Abstract: Integrating large language models (LLMs) into automatic text scoring (ATS) requires a human-AI collaborative process design and system construction. The LLM-empowered human-AI collaborative ...