Smart Brain, Downgrading Operations, and GANs in Future Wars
Issue 2, 10 October 2021
Smart Brain, Downgrading Operations, and GANs in Future Warfare
This week, the PLA Daily carried opinion pieces on asymmetric warfare, intelligent warfare, "language intelligence" and its importance in future warfare, foundational models in AI, and Network Information Security.
(Image Source: pngtree.com)
First, an article in PLA Daily written by scholars from the National University of Defence Technology talks about the importance of language in the future "informatized" and "intelligent" warfare. The article argues that in the future scope of language will exceed communication and will be integrated with various branches like integrated with computer science, cognitive science, psychology, etc. The authors postulate the possibility of language as a "soft kill" (软杀伤) weapon and importance of "language intelligence" (语言智能) in a range of fields, including life sciences and military. The author argues that countries like the United States (US), France, Russia, and Japan emphasize the use of Artificial Intelligence (AI) to process linguistic information and assist battlefield decision-making. Hence, it is important to develop an application of AI to create a "smart brain" (智慧大脑) for national defense. For the "smart brain" (智慧大脑) to be effective, it is necessary to make efforts in areas where "language intelligence" (语言智能) has advantages. The author argues that priority should be given to the development of intelligent and integrated equipment to deal with the enemy’s "key language" (关键语言) along with technologies like big data, speech recognition, voiceprint comparison, semantic analysis, etc. Furthermore, the article highlights the need to further innovative research and expand the military applications of natural language processing in fields like machine translation, public opinion monitoring, and speech recognition and develop intelligent mobile terminals and intelligent battlefield robots suitable for different scales of operations. Lastly, the authors call language a catalyst of military soft power.
Developing a "smart brain" (智慧大脑) and applying it for national defence doesn't seem like a science fiction dream for Chinese scientists. May be not immediately, but it is not entirely fiction to attempt such AI. Given that AI ecosystem in China is advancing in full speed, and encouragement from government, this seems plausible.
Second, another article on deep learning in AI highlights the importance of Generative Adversarial Networks (GANs). The author acknowledges the importance of AI and deep learning in military applications and future warfare.
GAN is a class of machine learning framework which uses adversarial learning method. It has two sub-models called Generator and Discriminator, which are engaged in a zero sum game. The Generator generates an output based on input and Discriminator uses the output generated by Generator and sorts them into binaries of real or fake. Described first by Ian Goodfellow along with others in 2014, it has rapidly advanced to become one of the most important breakthroughs in computer science. One of the applications of such model is to generate realistic images of people, objects, and scenes which cannot be distinguished as real or fake even by humans.
Moving on to asymmetric warfare, an article by scholars from PLA Army Research Institute (陆军研究院) illustrates characteristics and a new interpretation of "downgrading operations" (降级作战) in asymmetric warfare. The "downgrading operations" usually refer to the methods that degrade the enemy's environment and decision-making capabilities. As the article points out, the "downgrading operations" are designed to make the opponent "less fast, less tall and less strong" rather than making ourselves stronger and sharper. Downgrading the environment could mean destroying weapons and equipment or degrading conditions of personnel operating them. It also involves reducing the timeliness and accuracy of the opponent's decisions by increasing difficulties in obtaining information or polluting information needed for decision making. The aim is to make chaos. Although not explained sufficiently, the authors mention the use of AI in such "downgrading operations." Moreover, words like "mosaic warfare", "decision center warfare", and "distributed warfare" make an appearance as a prediction of dominant modes of warfare in the future.
Although the article doesn't mention of China's capabilities or future plans to carry out similar "downgrading operations", it is clear that Chinese scholars view such capabilities as crucial in future warfare. This is also evident given that authors claim the US is developing multi-domain tasks force and China wouldn't want to appear on a technological backfoot.
Next, an article on Network Information Security argues that Network Information Security is no longer a purely technical and information management issue, but a strategic issue. It involves national information security, economic and social development and stability, and even the success or failure of military struggles. The article claims that developed countries have strengthened the security of their networked systems by developing electronic weapons or anti-attack, anti-monitoring, anti-stealing, and anti-destroy technologies. The author makes it clear that China needs to develop such capabilities of its own as soon as possible.
The article gives off the impression that China doesn't possess the capability to fully secure its networked systems and anti-attack, anti-monitoring, anti-stealing, and anti-destroy technologies. However, given the emphasis of PLA and Communist Party of China (CPC) on informatized warfare and as evident from creation of a separate PLA Strategic Support Force (PLASSF), I wouldn't be surprised if Chinese scientists achieve this capability soon.
The state media is actively promoting the various use cases of AI and related technologies to showcase achievements and facilitate discussion on AI. For example, an article in ST daily this week claims that China's AI Judge "Xiaozhi" (小智) can complete 10 joint trials in 30 min. As per another report, China's self-driving buses have started trials in France. ST Daily also published an article discussing large AI models like OpenAI's GPT-3 and China's WuDao 2.0 (悟道 2.0). The article quotes China's leading AI researchers from Beijing Zhiyuan Artificial Intelligence Research Institute / Beijing Academy of Artificial Intelligence (BAAI) (北京智源人工智能研究院院长). BAAI is one of the leading AI research institutes in China.
BAAI was part of the team which developed China's most advanced Natural Language Processing (NLP) AI system WuDao 2.0 (悟道 2.0), which is claimed to be faster than Google's Switch Transformer AI. In June 2021, BAAI along with other industry partners launched China's first virtual student based on WuDao 2.0 (悟道 2.0). BAAI is also developing applications for the upcoming Beijing Winter Olympics.
Lastly, Chinese military watchers seem to be in awe of Russian military equipment and machines. PLA Daily has been carrying at least one detailed article on Russia's military weapons, aircraft, armaments, etc. This week features a long article on successful testing of Air-cushioned landing craft (also called Landing Craft Air Cushion (LCAC) ) type hovercraft Mordoviya (Zubr-class LCAC). It then goes on to explain how each country, including the US, attaches importance to LCAC hovercrafts and the US is even moving towards the advanced Shore-to-Ship Connector (SSC) to replace LCAC. The author mentions that the United Kingdom, Russia, the US, and South Korea have made good progress in making air-cushioned aircraft. In a rare instance, the article mentions India in the list of countries having the capacity to develop such hovercraft, along with Canada and Australia. Among the list of possible improvements for this type of hovercraft, the author mentions developing "unmanned" LCACs in the future.
Air-cushioned aircraft or LCAC are multi-purpose amphibious hovercrafts. They can be used to carry tanks, infantry fighting vehicles, missile launch vehicles, and troops. Such hovercrafts are considered good for surprise attacks from the sea. Hence, will be important for China's operations in Taiwan and in the South China Sea. PLA has Air-cushioned landing crafts called Type-726 LCAC or "Yema" LCAC which is independently developed and produced in China.
If unmanned LCACs are developed, then it will significantly improve China’s surprise attack capability in Taiwan strait. Improving speed and range of operations could also have along term applications in South China Sea and other naval oerations of PLA.
In Short
World Robot Conference 2021
The 2021 World Robot Conference was held in Beijing in 2021. The conference displayed a large number of robotic innovations. The robotic applications ranged from space systems to smart healthcare, One such interesting display was of a low-noise intelligent bionic underwater robot "Shark" that can operate in water at a depth up to 600 meters. It has a sound-absorbing material shell and bionic caudal fin propulsion system, with strong stealth ability. It is equipped with an advanced reconnaissance system. It can be used for reconnaissance, target search and rescue, and hydrological monitoring tasks while lurking underwater. The article cites it as "underwater detective."
(Photo: Simon Song/SCMP)
Zuchong Award (祖冲之奖)
Huawei and Pengcheng Laboratory jointly announced the "Zuchong Award" (祖冲之奖) for the Artificial Intelligence (AI) innovation. It is aimed to reward key technological breakthroughs in the field of artificial intelligence. The award is named after the ancient Chinese mathematician and astronomer Zu Chongzhi. The first award will be announced in 2022. Swiss company AI Witchlabs AG will be part of the process to recommend projects and assess awards.
Electronic Warfare Aircraft Near Taiwan
South China Morning Post (SCMP) reports that China is on expanding its airbase near Taiwan and has deployed J-16D electronic warfare aircraft to Xiangtang airbase in Nanchang, Jiangxi province near Taiwan. The report is based on the satellite image released by Canada-based Kanwa Defence Review and an anonymous military source in Beijing cited by SCMP.
Additional Readings
On the Opportunities and Risks of Foundation Models, Center for Research on Foundation Models (CRFM), Stanford Institute for Human-Centered Artificial Intelligence (HAI) Stanford University
10 trends for the future of warfare, World Economic Forum.