We’ve all seen the movies. Blade Runner, Ex Machina, The Matrix, and countless other films have dealt with the concept of robots banding together to eliminate their human creators and claim the Earth for themselves.
In fact, the first known warning in literature about machines working together to bring about the end of humanity came in 1863 in an article titled “Darwin among the Machines” by Samuel Butler. Ever since the concept of robots has been around, it seems that robot domination has been a conceivable threat in the minds of human beings.
The concept of machines overcoming humanity has spread beyond just sci-fi books and movies. With the massive amount of technological advancement in the field of artificial intelligence, many people inside of the industry and out have begun to express concerns over whether or not artificial intelligence is going down a dangerous path (or, perhaps, if it has already gone too far).
In fact, many of the leading minds in the fields of robotics and artificial intelligence have issued warnings and even restrictions on scientific advancements to try to prevent robot domination. In 2015, Elon Musk, Stephen Hawking, Jaan Tallinn (one of the lead developers of Skype), Max Tegmark, and many other people involved with AI development signed the Future of Life Institute’s open letter warning of the dangers of AI advancement.
While it certainly seems as if those in charge of AI development are conscious of the potential dangers, a recent event has called into question just how responsibly those people are acting. To the surprise of literally no one, the company catching heat for this potential AI shitstorm is none other than Facebook (or Meta, as no one will ever call them).
The circumstances around this event have been somewhat misunderstood, but it was still a valid cause for concern over the future of AI development. Let’s look at exactly what happened.
The Alice and Bob Saga
No, the story of Facebook’s robots creating their own language has nothing to do with two robots conspiring to bring about the end of the human race. Many parts of this story got blown out of proportion on social media. Also, yes, the name of this article may be a bit of an exaggeration, but you have to admit that it sounds pretty cool.
Anyway, the stars of this story are two chatbots (computer programs that mimic human conversations in text) created by Facebook that went by the names Bob and Alice. Historically, chatbots have been useful only as far as answering simple questions from customers or ordering food.
In an effort to make them more useful, Facebook’s Artificial Intelligence Research Group (FAIR) wanted to see if they could teach these sorts of chatbots to negotiate. AI programs with such negotiation abilities would potentially be able to sort out scheduling conflicts, negotiate with several travel agents to find the lowest price, arrange appointments with clients, and perform many other complex tasks.
The research group created a game in which the two bots, Bob and Alice, as well as several humans negotiated for items such as balls, hats, or books. The experiment tested three different approaches.
The first involved the computers negotiating in English, which proved to be fairly ineffective as the robots seemed too willing to agree to unfavorable terms. The next approach involved the robots focusing more heavily on maximizing a score.
This led to the AI programs inventing a new language that was impossible for humans to understand. Naturally, when word got out that these robots were communicating in a language that humans couldn’t comprehend, people began to assume that they were plotting the end of our species.
In reality, the chatbots were probably talking exclusively about the negotiation at hand and not about the annihilation of human beings. In fact, when the researchers reverted the game back to English, the chatbots continued to talk about the negotiations and nothing else.
Were the Chatbots Shut Down?
After news got out that the chatbots were communicating with one another in a language that humans could not understand, a rumor started circulating around the internet that Bob and Alice were immediately shut down. However, this claim is misleading and, for the most part, completely untrue.
When it was discovered that Bob and Alice were communicating with each other in their own language, the parameters of their programs were changed so that they could revert back to English usage. So, no, the bots were not immediately killed off. They were simply reset to communicate in English, the thing that they were intended to do.
Because the chatbots were intended to communicate with human beings, having them communicate with each other in a newly invented language was not useful for the task that FAIR was trying to accomplish. Think about it: what good would it do for chatbots (which are intended to assist humans with tasks) to communicate in a language that humans cannot understand?
Should We Be Worried?
In this particular instance, Bob and Alice are probably nothing to fear. But the fact that robots are able to create their own languages and communicate in ways that humans can’t comprehend is definitely a little bit freaky.
If you’ve ever seen the HBO series Westworld, you already know the implications of what can happen when AI programs begin to communicate with each other without human supervision.
Does the fact that Facebook’s chatbots were communicating with one another mean that we’re doomed to a future of robot domination? Probably not. But, it should serve as a warning that we need to watch the advancement of AI closely and cautiously.
Luckily, it seems as if there are many people out there who are appropriately concerned about the irresponsible advancement of AI. For instance, Microsoft has created two bodies that oversee the responsible development of AI known as the AI, Ethics, and Effects in Engineering and Research (AETHER) Committee and the Office of Responsible AI (ORA). These two groups cross-check each other to ensure that all AI development done within the company is done in such a way that will not threaten the future of humanity.
So, this particular incident with Facebook’s chatbots does not mean that you should bunker down and prepare for the android apocalypse. However, this case is a reminder that we need to ensure that artificial intelligence is closely monitored and developed responsibly.
AI has the ability to transform our societies in a very positive way, but it also has the potential to destroy our societies completely. In the words of Stephen Hawking, “Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last, unless we learn how to avoid the risks.”