Trustworthy automated philosophy? — LessWrong