Arshavir Blackwell, PhD, is an AI researcher and writer exploring how large language models think—and how we can make them understandable to humans.