x
A more effective Elevator Pitch for AI risk — LessWrong