Published SP’s Military Year Book 2019- Section- Technology-Pg 121 http://www.spsmilitaryyearbook.com/
Contrary to the more fantastic predictions for AI in the popular press, the Study Panel found no cause for concern that AI is an imminent threat to humankind. No machines with self-sustaining long-term goals and intent have been developed, nor are they likely to be developed in the near future. Instead, increasingly useful applications of AI, with potentially profound positive impacts on our society and economy are likely to emerge between now and 2030……
One Hundred Year Study on Artificial Intelligence (AI100) – Report of the 2015 Study Panel, Stanford University
Artificial intelligence AI algorithms have permeated the lives of the common man to a significant degree. From reducing typing effort by suggesting words, smart keyboards (voice activated typing), voice-activated assistants, background algorithms pushing out targeted advertisements or buying ideas & choices to smart devices & banking AI is slowly impacting the lives of citizens in many positive ways. While Machine Learning, ML is fundamental to AI applications of specific nature, some other disciplines of AI include: Natural Language Processing (NLP), Human-Robot Interactions & Robotics, Computer Vision, Multi-agent Systems, Search and Planning, Knowledge Representation and Reasoning (KRR), and Social Media Analysis.
Belfer Center of USA carried out a study ‘Artificial Intelligence and National Security’ for the Intelligence Advanced Research Projects Activity (IARPA) and has brought out four critical drivers behind the rapid progress in AI technology namely: decades of exponential growth in computing performance, increased availability of large datasets upon which to train machine learning systems, advances in the implementation of machine learning techniques, and significant & rapidly increasing commercial investment. Some characteristics of AI are unique for example: many of its application can be used both for the civilian and military applications with ease (e.g. image recognition algorithms); AI works in the background implying thereby that the algorithms are embedded in the software therefore it may not be obvious from the product per se that AI has been used in it; and the fact that AI has applications in almost all the fields of science and technology. Amidst the arena of research space occupied by AI, lies Artificial General Intelligence (AGI) which aims at building machines which can perform any function and task that the human being can do.
Militaries are looking at AI for enabling new force-multipliers as well as for achieving economies of scale and bringing down the acquisition costs of existing AI capabilities. Many weapon systems are operational today which have been enabled by AI, as an illustration: Sea Hunter, a trimaran of the US Navy, (a Medium Displacement Unmanned Surface Vehicle, MDSUV) became the first ship without a crew to sail from San Diego, California to Pearl Harbour, Hawaii and back. The US Navy has been developing it as an Anti-submarine warfare Continuous Trail Unmanned Vessel (ACTUV). In future Sea Hunter type vessels would provide useful options for monitoring submarine activity across the oceans and for Electronic Warfare Support measures. The South Korean military robot sentry SGR-A1, designed by Samsung, can challenge humans for a spoken password and, if it does not recognise the correct password, it can shoot with rubber bullets or lethal ammunition. It has been deployed in the demilitarised zone between North and South Korea. Also, the X-47B; the experimental strike aircraft developed by Northrop Grumman under the US Navy’s Unmanned Combat Air System (UCAS) Carrier Demonstration program; has successfully demonstrated aircraft carrier launch & recovery and autonomous aerial refuelling. The US Navy has decided to change the role of X-47B from a surveillance and strike aircraft into reconnaissance and aerial refuelling drone with ‘limited strike capability’.
An essential aspect of the AI programs currently being operated by the defence forces is that none of them can be said to perform like a human being. None of them can decide their own goals or decide what they would do independent of human intervention, in short none of the weapons can be classified as genuinely autonomous and possessing human cognitive capabilities (or having an AGI lineage!).
At this juncture, it would be appropriate to look at AI developments in China, Pakistan, and within India itself.
“Our adversary on the northern border (China) is spending huge amounts of money on Artificial Intelligence and cyber warfare. We cannot be left behind,” said Bipin Rawat. “It is time for us to also focus on AI and on Big Data Analytics rather than just keeping it confined to mere definitions.”
General Bipin Rawat, Chief of the Army Staff, 2019
The People’s Republic of China. In the defence sector, China has been focussing upon AI applications in unmanned systems, cybersecurity, facial recognition & surveillance, and cruise missiles. To make up for shortages in AI specialised workforce several Chinese institutions have commenced programs, for instance, Tsinghua University is running an advanced laboratory specialising in military intelligence and The PLA National University of Defense Technology has started an institute for intelligent sciences. China had released its AI Development plan on July 20, 2017, and it is understood that China is going to invest ~$150 billion in AI. As brought out by Webster et al. in their article “China’s Plan to ‘Lead’ in AI: Purpose, Prospects, and Problems” in New America, 01 Aug 2017, The Chinese strategic objectives and timelines are:
2020
• Overall technology and application of AI will be in step with globally advanced levels.
• Cultivate the world’s leading AI backbone enterprises.
• “The scale of AI’s core industry will exceed 150 billion RMB (~$21.7 billion), and exceeding 1 trillion RMB (~$150 billion) as driven by the scale of related industries.”
2025
• Breakthroughs in basic theories for AI, such that some technologies and applications achieve a world-leading level and AI becomes the main driving force for China’s industrial upgrading and economic transformation.
• “The scale of AI’s core industry will be more than 400 billion RMB (~$58 billion), and the scale of related industries will exceed 5 trillion RMB (~$726 billion).”
2030
• China will be the world’s primary AI innovation centre, achieving visible results in intelligent economy and intelligent society applications and laying an important foundation for becoming a leading innovation-style nation and an economic power.
• “The AI, core industry scale, will exceed 1 trillion RMB (~$150 billion), with the scale of related industries exceeding 10 trillion RMB (~$1.5 trillion).”
The PLA hopes to capitalise on the civil-military industrial complex to achieve superiority in AI military applications and switch from “informatized” warfare to “intelligentized” warfare. An illustration of the serious intent of the PLA regarding its foray into strategic AI can be seen from an article in South China Morning Post which claims that China is upgrading its computer systems on its nuclear submarines with artificial intelligence embedded systems to assist the Commanding Officers in decision making.
Pakistan. Pakistan is going to invest $3.3 million over the next three years in AI research and capabilities. The AI project would be carried out at six universities selected by Pakistani Higher Education Commission. The AI program aims to improve capabilities in industrial sectors, warfare, and surveillance. In all probability, it should be assumed that China would be providing Pakistan with necessary know-how in the years to come about AI developments in both the civil and military domains.
India. The Indian Finance Minister in his budget speech for 2018-2019 had mandated NITI Aayog to establish ‘The National Program on AI’ keeping in view that India, as the fastest growing economy, should have a large stake in new and emerging technologies. The NITI Aayog has taken a three-pronged approach namely: undertaking exploratory proof-of-concept AI projects in various areas, crafting a national strategy for building a vibrant AI ecosystem in India, and collaborating with various experts and stakeholders.
NITI Aayog in its discussion paper on National Strategy for Artificial Intelligence has recognised that Artificial Intelligence (AI) has the potential to be disruptive. Further, with intelligent machines enabling high-level cognitive processes like thinking, perceiving, learning, problem-solving & decision making, coupled with advances in data collection & aggregation, analytics & computer processing power, AI presents opportunities to complement and supplement human intelligence and enrich the way people live and work. The NITI Aayog in its paper on AI focuses on how India could leverage the transformative technologies to ensure social and inclusive growth in line with the development philosophy of the government. The guiding thrust for the AI strategy was to identify applications with maximum social impact, absorbing from the recent technology advancements in AI elsewhere in the world, and leveraging approaches that democratise access to and further development of AI.
India has allocated $477 million in 2018 for its Digital India program, which is a program initiative to promote AI, machine learning, 3D printing, and other technologies.
“New and emerging technologies like AI and Robotics will perhaps be the most important determinants of defensive and offensive capabilities for any defence force in the future. India, with its leadership in information technology domain, would strive to use this technology tilt to its advantage,”
Narendra Modi, Prime Minister of India, 2018
Defence R&D. The strategic implications of AI in the perspective of national security have been studied by a multi-stakeholder Task Force comprising the Government, Services, Academia, Industry, Professionals and Start-ups under the Chairmanship of Shri N Chandrasekharan, Chairman, Tata Sons. The Task Force has prepared a road map for AI in national security and has already submitted its final report to Ministry of Defence. Its Terms of Reference included a global scan of AI applications, study of level of AI development in India in general and specifically in the context of defence needs, and to make recommendations relating to making India a significant power of AI in defence, specifically in the area of aviation, naval, land systems, cyber, nuclear & biological warfare including both defensive & offensive needs including counter AI needs; recommendations for policy and institutional interventions required to regulate and encourage robust AI-based technologies for defence sector; working with start-ups / commercial industry and recommendations for appropriate strategies of working with start-ups. It is understood that the Task Force deliberated upon issues such as; LAWS- Developing lethal autonomous weapon systems for air, ground and underwater for both human-in-the-loop and human-on-the-loop scenarios; Simulated War Games and Training- A key area where AI can play a crucial role in training the forces in a simulated environment; leveraging AI for unmanned surveillance; AI and Cybersecurity; AI in Intelligence and Reconnaissance; and AI in Aerospace Security.
Centre for Artificial Intelligence and Robotics (CAIR) of the Defence Research and Development Organisation (DRDO) in India researches specific areas of AI for the defence forces. The focus areas include; Communication and Networking; Communication Secrecy; AI, Robotics & Control Systems, and Command, Control, Communication & Intelligence Systems. Some of the projects of CAIR are:
-Network Traffic Analysis (NETRA) which can monitor internet traffic. It analyses voice traffic passing through software such as Skype, Google Talk and can intercept messages with keywords like attack, bomb, blast, kill in real time.
-AI techniques for Net-Centric Operations (AINCO). It is a suite of technologies for the creation of a knowledge base, semantic information reception & handling, interference reasoning and event correlation.
-RoboSen is a mobile robot system targeted at patrolling, reconnaissance and surveillance. It is capable of autonomous navigation in semi-structured environments with obstacle avoidance capability and continuous video feedback.
-Unmanned Ground Vehicle (UGV) for low-intensity conflicts and surveillance in urban areas.
Lethal Autonomous Weapon Systems, LAWS. A global debate has erupted concerning the moral, ethical and legal issues involved in the development and deployment of the LAWS. Currently, no such system exists, however, countries have deployed protective systems, with limited autonomy, in defensive roles. These systems do not search and attack targets but undertake a predetermined response to specific provocations.
India faces two hostile neighbours with unresolved border issues and an active insurgency in many districts which has put its military and paramilitary forces under a veil of constant threat. There is a need for India to develop weapons to defend its vast land border, coastline and assets in space with minimal risk to its forces. This would be best achieved by utilising robotic sentinels which can respond effectively and neutralise the arising threats. The DRDO had indicated in 2013 that robotic soldiers would be ready for deployment by 2023.
India has stressed the fact that technology such as that being developed for LAWS has both peaceful and military uses. In 2016, during the Convention on Certain Conventional Weapons, India has said that there is a need for “increased systemic controls on international armed conflict in a manner that does not widen the technology gap amongst states or encourage the increased resort to military force in the expectation of lesser casualties, or that use of force can be shielded from the dictates of public conscience.”
Some AI programs for the Naval Operations
AI can be a boon to the Naval Operations, the backbone to a successful AI program is the requirement of adequate and appropriate data, which may not be accessible readily due to security protocols. There may be a need to migrate the databases to a platform which is conducive to AI applications. AI would need a continuous feed of humongous data, which the Navy would have to acquire, access or procure at certain costs. It would be pragmatic for the Navy to set up a mechanism for deciding how much and what type of data would be required for specific AI applications. As far as feasible, the costs of AI acquisitions should be reduced by adapting the algorithms from those already in commercial use and only in exceptional cases should the algorithms be drafted anew. As an illustration of applications of AI in Naval Domain, three areas where AI would make a difference in naval operations are discussed below.
-Multi-Domain swarming threats. One of the situations that a warship or even a group of warships may face is saturation of its defensive systems by swarming the battlespace with fast attack boats or drones. The success/ partial success or failure would depend upon the response of the warship to engage multiple targets by allocating its resources (sensors and weapons) in order of emerging threats and their damage potential. Currently, the capabilities of the Warships to effectively tackle a saturating battlespace are limited at best. The swarming would in all probability occur across multiple domains of air, surface and subsurface requiring the warship’s weapons and decoying systems to engage simultaneously across all the multi-domain threats. The problem would get much more complicated with each type of threat deploying its offensive weapons against the warship, requiring not only offensive cum defensive response but also survivability effort to mitigate damage which could be sustained in such an environment. Thus, there would be a need to have a defensive plan such that these threats are prevented from emerging within the circle of the safety of the asset. The allocation of resources/defensive weapons to tackle imminent threats presents a massive decision-making problem once the number of targets and available weapons to engage them increase. The humongous effort at decision making that would be required has been evaluated by the US Navy’s Office of Naval Research, ONR. For illustration, for a multiple incoming threat of 13 targets to be engaged by 4 weapon systems, the number of unique matching combinations would be 52; however, the unique weapon-target matching sets to counter all threats would be 600 billion! It is humanly impossible to find a solution to such a problem since other essential factors like engagement envelop, range, type & quantity of munitions also have to be considered in the dynamic scenario. Leading Navies are considering the use of multi-capability launchers to launch different types of munitions from the same launcher, implying that the same launcher could be loaded with anti-air, anti-surface as well as anti-subsurface munitions. Thus, the operator would also have to keep track of each type of munition from each launcher during the engagement. Needless to say that AI could prove to be a helpful tool in resolving decision-making problems under a swarmed battlespace and enable a much faster kill chain response.
-Scheduling and decision support in Electro Magnetic Spectrum, EMS operations. The number of single-purpose RF antennae on a warship has increased manifold leading to associated issues like cross interference, modification to superstructure design, weight, and cost. The antennae are optimised for polarity, radiation pattern, frequency and power. However, since they are located on a warship where space is a constraint, other problems arise such as; null zones due to ship’s superstructure, unpredictable patterns for transmission and reception; interference from similar frequencies from other antennae, high power transmission interference with other receivers operating at lower or higher multiples of the larger antennae, power management difficulties encountered by omnidirectional and directional antennae; and inefficiencies in power & frequency management.
The warship at sea has to resort to a compromise while operating a broad mix of antennae; however, determination of the optimal mix is a tricky proposition since the change of frequencies, deconfliction, power and direction have to be optimised. Today, multifunction antennae are being designed and even utilised where multiple frequencies are transmitted from different sectors and are also received, two similar frequencies can also be transmitted separately and directionally. The communication systems are hardened against snooping and jamming in real time. The warships also have the capability to track multiple contacts across the electromagnetic spectrum and also share data in real time with other platforms and operation centres using high-bandwidth satellite communications and Internet protocol-based radio networks. There is a need to have a scheduling and decision support tool which can efficiently manage the communication and data resources. The tool should be able to respond to dynamically changing requirements of requests, spectrum availability, system failures, mission requirements, frequency agility and allocation of antennae. Compounding the current systems capabilities with those in future systems; which may utilise lasers, ultraviolet, infrared or even X-rays & Gamma rays; lifts the scheduling and allocation problem in to the domain of AI.
-Time-critical decisions during landings on aircraft carriers. On an aircraft carrier, landing operations are one of the most critical operations because of various factors like, space constraints of the flight deck, tactical situation, own manoeuvring, pitching and rolling, electronic emission policy in force, fuel and munition on the deck, position of other aircraft, status of aircraft in the air, and preparations on the deck. The officer in charge of landing operations is the Landing Signal Officer, LSO who ensures the final approach and a safe landing of the aircraft. LSO with his own flying experience can predict the trajectory of the aircraft and enable a smooth landing. Towards the end of the final approach cognitive demands on the LSO increase manifold especially in bad weather, or at night because he has to rely upon his instrumentation desk much more before he can decide whether to abort or allow a landing. The AI can play a significant role in designing a decision support tool for the LSO, which would permit taking better and quicker time-critical decisions where errors can result in the loss of lives.
Conclusion
The US DOD recognises AI (and not AGI) as an essential technology enabler for: network-enabled, autonomous weapons capable of operating in future cyber and electronic warfare environments; strategies for collaboration between manned and unmanned platforms; assisted human operations, human-machine collaborative decision making; and autonomous learning systems. As can be seen, the realisation of the above capabilities is not linked to the progress achieved in Artificial General Intelligence, AGI.
The way wars are fought in future is bound to undergo a quantum change with AI enabled sensors, weapons and munitions flooding the battlespace. There is also a distinct possibility that the battlespace would transgress more and more into the private domain as both the offence and defence capabilities could be operated from desks deeply ensconced in humble surroundings. However, the day of the complete ‘AutoHunom’(AGI enabled autonomous weapon system) is decades away.
At present, even an AI of tremendous power will not be able to determine outcomes in a complex social system, the outcomes are too complex – even without allowing for free will by sentient agents…. Strategy that involves humans, no matter that they are assisted by modular AI and fight using legions of autonomous robots, will retain its inevitable human flavor.
Kareem Ayoub and Kenneth Payne, 2015. “Strategy in the Age of Artificial Intelligence”.
Unbiased Jottings on Global Maritime Issues 