x
Loophole in AI, how to fix them? — LessWrong