Earning money with/for work in AI safety — LessWrong