I wanted to give some methodological reflections on studying Large Language Models- not studying how to make them or improve them, but studying how they behave, why they behave that way, their capabilities, and the philosophical implications of this.
