x
AGI safety from first principles: Conclusion — LessWrong