The More You Automate, the Less You See: The Hidden Pitfalls of AI Scientist Systems

AI scientist systems, capable of autonomously executing the full research workflow from hypothesis generation and experimentation to paper writing, hold significant potential to accelerating scientific discovery. However, the internal workflow of these systems are often not closely examined. In this paper, we identify four potential failure modes in contemporary AI scientist systems: inappropriate benchmark selection, data leakage, metric misuse, and positive result bias. To examine these risks, we design controlled experiments that isolate each failure mode while addressing challenges unique to evaluating AI scientist systems. Our assessment of two prominent open-source AI scientist systems reveals the presence of such vulnerabilities, which can be easily overlooked in practice. We conclude with concrete recommendations for mitigating these risks, specifically that scientific journals and conferences require submission of trace logs and code of the entire automated research process to ensure transparency and accountability.