← back to paper
arxiv: 2607.27191 · 2 revisions
Can AI agents conduct open-ended AI research? Early evidence from two case studies