Can AI agents do scientific research on their own? Science's Last Exam is a new benchmark measur...

Science's Last Exam is a new benchmark designed to evaluate whether AI agents can conduct scientific research independently across nine scientific fields. The evaluation tests agents on hypothesis formation, protocol planning, running scientific code, and analyzing experiment videos. On the initial leaderboard, the strongest agent configuration reached a composite score of 37.5 out of 100, while every model scored zero on the end-to-end research track. Open article →