A Chinese achievement in the field of Artificial Intelligence (AI) has widely been reported in the media recently. Be it remotely controlled Tank to demonstration of Swarms of Autonomous UAVs during Air Show, China is building them fast and possibly in numbers while preparing for ‘Intelligized Warfare’. These technology demonstrations are not very far from field deployment and are a cause for worry for several countries. Chinese State Council also released a detailed plan for development of AI for development of nation, including military application. It is an ambitious plan but China has a strong foundation in its Academia, Public Sector Industry and Startups to make it possible and enable China to become a Global Leader in AI. It is therefore important to examine this foundation of Industry, Startup Ecosystem, Academia and their mutual cooperation to truly understand the potential of China and be able to predict the military deployment of AI.
In Jul 2015 an open letter was published during 24th International Joint Conference on Artificial Intelligence (ICJAI-2015), a leading conference on Artificial Intelligence (AI) expressing fears of a global military AI arms race. The signatories included eminent scientist late Dr Stephen Hawkings, Apple cofounder Steve Wozniak, world leading AI researchers Stuart Russell, Nils Nilsson and over 22,000 others [1]. This fear was based on the fact that AI can bring autonomy to weapon platforms and revolutionise other aspects of warfare. The sheer impact of AI on battlefield is now being considered as the third revolution in warfare after gunpowder and nuclear weapons. In spite of all the philanthropic intents of leading scientists of use of AI, all major military powers are working hard and fast to build the cutting edge systems for AI and autonomous weapons. Unlike nuclear weapon, AI does not require hard-to-obtain raw materials and can become significantly cheap for rogue states to proliferate and non-state actors to obtain.
Autonomy in weapon platforms will give tremendous advantage to any defence force. Autonomous drones can find a target and destroy it even before any human intervention is possible. Such autonomy can be built in aerial, terrestrial, amphibious, floating or sub surface platforms and has the potential of shortening the OODA (Observe Orient Decide and Act) loop by several factors, effectively taking the human from ‘in-the-loop’ to ‘on-the-loop’. In addition to autonomy, AI will affect the entire spectrum of war fighting, from operations planning, logistics planning to human resource management and training. Improvement in image recognition can exponentially enhance ability to observe and gather intelligence. AI can help provide better planning tools which help carry out operational or logistics plan with more efficiency.
Dual Use of AI
It must be clearly understood that AI is a dual use technology. Image recognition using Deep Neural Nets can be used to guide a blind person to navigate a city and can also be used to guide a missile! Tech used in self driving cars can be extrapolated to self driving tanks! Natural Language Processing and Voice Recognition has already enabled real time voice translation, it can be a very effective tool for surveillance and intelligence gathering. Any advancement made by China for apparently non military application must be viewed in light of its possible military use.
Ambitious Political Leadership
China with its growing ambition of global and regional dominance has identified AI as a strategic force multiplier and has been frantically working towards developing the technology. Chinese Premier Li Keqiang mentioned AI being an opportunity to build strategic capability during his speech at National People’s Congress on 5th March 2017 [2]. Chinese President Xi Jinping has also said that the robot revolution is expected to herald the Third Industrial Revolution [3]. This clearly shows the importance given by China to AI at the highest level. This enhanced strategic importance of AI should be seen in the light of the fact that China announced reduction of as much as 300,000 troops in 2015 while increasing their defence budget by 7% in 2017 in spite of an economic slowdown. Both these factors indicate towards China’s intention to invest in modernisation of People’s Liberation Army (PLA) and AI will be integral part of this modernisation effort.
National AI Development Plan
Chinese State Council announced a detailed plan to develop AI in China in July 2017 [4] and which included development of AI for Military Applications. The plan acknowledges that AI is a Strategic Technology and is the new focus of international rivalry. Among other Industrial and Social applications of AI, China has clearly articulated a need to Safeguard of National security in their plan. China plans to match the cutting edge AI technology with the world by year 2020 and by 2030 make some major breakthrough. AI being a dual use technology, it is clear that any breakthrough in the name of Social or Industrial advancement will soon be imbibed in PLA as defence technology. The Plan goes on to touch upon ‘over the horizon’ technologies like Quantum Computing which will bring quantum jump in AI capabilities once in production stage.
Military-Civil Fusion in National AI Development Plan
The plan explicitly brings out key steps to be taken to establish communication and cooperation mechanism among Military, Industry, Research Establishments and Academia. The cooperation will promote two way sharing of AI technology. This step is necessary to enable seamless dual use of AI.
Policy Initiatives for Strengthening Indigenous Research & Development
China aspires to develop AI indigenously and strengthen their technological capabilities. China believes that indigenous research rather than procurement from foreign sources is the most effective way of strengthening the nation as well as defence forces. China worked towards it meticulously by creating programmes to encourage research and strengthen universities as the primary drivers of national development. Strong academia can further contribute towards strengthening industry. Notable programmes include 863 Programme – National High Tech R&D Programme, 973 Programme – National Basic Research Program of China, Projects 985 and 211 – to promote and develop the Chinese higher education system.
863 Programme was initiated in 1986 with the primary objective to ‘boost innovation capacity in the high-tech sectors, particularly in strategic high-tech fields, in order to gain a foothold in the world arena; to strive to achieve breakthroughs in key technical fields that concern the national economic lifeline and national security’. [6] Information Technology was identified as the first major task of the program. 863 Programme is considered to be one of the major contributors to the development of robotics and AI related technologies in China with both civil and defence applications.
China launched ‘Internet+’, a three year AI implementation programme, effective from2016 to 2018 [5]. In addition to the programme’s application in industrial and automotive sector, security is a sector the programme seems to particularly focus
on. The programme is to provide data, mass training and infrastructure for deep learning applications. This programme is considered to be a precursor for the detailed plan for development of AI announced by the Chinese State Council [4].
Chinese Higher Education, Research and Development Backbone
Funds from these programmes are mostly spent on research and development being conducted at various universities and educational institutes of China. There are more than 500 universities in China, among them; the elite are an alliance of nine universities called C9 universities, the Ivy League of China. There are 39 ‘Project 985’ universities which aspire to be of international standards. The 71 ‘Project 211’ universities have enhanced research standards[7].
Four of the C9 League universities are among the top 100 universities of the world as per QS World Ranking 2016 with Tsinghua University topping at rank 24. It may be interesting to note that QS World Ranking of IISc Bangalore is 152 which is the highest for any Indian institution. Other universities to figure in top 100 list of QS World Ranking 2016 are Peking University, Fudan University and Shanghai Jiao Tong University. [8] AI is part of curricula in all of them with most having a dedicated centre or institute for AI research. C9 League universities collectively attract about 10% of the China’s R&D budget and produces about 20% of the scientific papers[9].
A strong foundation of Academic Institution not only provides trained manpower required for the industry, it also provides possibilities of R&D for defence projects in collaboration with industry. It may be pertinent to note that most Ivy League universities of USA have considerable amount of defence research to their credit. Strengthening the education and R&D at the universities is possibly the best single step that can be taken to strengthen a nation.
A good indicator of China’s pace of AI research is SCImago Journal and Country ranking which draws its data from over 21,500 titles from more than 5,000 international publishers [17]. The figure below shows the number of citable papers published by China, USA and India in category of Artificial Intelligence of subject Computer Science from year 1996 to 2015 in groups of five years. It is interesting to note that China overtook USA after 2005 and is doing more research than USA ever since. China published more than four times the papers published by India between the years 2011-15. In year 2016 China published almost double the number of papers published by USA [45].
AI in Military Higher Technical Education in China
In addition to the leading universities doing a considerable amount of research in the domain of defence and AI, there are universities and institutions dedicated for defence education and research. Notable among them are National University of Defence Technology, Naval University of Engineering, Naval Aeronautical and Astronautical University, Armoured Engineering Academy, China Aeronautical Institute, Nanjing Artillery Academy, PLA Academy of Military Science, Air Force Engineering University and Military Transport University. It may be worthwhile to note that Military Transport University and National University of Defence Technology participated in an autonomous vehicle challenge held in China recently conducted on the lines of DARPA Grand Challenge.
An estimate of research and development at military technical training establishments can be made by carrying out search of scientific papers at IEEE Xplore Digital Library, which shows over 5,400 papers with author affiliation to National University of Defence Technology, China. This is even more than the best technical establishment of India – IISc with 5,000 papers. Over 1,000 papers have author affiliation to Naval University of Engineering. Their research covers areas like Battlefield Target Evaluation [10], AI for Cognitive Radio [11], Autonomous Vehicles [12], Case Based Reasoning for Automatic Weapons [13], Autonomous UCAV [14], Autonomous Bionic Underwater Robot [15], Equipment Fault Representation and Prediction [16], etc.
The China Brain Project
The most ambitious effort by China in the field of AI is a combined neuroscience effort to understand human brain. China Brain Project [18] kicked off in 2016 as a 15 year plan to study the basic neuroscience, diagnosis of brain diseases and brain inspired computing for AI. The effort is being led by Institute of Neuroscience, Chinese Academy of Sciences, China, with collaboration of other educational institute and private companies like Baidu. It is also expected to have support of the Chinese military. The project aims to better understand working of brain at multiple levels and derive cognitive models and architecture for brain inspired chips. The project will also help build intelligent and interactive robots who can work well even in uncertain environment which can subsequently lead to development of ‘General AI’, the holy grail of AI research. ‘General AI’ is a system which is at least equal to human intelligence in every aspect.
Chinese Defence Public Sector
Academia has lead the research in China, but it is the industry which can finally translates these research into production grade, reliable and rugged weapon systems. Biggest companies in China are the public sector giants and quite a few of them are dedicated to production of defence systems. Notable among them are Aviation Industries Corporation of China (AVIC) with revenue of about $60 billion, China North Industries Group Corporation (NORINCO Group) with revenue of about $58 billion, China Electronics Technology Group (CETC) with revenue of about $26 billion. All three of them are fortune 500 companies. To put things in perspective, Bharat Electronics has revenue of about $1 billion and Hindustan Aeronautics Limited has revenue of about $2.6 billion.
NORNICO Group has demonstrated unmanned armed ground vehicles Sharp Claw 1 & 2 [19]. Sharp Claw 2 has the capability to navigate autonomously. It is a smaller vehicle and piggy backs Sharp Claw 1. They also demonstrated an autonomous ground logistics vehicle called Crew Task Support Vehicle during Zhuhai Air Show 2014 [20]. China Aerospace Science and Industry Corporation is apparently working on an intelligent cruise missile endowed with AI [21]. The missile will apparently be capable of selecting its target in flight and then destroy it. CETC research labs are working on Military Internet of Things[22] and in collaboration with Tsinghua University demonstrated swarm intelligence by flying more than 20 autonomous drones in a swarm [23] during Zhuhai Air Show 2016.
Chinese Private Sector and Startups
While the Chinese public sector takes step in the right direction in their own pace, it is the private sector and the technology start-ups which gives them the technical edge and boost of energy especially in the field of AI where initial investment required to carry out research and development is much less as compared to other fields. Baidu, Alibaba and Telcent are now globally known companies that grew on the internet boom and on Chinese policy to keep companies like Google out of their borders. Chinese startups have also gained much grounds in the field of AI like Turing Robots, DFRobots, iFLYTEK, Tianze Information etc. They all have created products and services using AI and Machine Learning. Startups in China are being supported by a very healthy startups ecosystem with the venture capitalists focusing a lot on AI.
Robin Li’s Baidu
Baidu with a market cap of over $60 billion is popularly known as Google of China.
Baidu is actively promoting AI in China. Services offered by Baidu in China include
search, maps, shopping, food ordering, etc. Baidu sees AI and allied technologies as essential for future sustenance and has invested heavily in research and development. It has established three research labs – Big Data Lab Beijing, Institute of Deep Learning Beijing and Silicon Valley AI Lab in California and has employed an estimated 1300 researches[24]. In addition to in house research,Baidu collaborates with a number of other research establishments and universities around the world.
Baidu’s research work in AI covers deep neural networks, speech understanding, image recognition, healthcare etc. In addition to the team of scientists that Baidu has, its research is facilitated by the enormous amount of data it has access to because of the large number of services being offered by Baidu[25] coupled with absence of Google and Facebook in China.
Kai-Fu Lee of Sinovation Ventures
Another stalwart in China’s AI Startup community, Kai-Fu Lee is a Venture Capitalist focused on AI startups. He is known to recruit young Chinese students from MIT for his AI Institute in China [47]. His company Sinovation Ventures has about $1.3 Bn worth of Asset Under Management.
A healthy Startup Ecosystem betting heavily on AI shows the confidence of the Market in the technology. A number of startups as mentioned above have been able to develop cutting edge solutions with AI. With their possible dual use and authoritarian nature of Chinese Government it is highly likely to see these technologies adapted for military use.
Civil – Military Fusion
Cooperation between Chinese government and the citizens exists through the Chinese People’s Political Consultative Conference (CPPCC), a political advisory group. The group include Robin Li, the founder of Baidu who advocates for increased involvement of government in development of AI. His call is also echoed by heads of Xiaomi and Geely Automobiles [26]. Robin Li is one of the early influencers of China Brain project. He invited the Chinese military and National Development and Reform Commission (NDRC) to participate in the China Brain project[27]. All the technologies being developed by Baidu can be used for military applications. With increased initiative from government’s and military’s side for partnership with civil industries, it is more likely than ever for technologies developed by Baidu being used in military applications. The civil – military
partnership initiative has been taken up at the topmost level with Chinese President Xi J nping heading the recently launched central commission for integrated military and civilian development [28].
Such civil – military partnership will not only bring relevant technologies to the doorstep of PLA, it also ensures a healthy collaboration among the leaders of industry and PLA. Industry leaders like Robin Li, CEO, Baidu has also called upon collaboration with military. Conglomeration of such stalwarts supported by the top military and political leadership enabled by empowered forums is one sure way to transform a nation.
Informed and Involved Military Leadership
Leadership within the PLA can be considered ‘Scholar Warriors’. Maj Gen An Weiping[29], Deputy Chief of Staff, Northern Theatre is a graduate of National Defence University of Science and Technology and has been referred to as a PLA scholar as early as 2004[30] even in a few western publications. He has commented on necessity of cyber security and ‘intelligization’ of battle field. Maj Gen Li Deyi [31] is an academician and director of the Chinese Association of Artificial Intelligence. Maj Gen Hu Xiaofeng[32] is a professor at National Defence University and is regarded as an expert in simulation and war-gaming. Industry leaders like Robin Li, CEO, Baidu has also called upon collaboration with military.
Conglomeration of such scholars supported by the top leadership and enabled with
empowered forums is one sure way to transform a nation.
Upcoming PLA’s Intelligent Military Revolution
PLA is looking forward to an Intelligent Military Revolution by using AI in a networked battle space[33]. They plan to put more machine soldiers in battlefields of 21st century than the human soldiers[34]. PLA is planning to move towards an integrated ‘System of Systems’ approach to enable joint operations in a highly autonomous environment[35]. Establishment of ‘Intelligent Unmanned Systems and Systems of Systems Science and Technology Domain Expert Group’ by PLA [36] indicates an organisational resolve towards transformation to a digitised and ‘intelligentized’ fighting force. There have been quite a few news crumbs of successful demonstrations of China’s ability to develop and integrate new systems have emerged from various quarters as discussed earlier, like the swarm of UAV and autonomous land vehicles, however, there may be a large number of projects which are yet opaque to the rest of the world. Given that China has created a very large base of R&D in universities, R&D Centres and industry technologies developed in the name of civilian use can very easily be used for military application. Demonstrated technologies are therefore only a tip of the iceberg.
China Vs Rest of the World in AI
China may have lagged behind the developed countries like USA in the theoretical aspect of AI, as the research in China did not bring any radically different AI technique, but they seem to have mastered the art of implementation and in some instances left the USA behind [37]. A custom made computer for neural network ‘DianNao’ made by State Key Lab of Computer Architecture won the best paper prize in a renowned conference ASPLOS in 2014 [37]. China has maintained lead in the top supercomputer list with top two slots held by Chinese supercomputer Sunway TaihuLight and Tianhe-2 [37, 38]. Top Indian supercomputing facility located at IISc is at 133rd place. India has 28 supercomputers installed [39] while both China and USA has 171 each [38].
Chinese tech startups working in AI domain have also demonstrated capabilities which are second to none in the world showcasing their strength in application of technology in real world. Speech recognition capabilities of a tech startup iFLYTEK is considered one of the best in the world. Telcent’s deep learning platform Mariana and Huawei’s self learning AI assistance called MoKA are also considered state of the art. Tencent’s artificial intelligence (AI) Fine Art stole the limelight after its stunning 11-game winning streak, at the 10th Computer Go UEC Cup, which ended on March 19, 2017 in Tokyo[40].
Another significant achievement for China is the Imagenet Large Scale Visual Recognition Challenge (ILSRVC). In 2015 there were 25 teams from China, out of total 67, and they won in four out of five main categories. In 2016, there were 33 teams, out of total 82, from China and they won in every category[41]. In 2017 the Chinese teams were winner in only four of the eight subcategories but other winner team have most individuals apparently of Chinese origin [46].
USA’s Concerns
USA now considers China a real threat to their dominance in the field of AI and has now started closely following the AI related development in China because of their strategic importance. In Oct 2016 a Research Report was submitted on behalf of the U.S.-China Economic and Security Review Commission regarding China’s Industrial and Military Robotics Development. The report extensively discussed the robotics and the AI related developments in China and their implications with respect to USA [42]. In Feb 2017 a testimony was given before the U.S.-China Economic and Security Review Commission regarding Chinese advances in unmanned systems and the military applications of AI [43] by Elsa B. Kania. The testimony recommended extensive monitoring and analysis of the research and development activities in China, and continued funding of critical technology domain with capabilities associated with the Third Offset within USA.
Conclusion
Looking at some of the success stories emerging out of China; their work is yet to reach deployable state in most of the cases. Most of the autonomous platforms demonstrated have either limited application or are not yet mature for deployment.
However, one must also take into account the existence of political will and a very large research and development base with the Chinese academia. With the pace of AI research already more than that in the USA, coupled with the Chinese strength in local innovation, it is possible to see induction of autonomous platforms in the PLA within next five years. A discussion paper by McKinsey published in Apr 2017 points out that China has a huge potential to be a global leader in AI in civil domain [44]. This potential will also be harnessed by those working towards military applications of AI.
China has come a long way and this is a result of years of effort of driven by top Chinese leadership, which helped formulate the right policies, which developed Chinese universities and institutes into top class research and development centres, which produces technology experts, evangelists and drivers, which resulted in fruitful collaboration through right forums, which resulted in formulation of right research projects which finally results in innovation, which in turn provides better opportunities to policy makers and enable them to correctly visualise the technology imperatives. Such a cycle is self sustaining and is capable of providing technological solution at a faster pace for a prolonged period of time.
The comparison of combat potential of any two adversaries is usually done by comparing the quantity and lethality of weapons platforms held with their defence forces. But it may be noted that defence forces ‘project’ the strength and will of a nation, the strength of any nation lies in the capability of the industry and academia to either provide revenue for purchase of weapon system or produce indigenously. It may therefore be prudent to compare the research potential and the capability to convert research into deployable systems. In today’s world with exceedingly fast pace of technological advancements, strategic advantage is not just existence of a particular technology in one’s armoury, but also the ability to create and deploy new ones at a pace faster than the adversary.
