ForeSci
Emerging1papers using it
2026first seen
ForeSci is a temporally controlled benchmark containing 500 tasks across four AI domains, used to evaluate the ability of LLM agents to make forward-looking research judgments based on historical evidence.