Reproduce “Large Language Models Can Explain Experts Better Than Experts Themselves” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mina Cho, Russell J. Funk, Alok Gupta, Mochen Yang
The authors test whether an LLM can turn tutoring conversations into reusable guidance, rather than relying only on experts’ written framework. GPT-4o uses conversation-derived guidance to answer mistakes and support novice tutors. In math tutoring, it improves response quality and reduces the expert–novice gap versus the human framework, but the task does not establish equivalence or broader transfer.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.