Tag
This study examines how LLMs suggest research methods (datasets, models, metrics) when prompted only with a research question, finding that LLMs exhibit a strong provider bias and propose a much narrower range of methods compared to actual papers, potentially narrowing researchers' methodological search space.