昆明理工大学学报(社会科学版)

随着ChatGPT的迅速发展，其在司法领域的应用再次引发了人们关于人工智能介入司法的思考。在新的人工智能技术介入下，加上对量刑规范化改革和法律形式主义的追求，人工智能辅助量刑系统得到进一步的发展。但是，由于数据的公开性和结构化不足、技术本身的复杂性和不可解释性、数据和算法设计者的价值偏见、技术的中立性和依赖性，致使在发展过程中人工智能存在数据、算法和应用上的风险，数据低质且匮乏、算法歧视和算法黑箱的存在、法官应用过程中过度依赖人工智能、侵害被告的诉讼权利。为此，建立智能辅助量刑系统应当确保数据的全面性和真实性、树立正确的算法伦理观、适度公开算法；法官应用过程中应当坚持法官主导审判、人工智能辅助裁判，为被告提供专家辅助人并保障被告的异议权。

With the rapid development of ChatGPT, its application in the judicial field has once again aroused people's thinking about the intervention of the artificial intelligence in the judicial field. With the intervention of new artificial intelligence technology, coupled with the pursuit of standardized sentencing reform and legal formalism, artificial intelligence assisted sentencing system has been further developed. However, the lack of openness and structure of data, the complexity and uninterpretability of technology, the value bias of data and algorithm designers, and the neutrality and dependence of technology—— all these may have caused such risks of data, algorithm and application in the development process of artificial intelligence as that of low quality and lack of data, algorithm discrimination and the existence of algorithm black box, excessive reliance on artificial intelligence in the application process of judges, and infringement of the litigation rights of defendants. The establishment of intelligent auxiliary sentencing system should ensure the comprehensiveness and authenticity of data, establish correct algorithm ethics, and moderately disclose the algorithm. In the process of application, judges should insist on judge-led trial, artificial intelligence assisted judgment, provide expert assistants for defendants and protect their right to dissent.