2 ms·Study: LLMs are prone to social identity biases similar to the way humans are1 points by giuliomagnifico 2y agogiuliomagnifico 2y agoBut fortunately: > AI biases can be reduced by carefully selecting the data used to train these systems