x
Testing ChatGPT for white lies — LessWrong