【摘要】 生成式人工智能已经成为人工智能的最前沿技术之一，被认为是通往通用人工智能的重要阶梯。对2023年生成式人工智能的主要发展动向进行了综合评述。首先盘点生成式人工智能技术当前发展状态；其次概述生成式人工智能技术未来发展趋势；最后探讨生成式人工智能的军事应用潜力。综述表明，多模态性和可解释性是当前生成式人工智能技术的热门方向，生成式人工智能技术的应用使强调指令理解推理的机器人、强调归纳推理的基础科学等领域不断取得突破，面向终端用户的“轻量化”和面向开发者的“开源化”势头正盛，与此同时，对生成式人工智能技术应用的监管也迫在眉睫。综述认为，深度融合多模态、建设高质量数据资源及评估手段、强调与环境自主交互、在边缘侧低成本应用部署、平衡技术发展与规制治理等，将是生成式人工智能技术在未来的重要发展方向。更多还原

【Abstract】 Generative Artificial Intelligence(GAI) has become one of the cutting-edge technologies in Artificial Intelligence(AI), and is considered as an important step towards Artificial General Intelligence(AGI). This paper makes a comprehensive survey of the main development trends in the field of GAI. Firstly,the development trends of GAI are reviewed; Secondly, its further development directions are summarized;Finally, the potential military applications of GAI are explored. This survey shows that multimodality and interpretability are currently the key directions of GAI; GAI is now accelerating the breakthroughs in domains of robots that emphasize instruction comprehension and reasoning, as well as basic science that emphasizes inductive reasoning; lightweight GAI for end-users and open-source GAI for developers have become popular trends; and the supervision of the application of GAI is also urgent. This survey believes that deep integration of multimodality, construction of high-quality data resources and evaluation methods, autonomous interaction with the environment, low-cost application deployment at the edge, and balancing technological development with regulatory governance, will be important directions for the future development of GAI.更多还原