Will AGI surprise the world? — LessWrong