Does Asking Clarifying Questions Increases Confidence in Generated Code? On the Communication Skills of Large Language Models

As the responsibility of software developers encompasses more than just writing code, current Large language models (LLMs) have significantly improved the ability LLMs cannot fully replace professional software developers [4, 29]. to perform tasks in the field of code generation. However, there At a high level, the gap lies in several critical aspects of software is still a gap between LLMs being capable coders and being top-tier development beyond coding, such as effective communications, software engineers. Based on the observation that top-level software requirements, design, domain knowledge, and the broader context engineers often ask clarifying questions to reduce ambiguity of relevant projects and components, etc. [23, 31, 32, 35]. In this in both requirements and coding solutions, we argue that the same paper, we are interested in applying the communication lens to should be applied to LLMs for code generation tasks. By asking inspect the gap, given that effective communication is a critical probing questions in various topics before generating the final code, capability that connects all of the above-mentioned parts to coding.