Thinking in another language could reduce biases — LessWrong