Facebook has experienced what has to be the first uprising from the computers recently. The social media giant’s AI started chatting in code words that were difficult to decipher, which was eventually abandoned.
Facebook Artificial Intelligence Research (FAIR) started an experiment to let AI bots negotiate with each other, which would enable them to learn “complex communication and reasoning skills. Two separate experiments were conducted in this regard — the first one lets the bots communicate in English, which could be understood by humans. The second experiment would remove the language barrier, i.e. let the bots talk in a language they wish.
The second experiment was done to make the bots learn and develop communicating skills. However, on inspection, they found that their communication was becoming conspicuous. Consider the following sample from the conversation:
Bob: “I can 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.”
Now, it is clear that the statement’s meaning cannot be understood. Therefore, the researchers decided to abandon this experiment before these AI bots started discussing something that should not be discussed. The researchers have still not been able to decipher what the conversation was on about.
Although it was good for development in the field of AI, it was certainly difficult to understand what the bots were talking about. Were they talking about preserving the planet by getting rid of the humans? Who knows!
