Artificial Intelligence (AI) is amongst the latest technology that is revolutionising both the commercial sector and daily household as well as military and homeland security arena. It provides limitless autonomous mission functionalities in time and space domains. This enables the agencies to carry out uninterrupted tasks of surveillance, sensor fusion, intelligence, enhanced decision support, operational logistics and sensor-weapon-shooter.

In early 1990s, Tac C3I and C4I2SR (many Cs continue to get conceptually added with time and in technology race) programmes of Indian Armed Forces started as a programme to achieve Net-Centric Operations (NCW, Effect Based Operations and many more acronyms). Their full effective operationalisation in battlefield is still at a distant future.

In today’s modern digitised battle and operational space, India has a modernisation statement that “AI technology will Transform Future Warfare”. In contrast, USA Department of Defence (DoD), primarily to counter AI advances of China has decided that “US Armed Forces has to invest in AI in battle space”. The difference here is between ‘will’ and ‘has’.

While India continues to procrastinate to use AI in battle space, USA has moved ahead to implement it. Currently, India has permitted use of drones (semi-autonomous) for delivery of medicines and vital organs. But the crashing of a drone in end-December 2022 on Delhi Metro tracks has raised many security and safety questions. It remains to be seen whether the decision makers take a step forward or retract. Hopefully would be the former.

On the other hand, as recently as November 2022, California has passed a law to allow use of Killer Police Robots. Of course, the house continues to be divided on ethical and moral issues. Kamikaze drones demonstrated during the Army Day’s Parade of 2021 was creditable. Hopefully it progresses beyond demonstration mode.

India’s AI programme has to address entire battle space. China is India’s major adversary and is way ahead in Military AI domain.  There is an urgent need for India to take two quick concurrent hops to achieve remotely piloted (semi-autonomous) and autonomous AI.

Government, Military and Homeland Security users, R&D and Industry can no longer be content with seminars, conferences and discussions and technology demonstration. They need to invest in AI operational concepts, indigenisation in AI assisted platforms that can perform specific military, security and civil tasks. R&D in AI is expensive and India, in comparison with China is behind in AI.

Import of AI systems and platforms, that are not under export ban, would need to be customised as per Indian SOPs and battle, drills and environment. Further, to meet the flux in digital space, AI ecosystem has to relearn and reform throughout the life cycle. Current fragmented organisation is unlikely to meet the goals of AI. Information Communication Technology (ICT) and AI tasked Armed Forces are in watertight silos of the three services.

The degree of difficulty to integrate and synchronise various entities increases many-folds due to further sub-allocation of development tasks and resultant independent approach by various line directorates. Armed Forces need to assign consolidate ICTisation (AI, C3, Cyber and Cloud) efforts under one umbrella. Going by the best practices in modern armed forces of the world, Signals are the CIO and responsible for infrastructure development; its operation, protection and hosting applications.

In Cyber and IoMT world single agency coordination and control are a must. At National level GoI, under NSA’s leadership needs to set up a national task force on AI. MoD would be best suited to head this task force. The current task force setup by Ministry of Commerce and Industry has ‘Vision’ and ‘Mission’ that is for economic benefits. Their vision can be addressed by simple semi-autonomous AI, (www.aitf.org.in). National Security is not their priority. Neither they have the domain expertise to lead it.

AI is a dual use technology. Complex Military and Security models can easily be downsized to meet civilian use and achieve their main objective of ‘Economic Benefits’. A snap shot of their objectives and roles is given below.

AI Limitations in Military Use

AI platforms in fully autonomous military role are still bordering science fiction. AI functions perfectly in a controlled and uncontested domain. Tesla autonomous car performs well in an environment where majority of the people follow laws and rules. Tesla in Indian roads where road space is aggressively contested by drivers and traffic rules are broken more than followed, would under perform.

Adversaries at war never follow conventions and annals of military history indicate that rules are often violated. Battle space is highly contested, subjected to constant targeted Information Warfare. Attacks to degrade and deny communication infrastructure will enhance AI complexity many folds. The major resultant drawbacks of AI in intense digital battle fields are: –

• Restrictions of sharing of operational data: live and operational data is not made available to the AI developers. Absence of test data makes the AI models unreliable.
• Considerable Computing power and data availability at the edge: Computing power of modern processors is considerably large. Even the mobile handsets have processors that are AI capable. Storage too has miniaturised considerably. The only limiting factor is connectivity (Bandwidth) to the cloud. A trade-off has to be made between AI at the edge and in the cloud.
• In contested areas AI (model) can easily be fooled.
• Inability to multitask: AI can only perform a task that it has been trained for. E.g. an AI model trained to recognise a ZTZ99 (type99) tank of China may not recognise Type 69 MBT of Pakistan, if not already trained for.
• AI has limitations of fully understanding the inputs and the context within the inputs. Even images are converted into numeric representations for computational processing. During these conversions context is often lost.
• Susceptible to Attack by Adversaries: In Battle space AI will be subjected to attacks by adversaries. Evasion, Inference, Poisoning and Extraction attacks can easily be carried in the Cyber space to fool the AI system. Robustness of AI in fully autonomous mode will remain a major challenge.
• Indian AI Ecosystem: Transformation from linear thinking and manual action to lateral thinking and automatic process is a major behavioural change. Considering the Indian social fabric this would be time consuming and critical for adaptation by both developers and end users.
• Accuracy, Precession, Sensitivity and Specificity assume different importance in various scenarios. False-Negative signals, while screening for terrorist entry into a vitalarea is undesirable in an AI model screening for terrorist. Similarly, false-positive signals while carrying out seek and destroy AI mission will have disastrous results in a populated area. Cent-percent efficacy of an AI model, in contested environment is still suspect.

Organisational Limitations

Study of various organisational models of the world are a pointer towards considerable dissimilarities between the Indian Military and that followed by modern western armed forces. Though Indian Armed Forces are the best judge to decide the organisation, they need seriously to consider course corrections, also.

Fragmentation of Cyber Space into various agencies may not be a good idea. Creation of Defence Cyber Agency (DCA) with the smallest Service as the lead may prove to be restrictive in consolidation of rapidly and ever-growing digital battle space.

US has a global reach. Accordingly, the US Cyber Command (USCYBERCOM) is one of the eleven unified cyber commands. These commands span from Army – Air Force, Fleet Commands, Marines and Coast Guards. USCYBERCOM plans, coordinates, integrates, synchronises and conducts activities to direct the operations and defence of various Defence information infrastructure. Organisational changes to Cyber Space agency that will perform the above-mentioned tasks cannot be delayed any further. DG Signals, being the most skilled and resourceful organisation in ICT is clearly the front runner for this role. Equitable distribution of military turf cannot be the basis of allocation of responsibilities.

Some Important Aspects that need Consideration

Digital Modernisation strategy must be simple, as it is the cornerstone for advancing a digital environment that would offer definite strategic and tactical advantages to the joint forces in the modern battlefield.

Digital Modernisation Strategy must be in sync with National Defence Strategy. In modern digital environment, Digital Modernisation Strategy provides a roadmap to implement National Defence Strategy. It is mandatory that the National Defence Strategy should be appropriately modified with Cloud, AI, command, control, communications and cyber security as the structural fulcrum. These fulcrums are equally applicable for civil and administrative functions and businesses.

Way back in 1999 David S Albert, John J Garstka and Fredrick P Stein, in their book ‘Network Centric Warfare’ enunciated physical communications being the “Entry Ticket” to Network Centric Warfare. This prerequisite still remains the same, with of course few add ons.

Communication infrastructure has now become all pervasive to include Cyber, AI, Cloud and C3 (command, control and communications) as the primary objective. Digital Modernisation also has these as the foundation. Operationally, Lethality (Kill, Success) in competitive environment by millions of interconnected entities is determined by their ability to act in unison and  in partnership. They need to learn and adapt to the highly contested battle space through integration and synchronisation.

AI will replicate the Battle Drills, SOPs and good old OODA loop in digital battle space. Digital Battle Space is in a state of continuous flux; the process needs to learn, adapt and reform. Similarly, the process that has competitive advantage in the battle space, survives as well as outperforms in highly contested digital battle space and has a skilled, trained and adaptive workforce to exploit the systems which will decide victory or defeat.

Optimum utilisation of existing ICT resources and their reformation with advances in cyber space is of paramount importance. Reorganisation and role of DG Signals, led by a SO-in-C needs a major refresh to make it suitable to plan, coordinate, integrate, and synchronise operations in cyber space. ICT infrastructure may be managed and operated in Service Specific compartments. But total denial of access to the resources (virtualised infostructure) to other services will never lead to an integrated modern battlefield.

Apex body has a major task cut-out to build trust amongst the three services in each other’s cyber security policies, rules and procedures. Common Tri-service backbone of Network For Spectrum (Project KRANTI) is a good beginning in this regard.

Information Warfare in Modern Digital Battlefield is no longer confined to SIGINT, ELINT, PsyOps, OE Ops etc. In data infested modern digital battlefield, the battle-space is densely populated by thousands of military interconnected entities. Identification of targets based upon search and monitoring of Electro Magnetic Spectrum (wireless) will not yield any results.

Cyber space (Cyber RF) gains more importance in modern digital battlefields unfolding in the near future. EW units need to be made capable of performing IW related tasks in Cyber Space. Electronic Support Measure and Electronic Counter Measures besides Electronic Counter-Counter Measure tasks can no longer be confined to wireless communication infrastructure.

The need is to address entire EM and Cyber space to include Remotely Piloted Vehicles, Autonomous Unmanned Platforms, Virtual ICT information Infrastructure-Cloud, AI, C3 etc.

Cyber space, due to ease of denial is being targeted on a daily basis. So, AI, cyber warfare and Digital Battlefield appears to be an ideal tool to employ in No-War-No-Peace, Restricted Warfare and conventional war. This is likely to change soon. Growth in intensity of cyber-attacks will soon have well defined red-lines. Retaliatory actions will soon become a norm. Operators in grey zone will not go unpunished.

India’s major adversaries are nuclear powers. Under these circumstances placing vulnerable AI systems in contested domains and making them responsible for decisions will have disastrous results. The consequences of faulty decisions, even in conventional war, between two nuclear power states may lead to a full-scale nuclear war. Humans must take and remain to be responsible for Key Decisions.

Indian Armed Forces need to address transformation in a modern AI enabled Digital Force on priority. The approach has to be one of collaborative cooperation and not competition. Coordination, Integration and Synchronisation are basic principles of digital battlefield and AI is a catalyst for this.

These basic principles have to be deep-seated while enhancing military capability building and force restructuring. Military thought process and ethos have to move towards integration and synchronisation. The first step is trust in each other’s professional capability and competence and then to overcome organisational and structural weakness in a service by the strength of the other.

-The writer is a Signals veteran. He has vast experience in Electronic Warfare in CI/CT and conventional operations. He led the implementation of ICT infrastructure during test-bedding of Tac C3I systems of the Indian Army as a Director. He was Project Director, ASCON Phase-3 and as MoD Rep was involved in finalisation of NFS scope and budget. He holds PG in Computer Science from IIT Madras and PG in AI from Austin University, US.