In the "The Wandering Earth" series of movies, the artificial intelligence MOSS attracted the attention of the audience. Not only is it self-aware and self-renewing, but it can also make decisions that the system deems right in the shortest amount of time, automatically generating relevant content to answer the questioner's question.
With the continuous breakthrough of artificial intelligence (AI) technology, MOSS is expected to appear in real life. And now, generative AI with similar capabilities to MOSS is making its way into the limelight.
Since the release of ChatGPT last year, a number of generative AI chatbots have continued to appear. These generative AI chatbots can not only provide high-quality real-time interaction and feedback on human cognitive needs, but also have strong topicality and entertaining conversational capabilities, and have quickly become the "top stream" in the technology circle. At the same time, generative AI, with its powerful network penetration capabilities, public opinion cognition capabilities, processing and analysis capabilities, may promote the transformation of modern warfare in a more diverse form, more insidious means, and more difficult to resist.
Demystifying generative AI
Today, AI is not a new term. Mobile phones have AI voice assistants, cars have AI navigation, and even canteens have AI chef ......AI seems to be everywhere in our daily lives. Actually, AI is a broad term. To put it simply, AI is the sum total of a series of technologies that include researching, simulating, and expanding human intelligent behavior.
ChatGPT, which will become popular all over the Internet in 2022, is an example of generative AI. Generative AI is a specific type of AI technology. Based on the Xi of known information, generative AI jumps out of the circle of imitating human behavior, trying to generate new data information for human reference from the perspective of human thinking.
The key word for generative AI is "innovation". It works by using computer programs to simulate human creative thinking after pre-training on large data, so as to create a certain logical and coherent language text, images, audio, and other content. In this creative process, large models are the soul of generative AI.
The so-called large model refers to a deep Xi or machine Xi model with tens or even trillions of parameters. Generative AI uses large models to model the complexity of databases with massive high-quality datasets, and uses powerful computing power to estimate model parameters and find relationships between data. Taking ChatGPT as an example, its model architecture is generated based on natural language processing and deep learning Xi technology in AI technology, with 175 billion parameters, and can learn Xi language rules and patterns in these data through pre-training on huge datasets. It is also optimized with a "human-in-the-loop" approach, providing highly realistic dialogue scenarios by interacting with users for their own feedback and improved output.
Currently, to chatgpt-4The generative AI represented by 0 has realized the automatic processing of multimodal content, which can generate specific text, audio and other information for different dialogue scenarios, and can also automatically generate conversations similar to human language according to the context to communicate closely with users.
It can be said that generative AI is a product that integrates big data, large models, and large computing power. It can also be said that it is precisely because of the current growth of social information data and the rapid development of network computing technology that generative AI has eaten the "dividends" of the times.
Generative AI has sparked thinking and exploration in almost all fields. Some experts say generative AI could replace hundreds of millions of full-time jobs. In the financial sector, generative AI is able to generate risk assessments and investment strategies through the analysis of large amounts of data, market trendsIn the field of art, generative AI can generate ...... of works with the artist's unique style according to the specific needs of users through the Xi of public works such as artists' cultural creations, songs, and paintingsNot only that, generative AI can also be integrated with emerging technologies such as the Internet of Things and cloud computing to form a more complete and comprehensive technology ecosystem. It is no exaggeration to say that generative AI has endowed AI technology with more powerful functions and broader development prospects, further expanding the impact of AI technology on human production and life.
It may become a subversive force on the battlefield in the future
In the early morning of January 3, 2020, the commander of Iran's Quds Force, Qleimani, was suddenly attacked outside Baghdad International Airport in Iraq. The extensive use of information and intelligent technologies such as 5G, the Internet of Things, and artificial intelligence has made the current form and means of warfare undergo subversive changes.
It has been said that in future wars, the role of data will go beyond ammunition. Artificial intelligence technology further empowers data, making data the source of intelligent warfare. For example, the US military's MQ-9 drone assassination of Soleimani was achieved through a large number of collection and analysis of his life Xi intelligence data, reconnaissance and monitoring of his behavior data, etc., to achieve the overall tactical goal of "predicting in advance, cruising and locking, capturing and tracking, and killing accurately".
In the era of information and intelligence, unmanned underwater vehicles, unmanned surface ships, unmanned aerial vehicles, and other unmanned early warning and detection forces have been further strengthened, satellite technology has continued to develop, and land-based, sea-based, and air-based reconnaissance forces have been upgraded. At the same time, the expansion of channels has also brought about the explosive growth of military data.
In the face of massive and rapidly changing battlefield data and information, it is particularly important to analyze and judge the battlefield trend more accurately and quickly. The main challenges facing the future military big data application are how to use relevant network technologies to research and mine massive data, clarify the logic between data, find out the secrets hidden in the data, and transform them into intelligence to assist commanders in decision-making. Generative AI, with its powerful information processing and output capabilities, is the key to solving this problem.
Generative AI can perform unsupervised pre-training on many different types of data, obtain valuable data from it, greatly shorten the processing time, and find and discover the patterns in it, generate combat plans according to the battlefield situation, and provide advice for operational command. Larger-scale neural network algorithms make generative AI closer to "general talents", which can not only process and create texts and images, but also study and judge the battlefield situation, the trend of war, and even implement cognitive confrontation, playing a special role in psychological warfare, warfare, and intelligence warfare.
At present, generative AI has attracted the attention of the world's military powers. In January this year, the United States added generative AI technology to its "technology watch list" and announced the creation of the Department of Defense Generative AI Working Group in August this yearAt the 2023 International Conference on Artificial Intelligence Journey held on November 24, Russia** Putin announced that he will soon sign and approve a new version of Russia's AI development strategy, which involves expanding basic and applied research on generative AI and large language modelsIn March, the UK** pledged £1 billion into supercomputing and artificial intelligence research in hopes of building its own generative AI chatbot.
The role of "virtual staff officers" is becoming increasingly prominent
From the perspective of theoretical conception and practical performance, the most prominent and main manifestation of the militarization of generative AI is the virtualization of the "staff officer" seat. As a rapidly evolving trend, "virtual staff officers" may play a pivotal role in future wars.
First, generative AI can take over traditional staff operations and ease the pressure on staff officers. As the nerve center of operational command, staff officers are responsible for the circulation and transmission of all kinds of information, and must accurately convey the intentions and operational orders of their superiors to the corresponding combat units. After pre-training, generative AI can rigorously and rationally formulate new military documents according to keywords or specific requirements, and integrate information into various forms such as situation images and data charts for display, freeing staff officers from traditional services such as plotting, calculation, and writing, and improving the efficiency of operational command.
Second, generative AI can guide commanders in military decision-making in the direction of warfare. Based on strong data support, algorithms, and the development law of thing generation, the credibility of generative AI's output results in military affairs will be further improved. It can analyze and process real-time information on the positions, actions, and capabilities of both the enemy and us, as well as the advantages and disadvantages of the situation, and based on the development trend of the war, it can form operational proposals and assist commanders in making decisions and commands. Advanced generative AI models can aggregate enemy offensive information, understand enemy action patterns, combat Xi, and even simulate the "way of thinking" of enemy commanders through algorithms.
In addition, generative AI can assist human-computer interaction and integrate control platforms. Generative AI is an ideal human-computer interface because it is a machine product that understands human language at the same time. The Russian military attaches great importance to the application of relevant models in anti-UAV, and Kaspersky Lab has developed the "Kaspersky Anti-UAV System" based on artificial neural networks by combining data models with traditional strike methods. By building an artificial neural network, the data received by various sensors around important facilities is analyzed and processed, and at the same time, special algorithms are used to quickly discover and identify drones, independently classify them, judge friends and foes, and respond in a targeted manner. This year, the U.S. military also used generative AI to generate reports on anti-drone operations. In general, generative AI can break down combat missions into sub-missions and use built-in algorithms to get the most optimized solution. Relying on its rapid response capability, it can accelerate the efficiency of the "observation, adjustment, decision-making, and action" cycle in operations, and help commanders cope with the tense rhythm of future wars.
Large-scale model technology, represented by generative AI, has the potential to be widely used in the military field and has a profound impact on emerging combat means and methods. Of course, problems such as "lack of samples", "no data", "non-openness" and "difficult sharing" still plague the field of military intelligence. At the same time, AI technology involves biometrics, automatic planning, intelligent control and other aspects, and it is possible to lose control in complex environments or when executing complex instructions
In fact, the pros and cons of the application of AI technology in the military field depend on the choice of human beings. What kind of path generative AI should take to the battlefield remains to be further explored.
*: Xinhuanet client
Tianjin several ports
Data-driven intelligence, innovation leads the future.
Editor: Pang Xiaoti Shi Yue.
Review: Deng Jinglong.