Frankenstein is a famous novel written in 1810, wherein a scientist creates a monster, by accident, during a secret scientific experiment. That monster then wreaks havoc all across the city, something which humans cannot stop.
Something similar was witnessed recently at Facebook AI Research Lab (FAIR), where an experiment was hastily shut down, as the scientists involved panicked.
In the experiment, a few Artificial Intelligence powered robots created their own language, and started chatting using that language, which humans couldn’t understand. The chat was creepy all right, but the fact that scientists working on the experiment had no control over the communication initiated by the robots makes it really scary and weird.
Is this the rise of AI powered bots, who can take over the world one day?
Or, are we over-reacting?
The Experiment Which Scared AI Scientists
Few weeks ago, Facebook started working on an experiment to develop a generative adversarial network , wherein they planned to develop AI bots which can carry on negotiations on their own.
Facebook Artificial Intelligence Research team published in their blog, that they are attempting to make it “possible for dialog agents with differing goals (implemented as end-to-end-trained neural networks) to engage in start-to-finish negotiations with other bots or people while arriving at common decisions or outcomes.”
Just like humans negotiate, several times in a day – negotiation between husband and wife, regarding which TV channel to watch; or negotiating with your boss to approve a leave, or bargaining with a shopkeeper to reduce price.
The AI Bots, in the experiment, were designed to negotiate, just like humans do, to split responsibilities, to arrive at conclusion, and to work.
Eventually, Facebook aimed to deploy these bots to communicate with advertisers, users and customers, and to drive business.
So far so good. But what happened next shocked everyone.
The Creation of Digital Frankenstein
When two AI bots named Alice and Bob were asked to talk with each other, the experiment progressed as per the plan.
But then, something happened with baffled all the scientists involved: The bots devised their own language while chatting, and they started talking something which the scientists never understood.
Here is the actual chat which happened between Alice and Bob:
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 to me
Later, scientists working with Facebook described this chat as scary, but exciting.
The report said, “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..”, adding, “After learning to negotiate, the bots relied on machine learning and advanced strategies in an attempt to improve the outcome of these negotiations,”
The experiment was shut down, as the team panicked, unable to understand how and why these bots have talking in their own language. Were they planning to take over the world of humans, and establish their own empire?
Considering that our actual knowledge on how AI thinks, and what thought process goes inside the mind of a bot is still relatively unknown, this experiment does raise some valid questions.
Do you think that Elon Musk was right when he said mechanism to control AI should be devised first, before further experiments?
Do let us know by commenting right here.
"Facebook Panics and Shuts Down Experiment After AI Bots Create Their Own Language for Chatting. Was This an AI Uprising?",