Imagine exploring the world's most complex software systems, like the Linux kernel, with the help of an AI assistant. That's the promise of Code-Survey, a groundbreaking new method that uses Large Language Models (LLMs) to analyze massive codebases. It's like giving a survey to thousands of developers at once, unlocking hidden knowledge about how software evolves. Code-Survey treats LLMs like survey participants, carefully designing questions to extract key insights from unstructured data such as commit messages and email discussions. This transforms chaotic information into organized datasets ripe for analysis. In a case study on the Linux kernel's eBPF subsystem, Code-Survey revealed surprising trends. For example, it pinpointed which kernel components are most prone to bugs and highlighted how new features impact overall stability. The analysis even uncovered hidden dependencies between different parts of the system. While traditional methods struggle to handle the sheer volume of unstructured data in large codebases, Code-Survey offers a powerful new approach. By combining AI with human expertise, we can gain a deeper understanding of how complex software systems work, leading to improvements in design, reliability, and security. Although still in its early stages, Code-Survey has the potential to revolutionize how we analyze and understand large-scale codebases. Future research aims to refine its accuracy and apply it to other major software projects, paving the way for more efficient and insightful code analysis.