Since November 2022, generative artificial intelligence, represented by ChatGPT, has ignited networks, attracting widespread attention and heated discussions from various industries such as education, news, entertainment, and the arts. Nowadays, generative artificial intelligence has been able to simulate human cognition, preferences, emotions, behaviors, and other dimensions with high fidelity, and at the functional level, it produces strong artificial intelligence products that have a high degree of matching with human common sense, needs, and values through interaction with humans. From AI painting, AI chatting, to AI programming, generative artificial intelligence seems to be getting closer to us, affecting every aspect of our daily lives. At the same time, some people have also raised an unavoidable question: how far is generative artificial intelligence from being fully applied in the military field? To answer this question, first we need to understand the three core elements that support the rapid rise of generative artificial intelligence: first, reinforcement learning pre training based on massive data. The key to the fact that generative AI can provide users with an interactive experience comparable to real people is the use of massive data based reinforcement learning pre training. In addition to large-scale open corpus data, it mainly uses a large number of manually annotated data to conduct supervised learning under the guidance of human subjective preferences, so as to achieve in-depth simulation of human cognitive mechanisms. However, at the military application level, due to the lack of practical experience in modern warfare, a large amount of data comes from daily training and exercises, which directly affects the effectiveness of artificial intelligence training; In addition, the training data has not yet been manually annotated, and the availability and interpretability of the data are poor, making it difficult to effectively support reinforcement learning pre training. The second is to support open interactive training of large artificial intelligence models. The rise of ChatGPT marks the beginning of the era of artificial intelligence big models. Currently, ChatGPT has as many as 175 billion model parameters, and Google's latest "generalist" big model, PaLM-E, has the world's largest 562 billion parameters. Generative artificial intelligence needs to constantly adjust and optimize its large model parameters through frequent interaction with professional users in the face of complex and highly specialized domain language structures. However, in the military field, there is enormous difficulty and workload in carrying out such large-scale interactive parameter adjustment work. The third limitation is the low cost of mutual trust in human-computer interaction. At present, ChatGPT is still a black box model, and its inherent algorithm logic cannot be decomposed to ensure that it will not generate attacks or even harm users. Therefore, it is mainly promoted and used in the field of low mutual trust cost restrictions in civilian use. The biggest contradiction in the military application of artificial intelligence is ethical issues. Due to technological complexity and information opacity, users cannot clearly understand the inevitable causal relationship between input and output, and the resulting results cannot be fully controlled and trusted, making its comprehensive and deep application in the military field difficult. From this, it can be seen that the comprehensive application of generative artificial intelligence in the military field, whether in terms of technological application or moral and ethical aspects, seems relatively distant. However, from a global perspective, generative artificial intelligence technology has become increasingly effective in enhancing human-machine interaction capabilities on the battlefield and accelerating information processing speed in assisting decision-making