I thought that was clear, try to show biases in LLMs with a concrete example.