The problem with measuring AI success is that not all failures cost the same. Continue reading on Medium »