Into the Unknown: Benchmarking AI agents on open scientific research problemsarxiv.org2 points·matt_d··0 commentsOpen articleSaveView on HN