How easy/fast is it for a AGI to hack computers/a human brain? — LessWrong