生成式人工智能研究进展  

Progress in Generative Artificial Intelligence Research

作　　者：王芳[1,2] 朱学坤[1,2] 刘清民 岳之楠 王美权 张馨月 杨天德 马鑫 张超 Wang Fang;Zhu Xuekun;Liu Qingmin

机构地区：[1]南开大学商学院信息资源管理系,天津300071 [2]南开大学网络社会治理研究中心,天津300071

出　　处：《图书与情报》2024年第4期45-64,共20页Library & Information

基　　金：国家社会科学基金重大项目“基于数据共享与知识复用的数字政府智能化治理研究”(项目编号:20ZDA039)研究成果之一。

摘　　要：快速地技术迭代与广泛应用表明生成式人工智能在驱动经济社会发展方面具有巨大潜力,但同时它也可能带来多种挑战与风险。在系统回顾相关学术文献的基础上,文章梳理了生成式人工智能的关键技术、技术发展历程及应用场景,分析了生成式人工智能在数据训练、算法模型、内容利用等方面存在的风险,总结了生成式人工智能风险治理的策略,包括实施全生命周期的数据质量控制,提高AI风险治理的技术能力,对组织结构进行适应性优化,强化问责机制和标准约束,同时提出未来生成式人工智能发展应坚持发展创新与风险治理并重的原则,完善政策法规与技术标准体系,建立多元包容的国际合作共治体系,充分激发企业活力。本文可以为生成式人工智能领域的学术研究、技术开发和政策制定提供参考借鉴。The rapid iteration and wide application of generative artificial intelligence(GAI)technology have demonstrated its immense potential in driving economic and social development.At the same time,it also brings challenges and risks.Based on a comprehensive review of relevant literature,this paper identifies the key technologies,development history,and application scenarios of GAI,analyzes the risks in data training,model building,and content generation,summarizes the risk governance strategies for GAI,such as implementing data quality control throughout the entire lifecycle of GAI,enhancing technical capabilities to manage AI risks,optimizing organizational structures for adaptability,and strengthening the constraints by accountability mechanism and technical standards.Additionally,this paper proposes that the future development of generative artificial intelligence should adhere to the principle of balancing innovation with risk management,improve the system of policies,regulations,and technical standards,establish a diverse and inclusive framework for international cooperation and governance,and fully stimulate the vitality of enterprises.This study is expected to provide reference for GAI related research,technology development,and policy making.

关 键 词：生成式人工智能 大语言模型 风险治理 AIGC AI幻觉 政策分析 

分 类 号：G354.5[文化科学—情报学]