This research examines how Large Language Models can be enhanced to function more like top-tier software engineers by developing better communication skills when generating code.

• Top software engineers ask clarifying questions to reduce ambiguity
• Research focuses on benchmarking LLMs' ability to communicate effectively in coding contexts
• Aims to close the gap between LLMs as functional coders versus skilled software engineers
• Proposes that better communication leads to better code generation

For engineering teams, this research matters because it points toward LLMs that don't just write code, but can participate more meaningfully in the software development process by seeking clarification when requirements are unclear.

HumanEvalComm: Benchmarking the Communication Competence of Code Generation for LLMs and LLM Agent