Reasoning of LLMs

Your friend is intrigued by the insightful answers provided by a Large Language Model (LLM) in their research project. They want to understand how the model arrives at these conclusions.

Since the LLM cannot directly annotate its responses, what strategy could you recommend to your friend to gain insights into the LLM's reasoning process?