ChatGPT is a generative artificial intelligence (AIGC) chatbot program launched by the American artificial intelligence company OpenAI. It was launched on November 30, 2022. It quickly attracted global attention after its launch and is widely regarded as an important milestone in the development of generative artificial intelligence technology and industry.
ChatGPT uses a large language model based on the GPT-3.5 architecture and is trained through reinforcement learning. The amount of pre-training data reaches 45TB and the corpus vocabulary reaches 300 billion, which can simulate natural human conversations with a high degree of reality. As of January 2023, the ChatGPT program has more than 100 million active users. Microsoft is highly optimistic about the development prospects of ChatGPT. It has announced the launch of the latest version of Bing search engine and Edge browser supported by ChatGPT, and is exploring the integration of ChatGPT into its office suite. It plans to launch the new Windows 12 operating system in 2024. Access a large number of AI applications.
ChatGPT is both practical and functional, and the industry has high hopes for its application prospects. It is believed that ChatGPT has a wide range of future commercialization scenarios and can be integrated into other applications. At present, discussions related to ChatGPT application scenarios are mostly focused on search engines, life services, subject education, daily writing, etc., including: chat robots; writing and debugging computer programs; creations in literature and media-related fields, including creating music and TV series , fairy tales, poems and lyrics, etc.; education, exams, answering test questions.
Some experts believe that although the current ChatGPT cannot bring about fundamental changes in production methods, it represents that the development of artificial intelligence has entered a new stage.
It is worth noting that generative artificial intelligence represented by ChatGPT has attracted the attention of the US defense department. The US Defense News Network reported on January 28 that Stephen Wallace, director of the Defense Information Systems Agency (DISA) under the US Department of Defense, said that the technology has recently been included in the Defense Information Systems Agency (DISA) technology Watch list.
The U.S. Defense Information Systems Agency regularly updates the watch list, and some technologies eventually become the backbone of national defense security, such as 5G, zero-trust digital defense, edge computing and other technologies. The U.S. Department of Defense’s third “Offset Strategy” states that the rapid development of artificial intelligence will define the next generation of warfare. American technology website Systelusa analyzes that ChatGPT’s ability to understand, respond and interact with humans will help enhance military intelligence and decision-making.
ChatGPT has the potential to bring about a model revolution in open source intelligence analysis using artificial intelligence.
First, with the support of artificial intelligence ChatGPT, big data technology is integrated with the Internet, and the processing capabilities of large-capacity data will be significantly improved. The automation and intelligence level of open source intelligence work will also continue to improve. The upgrade and development of content generation technology will provide The iteration and development of intelligence work provide new directions.
Second, ChatGPT has the potential to help solve limiting factors such as the disorder, complexity, large quantity, and difficulty in screening of open source information, and greatly improve comprehensive response capabilities such as precise capture, comprehensive sorting, rapid warning, plan generation, and emergency guidance of open source information. , thereby gaining the upper hand and the initiative.
Third, ChatGPT can improve the autonomous operation capabilities of open source intelligence systems. That is, the algorithm is repeatedly trained based on the continuously upgraded massive amounts of historical data, and finally achieves relatively independent independent accurate collection and analysis, and realizes the independent intelligent operation of the open source intelligence system.
Fourth, ChatGPT can improve the accuracy and standardization of intelligence analysis. The training algorithm of artificial intelligence is generated by repeated training based on past major events, massive data sets and expert experience. Although it lacks critical thinking and innovation in simulating unconventional events, its ability to analyze events under traditional modes far exceeds that of human beings. Therefore, using artificial intelligence training algorithms to assist intelligence personnel training can eliminate the understanding bias caused by human factors to a certain extent, making the analysis process of intelligence analysts more scientific and standardized.
Of course, there are still many problems with the practical application of artificial intelligence in the intelligence community. First, the premise of artificial intelligence application is the large amount of training data required for deep learning. Although various intelligence methods can collect large amounts of data, they often do not meet the requirements of machine recognition and calculation; second, the security of artificial intelligence still remains There is no guarantee. The development of computer attack and defense technology is a process of ebb and flow, especially for artificial intelligence with high complexity and opacity. Even if it is attacked or deceived, it is usually difficult to detect; thirdly, the problems brought by artificial intelligence The ethical issues that arise are inevitable in the intelligence community. On the one hand, leakage of the massive amounts of data collected will cause great social panic and unrest, and how to choose between data needs and civil rights remains to be discussed; on the other hand, the application of technical tools will inevitably infringe on the interests of intelligence personnel. , and the positioning of artificial intelligence has not yet been clear, and the debate between “artificial intelligence” and “intelligence enhancement” will continue.
Overall, the current generative artificial intelligence technology has many shortcomings in terms of development and application costs, information accuracy, and application models. In the long term, with the in-depth application of this technology, attention should be paid to many potential risks in the fields of ideological security, network and data security, technological security, social security and political security.
Judging from the general trend, the competition for national intelligence information will increasingly be characterized by competition between algorithms and computing power. Algorithms and computing power will become new national strategic resources after data.