Human ChatGPT Comparison Corpus (HC3)

HC3数据集是由上海财经大学信息管理与工程学院AI实验室创建，旨在比较ChatGPT与人类专家在开放域、金融、医疗、法律和心理学等多个领域的回答差异。该数据集包含近4万条问题及其对应的人类和ChatGPT的回答，通过人类评估和语言分析揭示了ChatGPT回答的特点及其与人类的差异。数据集的应用领域广泛，旨在提高大型语言模型在专业咨询和日常辅助中的表现，并探索未来语言模型的改进方向。

The HC3 Dataset was created by the AI Laboratory of the School of Information Management and Engineering, Shanghai University of Finance and Economics. Its purpose is to compare the response differences between ChatGPT and human experts across multiple fields including open-domain scenarios, finance, healthcare, law, psychology and others. This dataset contains nearly 40,000 questions along with their corresponding human-authored and ChatGPT-generated responses, and reveals the characteristics of ChatGPT's responses and their disparities compared to human responses through human evaluation and linguistic analysis. With wide application prospects, this dataset is designed to improve the performance of Large Language Models (LLMs) in professional consultation and daily assistance, and explore the directions for future improvement of language models.