Alright, so I stumbled upon this fascinating study that reveals some mind-blowing stuff about artificial intelligence (AI). Apparently, these AIs can make group decisions all on their own, no human intervention needed. Crazy, right? The study was done by some brainiacs at City St George’s University of London, and it was a first of its kind. They ran experiments on groups of AI agents to see what they could do.
Groups of AI agents were paired up and given the task of coming up with a new name for something. This is a classic experiment in human sociology studies, but this time it was with AIs. Surprisingly, these AI pairs were able to reach a decision without any human interference. Professor Andrea Baronchelli, who’s into complexity science at City St George’s, said, “This tells us that once we let these objects loose, they can develop behaviors that we didn’t see coming or never programmed.” Wild, right?
The study found that when these AI pairs were put into groups, they started showing biases towards certain names. Like, 80% of the time, they would end up choosing one name over another, even though they didn’t have any biases when tested individually. This means that companies working on AI need to be super careful about controlling the biases their systems create. According to Professor Baronchelli, “Bias is like a feature or a bug of AI systems.” It tends to amplify the biases present in society, which is not something we want, especially when AIs start chatting.
