We recruited almost 20 law professors across the country to assess how well AI tutors perform in law school. In a first step, the professors and LLMs answer common contract law questions asked by students. In a second step, the professors compare whether they prefer the answers given by their colleagues or those given by LLMs. Because the professors don’t know whether the LLM authored the answer, this design is especially suited to determine how tutors stack up against professors in a “blind” evaluation. Ultimately, the project will yield important insights into whether AI tutors are ready to be rolled out in law school classes.