans have this idea of decomposition, breaking ideas into simpler concepts, its because we have a limit to cognition.” The bots go on a reward and benefit principle, so if a means of negotiating is less efficient, they find another way to do it. Eventually there was no “reward” in using the English language, so they started to communicate differently. Tesla CEO Elon Musk has said AI is the “biggest risk we face”, and has asked America’s governors to implement regulations on the technology. “I have exposure to the very most cutting-edge AI, and I think people should be really concerned about it” Musk said at theNational Governors Association Summer Meeting in Rhode Island. But Mark Zuckerberg, CEO of Facebook, disagrees: “I think people who are naysayers and try to drum up these doomsday scenarios — I just, I don’t understand it. It’s really negative and in some ways I actually think it is pretty irresponsible”. While there isn’t a clear consensus, this kind of development definitely freaks me out, just a little bit. I don’t like people talking behind my back as it is, so I can’t say I’m comfortable with secretive robots either. But who know how they would respond to being able to understand the world. Maybe they won’t even want world domination, like the time thisrobot cop committed suicide by drowning itself in a mall fountain.