More Than Can Be Said: A Benchmark and Framework for Pre-Question Scientific Ideation
概要
arXiv:2605.06345v1 Announce Type: new Abstract: AI research agents have shown strong potential in automating literature search and manuscript refinement, yet most assume a clear and actionable initial input, operating only after a research question has been made explicit. In contrast, human researc…
