Artificial Intelligence (AI) Applications in Military Domain

Sub Title : Incorporating AI in defence systems: Need to keep pace with the developed world

Issues Details : Vol 14 Issue 2 May – Jun 2020

Author : Amit Baveja

Page No. : 63

Category : Military Technology

: June 1, 2020

AI is being seen as the next military advantage. The article details key military applications of AI and the advantages thereof. It is imperative that we act now to ensure that the pace of incorporating AI in our defence systems is in sync with what is happening in advanced countries and that we do not get left behind in this race. Despite its tremendous potential, there are also concomitant challenges for utilizing AI in military applications

The phenomenal expansion of AI based applications in diverse fields is revolutionizing multiple sectors including health & medical diagnosis, education, agriculture, predicting floods, budgeting, customer relationship management, supply chain management, creative arts, entertainment, and gaming, inter alia. It would not be an overstatement to say that AI is poised to disrupt our world and transform our lives irreversibly.

USA and China are presently the global leaders in AI. While the US growth has largely been driven by the tech giants such as Google, Amazon & Facebook, the US Department of Defense spent over $2.4 billion in 2017 on AI related technologies in the military, exhibiting the focus and intent of the government in leveraging AI in defence applications. China, on the other hand, aims to emerge as the global leader in AI by 2030. It has been funding university programs and incentivizing industry in a big way. UK, Germany & France have also been spending phenomenal amounts on R&D in diverse areas, including defence. With the global powers focusing on AI in a big way, India cannot afford to be left behind, especially in view of the enormity & diversity of security challenges that it faces across its wide geographical expanse. Needless to say, the growth of AI in defense can only be fueled by government funding & incentives. It is important to shape our priorities and act decisively to ensure that we do not lag behind in the AI domain lest we have to keep playing the catching up game as is happening in other areas of military technology.

The Niti Ayog, published a discussion paper on National Strategy for Artificial Intelligence #AIFORALL in 2018 wherein it recognized the tremendous potential of AI in transforming economies and laid down a pragmatic national strategy for building up an ecosystem for AI enablement and technological leadership for inclusive growth. The paper also acknowledged that India is a late entrant in the field and identified applications with maximum social impact as focus areas for AI intervention in the country. While the Niti Ayog paper did not list defence as one of the priority areas, needless to say, AI has phenomenal potential to transform the very nature of how militaries operate and how we can defend our nation.

The transformational potential of AI should be viewed as complementing human capabilities rather than merely replacing humans from various domains. Despite the ongoing debate about AI replacing human jobs, the fact remains that people would remain the drivers for adopting AI in various fields as well as benefit by capitalizing on its tremendous potential. The pertinent need for reduction in the size of the Indian Defence Forces has emerged as a major challenge, especially for the Army. The conflicting demands of ensuring national security and simultaneously cutting operating costs can easily be accomplished by using technological facilitators like AI. A scan of various AI applications can help us to classify them into two broad categories. On one hand AI can automate repetitive low value work, which humans generally find extremely boring, monotonous, repetitive, and arduous. On the other hand, AI also guides people on how to do their jobs better, thereby making them more efficient in carrying out their jobs.

Military professionals can identify multiple tasks in the two categories stated above. Additionally, the dangerous nature of military jobs that puts our soldiers in the harm’s way and exposes them to potentially fatal dangers; utilizing AI based systems to replace humans can help militaries preserve trained soldiers and save precious human lives. As General George S Patton had famously said, “No bastard ever won a war by dying for his country. He won it by making the other poor dumb bastard die for his country.” While in the yester years, the ways to implement this cliché may have been limited to meticulous planning and flawless execution by the maneuvering forces ably supported by fire support means, contemporary technology offers solutions in terms of AI powered systems to undertake various combat roles that can help in preserving priceless human lives.

AI Applications

AI has the potential to enhance the nation’s military power as well as alter the balance of power amongst nations. The changing nature of warfare wherein it is gravitating away from kinetic and maneuver warfare towards newer forms to include cyber, information and economic warfare, wherein at times we cannot even say whether or not we are at war, certainly brings to fore phenomenal number of AI applications to help achieve national goals.

Identification of military applications for AI and prioritizing amongst them may be done by using the twin criteria of (a) availability of a new technology that has the potential to solve the user problem, and (b) whether it addresses the needs that existing & available technology cannot address, may be used. Certain possible applications are discussed below:

Surveillance & Situational Awareness. The most obvious military task that comes to mind is surveillance or ensuring situational awareness. Every soldier who has manned an observation post or even an electronic/ electrooptical surveillance device would testify that it is extremely monotonous and arduous to continuously observe the area / screen for endless periods of time. Additionally, the old military adage that the first report is always wrong captures how fear and other psychological factors can skew the way humans interpret data. While modern surveillance devices have augmented the detection capability manifold, the fact remains that human omissions arising from inability to continuously observe can obviate the advantage of enhanced detection capability & make it absolutely meaningless. AI can enhance the efficacy of surveillance systems by ensuring zero misses in surveillance besides freeing a large number of soldiers for other combat tasks.

Information Warfare. AI systems can be designed to produce large amount of fake news and propagate it over multiple media, thereby lending itself to a wonderful tool of Information Warfare. Needless to say, such fake news can play a menacing role in social, economic, and psychological domains, thereby causing disruption and adversely affecting the adversary.

Training. Training needs of militaries, presently being met by field training as well as simulations can be better addressed by utilizing AI based solutions which can customize the content not only for each soldier (or each subunit/ unit), but also for different weapons, different locations over varying terrains and different adversaries, thereby augmenting the efficacy and value of training.

Wargaming. Wargaming is a key tool for militaries to provide experiential learning to commanders starting from the tactical levels at squads, platoons, and companies to operational levels. These also help in training the staff officers in sharpening their skills including helping them translate the commander’s intent into detailed plans, dissemination of the same and coordination with multiple stakeholders, which are extremely demanding in the present form. On the other hand, wargames have also been a time-tested tool to allow militaries to test new technologies and their impact on warfighting. Allowing AI algorithms to capture data over multiple wargames can help in progressive evolution of AI based systems that utilize referential learning to enable them to play an increasingly dominant role in actual operations.

C4ISR /C2I systems. Utilizing AI for facilitating combat decisions would help in effectively elevating the quality of inputs, their collation and analysis. It is likely to happen in an incremental manner starting with transmitting intelligence and orders; graduating onto collation of intelligence from multiple sources; further onto decision making by Manned-Unmanned Teaming (MUMT) platforms; subsequently to combat operations undertaken by totally autonomous, unmanned platforms; and finally in facilitating operational decisions at higher headquarters.

UAV Swarms. While unmanned systems lend themselves to utilizing AI, swarms are poised to usher in a revolution in unmanned systems. AI powered swarms of UAVs can be tasked to perform a host of functions ranging from surveillance and logistics to direct combat roles of delivering weapons on to pre-determined or impromptu targets. Identification and Security Access. AI applications using biometric recognition can be used in identification of soldiers and ensuring graded access to various military installations or military equipment for different personnel. These can also be used to accord varying rights to different members of a combat team or formation headquarters to take different types of decisions in varying combat situations.

Logistics. Logistic functions, employing both the push as well as pull models, have tremendous scope for utilizing AI based solutions. The sheer enormity of the data to be handled as well as the repetitive nature of the job that the logisticians perform make it a potential area for AI application.

Challenges for introducing AI in Military Applications

Despite tremendous potential, there are numerous challenges for utilizing AI in military applications, a few of which are listed below:

Data Centricity. AI can be best utilized in applications that involve processing vast amounts of data. As machines that can think, they need to be fed with phenomenal amounts of data to develop into effective systems. The primacy of confidentiality in the military domains remains a major obstacle in generating data and sharing it with researchers/ development agencies. This challenge, however, is not insurmountable and can be overcome by introducing confidentiality protocols and procedures for the researchers.

Transparency. Presently many AI systems suffer from a Black Box syndrome, wherein what is happening between the inputs and final outcomes is largely unknown. However, in the military domain, it is essential to know this due to the serious implications of blindly accepting the outcome without knowledge of the process.

Accuracy. The criticality of accuracy is also exceedingly high as it has implications on human lives as well as the security of the nation. While there is really no guarantee of even human decisions always being the best ones or even correct, there is bound to be reluctance in allowing machine learning algorithms to make such critical decisions.

Robustness and reliability. The vulnerability of AI based systems to cyber-attacks wherein its inputs can be manipulated poses a huge challenge adversely affecting the reliability of AI based solutions. This challenge would be significantly greater in critical applications of combat decision making, while its degree would be comparatively much lesser in support functions aspects like logistics.

Development Timelines. AI based applications have been developed much slower than what has been promised by various researchers/ organizations. A scan of the automobile industry wherein the leaders like Tesla, Google and Uber inter alia, have failed to meet their own ambitious deadlines of bringing driverless vehicles on the roads, is a testimony to the fact that AI based technologies do take a significant amount of time to mature, test and implement.

Ethical Debate. The ethical debate of AI versus humansor even AI replacing humans assumes a significantly different connotation in the business of protecting human lives or even ‘taking lives’. Can we delegate the authority of controlling missions or allow bots to take decisions where human lives are involved? This ethical debate extends to MUMT systems which are under development wherein a combination of manned and unmanned machines are assigned combat tasks.

Regulatory Obstacles. The absence of any dedicated procedural guidelines for AI systems in the ‘Draft Defence Procurement Procedure 2020’, which would remain the bible for procurement of new equipment & technology in the Tri-Services over the next few years, is going to be a major obstacle. The very nature of the development curve makes it pertinent to develop specialized procedures for development of AI based systems/ applications. The absence of defence as a priority sector in the Niti Ayog Paper also needs a de-novo look.

The Way Ahead

AI applications needs to be viewed in light of the transformative impact that they can have on warfighting. It would need to be included in the Niti Ayog’s roadmap for development of AI in the country as well as in the DPP 2020, so as to ensure that the existence of a broad framework for military applications.

A collaborative approach between the MoD, Tri-Services, academia and industry would be required wherein the MoD would need to play its role in terms of showing commitment to development of AI applications, funding the projects and facilitating incentives from other ministries & departments. The Tri-Services should lay down the priorities as well as provide handholding in terms of data, guidance and support as the AI systems are developed. Temporary project teams consisting of experts from military, academia and industry should be setup for each application which should work in sync with all stakeholders to meet time bound goals. The project teams should be given adequate autonomy & funding to facilitate achievement of these goals. Private sector may be accorded priority over PSUs so as to ensure that the goals are not scuttled by the constraints imposed on publicly funded bodies in terms of manpower hiring, procurements etc.

A graded approach, which allows evolution as well as diversification of developed applications and their introduction in the Forces as per their maturity would help in progressively enhancing the use of AI in military applications