x
Systems Analysis: AI Alignment and the Principal-Agent Problem — LessWrong