This is a great question that gets to the heart of how language models make decisions. The short answer is: **because