Study: LLMs are prone to social identity biases similar to the way humans are

But fortunately:<p>&gt; AI biases can be reduced by carefully selecting the data used to train these systems