AI scientists tested with diverse researchers
A new benchmark tests AI research agents by pairing them with simulated researchers who have different expertise and preferences.
A new benchmark tests AI research agents by pairing them with simulated researchers who have different expertise and preferences.