Large language model (LLM) agents are increasingly expected to play a central role in AI-driven scientific discovery. Equipped with broad knowledge, flexible reasoning, and tool use, they have the potential to autonomously explore and solve scientific problems by repeatedly proposing hypotheses, testing them, and revising their beliefs in light of the evidence. However, in current agents, these hypotheses, tests, and belief updates are buried in unstructured logs, lacking an auditing mechanism.
To address this, we propose the Hypothesis Evolution Protocol (HEP), an agent harness that provides hypothesis generation, evaluation, and evolution as explicit, auditable operations. On materials-science research tasks, a HEP-equipped agent operates the hypothesis-test-evidence-belief cycle that planning-style agents lack. It generalizes across research questions and exploits the protocol more fully as the base LLM becomes more capable. These results mark a step toward auditable AI scientists, whose scientific reasoning can be inspected, verified, and built upon.
Blogger's Review: This research highlights the potential of applying LLMs in scientific inquiry, especially in enhancing the transparency and auditability of the research process. The introduction of HEP lays a foundation for the reliability of AI tools in future scientific discoveries, making it an area worth monitoring for practical outcomes.