TRACES asks a fascinating question: Can AI discover something nobody knows yet?
Most AI benchmarks test whether a model can retrieve information or solve problems with known answers. But real science is different β researchers work with incomplete evidence, test hypotheses, challenge assumptions, and search for conclusions that can be verified.
Thatβs what makes TRACES so promising. If AI can investigate unknown problems rather than simply reproduce existing knowledge, it could become a powerful research partner in medicine, biology, physics, materials science, and beyond.
The most important part is distinguishing real discovery from a lucky guess.
This could be a major step toward measuring not just how smart AI is, but how useful it can be for advancing human knowledge.
Most AI benchmarks test retrieval β can a model find the known answer? However, the hardest problems in science require discovery, can a system earn an answer nobody has yet?
Meet TRACES π§ β the world's first benchmark for measuring discoverative AI: AI that can work through evidence, test hypotheses, and reach verifiable conclusions on problems without answer keys. Proposed by our founder @tianqiao_chen, who defined its six capabilities.
Three things published today: a definition of "discoverative intelligence", a rubric to tell sound investigation from lucky guesses, and a open call for both solvers and problems
*Website: https://t.co/iJ5rDV6qz9