AI Safety Newsletter #5: Two hopeful visions for our AI driven world