ASI-Bench: new test reveals AI still leans hard on human guidance
A new benchmark built by 40+ experts across 60 project-level research tasks tests whether AI can explore the unknown and run science end-to-end, not just apply learned knowledge. When methodological guidance is withdrawn, the best agent scores drop from 50.9 to 26.6, showing current systems remain far from autonomous research. ASI-Bench is open to contributors.