My policy on using LLMs in academic work

I firmly believe that human thought and guidance is essential to conducting and communicating research, and should therefore not be replaced by machines. The capabailities of frontier LLMs to produce complete mathematical arguments, and even intuition about a proof or line of research, have turned into powerful and useful tools in research. While I’m figuring out how to use these tools productively in my academic work, I am adopting the following guidelines:

• I choose to formulate and develop my research program myself.
• I only incorporate LLM-aided or -generated arguments in my work once I have fully digested them and can reproduce them using my own words and thoughts.
• I refrain from using any LLM output verbatim in my writing, whether in research, teaching or personal writing.
• I conduct peer review myself, using LLMs only to check technical arguments provided the paper is already available publicly.