Right from its Independence in 1947, India recognized that to advance in a rapidly changing world and establish itself as a nation without relying on the crutches of its erstwhile colonizers, it would need to focus on the development of its technological infrastructure. This was also a clear policy priority, and the early years of India’s nationhood involved planning and setting up what would be the first of many Indian Institutes of Technology (IITs) in Kharagpur (Council of Indian Institute of Technology n.d.). The decades following this saw a steady increase in India’s international partnerships around science and technology.

Another policy priority that remained consistently high on the agenda since Independence, given the geopolitically turbulent neighbourhood India found itself in, was that of national security. Surrounded by contemporary nuclear powers—Pakistan to the west and China to the east of its northern borders, both of which India fought several limited wars and border conflicts with—India’s drive for technological advancement also seeped into its security and military priorities (Weisman 1987). This is why, despite being a nascent nation at the time, India invested heavily in military technology, ultimately developing its own nuclear weapons in 1974 (Kristensen and Kile 2020) and its first guided missile system in 1983 (Rakshak 1998).

However, while India’s stake in the arms race undoubtedly accelerated during the Cold War era, its adoption of liberalization, privatization, and globalization (collectively referred to as the LPG reforms, National Council of Educational Research and Training 2023) in the early 1990s led to a shift in its security policy. While border clashes continued—and still continue at a reduced scale—trade relations with China and Pakistan improved irregularly but steadily following the LPG reforms. As a result, policy priorities for India seemed to shift from a security-dominant narrative to a more economy-oriented one, reminiscent of Immanuel Kant’s trade peace theory (Russett et al. 1998).

Over the next few decades, the race to securitize critical emerging technologies took a backseat for India in the face of renewed geo-economic aspirations, and investments in tech became primarily industry oriented. Simultaneously, the information technology (IT) boom hit India, and Bangalore became the Silicon Valley of India towards the beginning of the 2000s, taking India’s civilian tech and software industry to new heights both in metrics and magnitude. With the increased interest around artificial intelligence (AI) in the past decade, and especially in the past few years with the advent of generative AI, India is adjusting accordingly and now aspires to become a global AI superpower (Raja 2023).

Recently, military AI has been explicitly mentioned as part of India’s AI ambition (DDP MoD 2022). This ties into the larger trend of India beginning sporadic activity around military AI in the recent past, although not nearly enough to realize its ambitions. Several internal factors have acted as limitations, including the lack of a structured and organized national roadmap, sizable investments by the government, and the quality of AI talent.

This chapter explores these broad challenges by giving an overview of the defence AI ecosystem in India. It clarifies the ambitions and limitations that India is facing regarding its defence AI architecture, examines multiple aspects of its defence AI landscape as it currently stands, and concludes with a prospective analysis of new challenges and the future trajectory that India sees itself on. This chapter primarily uses open-source information unless explicitly specified.

1 Thinking About Defence AI

After being recognized as a key IT hub, India has repeatedly expressed interest in replicating a similar success in AI. In defence AI, it has a particular line of thinking which is congruent to both its primarily peaceful foreign policy of vasudhaiva kutumbakam (‘the world is one family’ in Sanskrit), and a pragmatic consideration of being competent enough to avoid being dominated by other States. The Defence Minister of India has occasionally stated that India has “no intention to rule the world, [but] we must develop the capability of our [defence] AI technology so that no country can even think of ruling us” (Press Information Bureau 2022d). However, the much higher prioritization of other general-purpose AI—with India exploring AI use (Deloitte 2022) in several industries, ranging from its signature digital public infrastructure (Singal 2023), public services (Elias 2023), healthcare (NASSCOM 2021), and education (India Today 2023), to finance (Menon 2023) and creative profession (Das 2023)—means that defence AI often gets pushed to the sidelines.

This is related to a market trend where India is known for its AI talent but not its AI innovation. India is now ranked amongst the top countries in the world for AI skill penetration and AI talent concentration (IndiaAI 2023) but comes up short compared to several other countries in terms of AI patents (Inside BigData 2020) and innovation (Cesareo and White 2023). It remains to be seen whether this is a function of its educational, skill and job market environments that prioritize building systems over creating them, of a proportional representation based on its large population, or merely a matter of time for it to catch up on innovation.

However, one recurring factor is that India has seen numerous knee-jerk developments and concentrated efforts around AI, but still has a hazy conception of how to establish a well-oiled AI ecosystem. India’s widely adopted ‘Make in India’ program for import substitution and manufacturing capability enhancement has led to several new startups around AI and software as a service (SaaS), and a sub-campaign to ‘Make AI Work for India’ has recently also gained support (D’Cruze 2023). Although the assumption is that most workstreams across the government will implicitly follow this campaign, there is no overarching governance system guiding this or explicitly linking this campaign to defence AI production.

The lack of institutionalization around AI poses specific limitations to how efficiently India’s AI ambitions are being realized. While the government appetite for defence transformation is present, with the Indian Prime Minister explicitly prioritizing replacing legacy military systems with updated ones in 2021 (The Indian Express 2021a), the fragmentation of its approach causes avoidable hurdles. Moreover, while the defence system brushes up on existing security developments, it risks being blindsided by emerging geopolitical and technological challenges for which government institutions are not prepared, leading to a pile-up of governance oversight.

This context shapes the way India thinks about the use of emerging technology for national security and defence. It is well established that India’s relations with its neighbours are rocky at best and hostile at worst. While this was a largely kinetic consideration mostly confined to India’s disputed physical borders up until recently, the geopolitical tussles have also gradually seeped into the virtual space. In 2023, India was the most targeted country in the world for state-sponsored cyberattacks—ahead of other popular targets like the US—and facing 13.7% of all cyberattacks globally (Roy 2023). Despite this concerning fact, India has not yet come up with its own cybersecurity law or policy document, instead relying on a dated IT Act from 2000 occasionally amended for current challenges, and the newly minted 2023 Digital Personal Data Protection Act, which primarily serves end-user protection (Ministry of Electronics & Information Technology 2024).

While cyberattacks are notoriously hard to attribute, the prevalent assumption is that most of these attacks were conducted by China and its occasional allies Russia and North Korea (Roy 2023). Considering that Pakistan is also one of India’s main geopolitical adversaries but is not included within the attack vector analysis signals that India’s updated security policies need to reflect its current realities, and not just its inherent rivalries. India has previously been cognizant of this fact and had attempted to respond to its military clashes with China in Doklam in 2017 (Chengappa and Krishnan 2017) and the Galwan valley in 2020 with a diplomatic and economic standoff as well as a ‘tech decoupling’ (Kumar 2020).

The latter was an especially gripping public move by the Indian government, which banned about 250 Chinese apps including TikTok, Shein, Aliexpress, etc., in four spates between 2020 and 2022 on the basis of the national security implications of these apps (Bhati 2022). This tech decoupling had borne very limited results since trade with China revived the following year (although the apps remain banned). However, one thing became increasingly clear to defence circles in India—security had definitively taken on a technological characteristic both on and off the battlefield, especially when engaging with more technologically advanced rivals (Off-the-record conversation with an Indian defence policy expert December 2023).

However, in the absence of a dedicated document delineating the thinking of senior military decisionmakers in India around defence AI, it is difficult to clearly see where India intends to take its nascent defence AI ecosystem. Some notional clarity about this is spread across press releases and leader statements, often indicating how defence AI is seen as an important tool for achieving military superiority, especially vis-à-vis over India’s adversaries. In parallel, AI is also seen as a substantive means to strengthen and modernize the country’s defence forces (DDP MoD 2022).

Moreover, there is also much confusion around where India’s responsible innovation and AI ethics priorities lie when it comes to defence AI. A prime example of this is India’s two-pronged approach on AI ethics and responsible AI (RAI) in the civilian and military spaces. India attended the Responsible AI in the Military (REAIM) 2023 Conference hosted by the Netherlands, where about 80 governments were represented. About 60 of them signed a Call to Action to include RAI considerations in their military ecosystems. Curiously, India was not amongst them (Government of the Netherlands 2023), and did not give a reason for not signing the Call despite having a clear RAI focus in its civilian AI landscape (NITI Aayog 2021a, b). This was either a result of contrasting priorities, a fragmentation of diplomatic decision making, or an example of India not aligning its security policies with majoritarian alliances in favour of its own security considerations. Such was the case when India was one of four countries to not sign the Nuclear Non-Proliferation Treaty (NPT) (Pai 2020).

Whichever the reason (or combination of reasons), the lack of either a comprehensive military strategy with respect to AI, or an updated AI strategy that includes the defence sphere, is a hurdle for India’s AI advancement. This is also set to be exacerbated by the advent of emerging AI technologies, like generative AI, and their new potential impact on defence, such as contextual threats of asymmetric conflict using AI like information warfare and deepfakes (Galston 2020). This has already been observed before—an Indian cybersecurity agency unveiled a wide disinformation network of bot accounts on Twitter running China and Pakistan-favourable material before and after the Galwan valley clash in May 2020 (Goyal and Priyadarshini 2020). This network comprised about 400-500 Twitter accounts that spread misinformation and released deepfake-videos of the clash in an attempt to implicate the Indian army (Goyal and Priyadarshini 2020).

Given this backdrop, India began thinking about civilian applications of AI in 2018 and released its national AI strategy the same year (NITI Aayog 2018). The strategy came at an initial stage of AI conceptualization by India and refers to AI as “a constellation of technologies that enable machines to act with higher levels of intelligence and emulate the human capabilities of sense, comprehend and act,” an understanding which has steadily become a lot more nuanced globally. While the area of robotics and AI was already established within India’s defence establishment, with a Centre for Artificial Intelligence and Robotics (CAIR) created under the Defence Research & Development Organization (DRDO) back in 1986, it remained mostly dormant until 2018 (Centre for Artificial Intelligence & Robotics (CAIR) (n.d.)). The following years saw an uptick in India’s defence technology developments, and an expansion of its slow AI revolution from the civilian space into the military one.

In 2020, the government organized the Responsible AI for Social Empowerment (RAISE) Conference (Press Information Bureau 2020b). NITI Aayog, the policy think-tank of the Indian government, also released a two-part report in 2021 on approaches toward operationalization of responsible AI principles for civilian AI. In a parallel process, defence AI achieved newfound prioritization in 2018—a multistakeholder taskforce on defence AI was created by the government under the chairmanship of renowned Indian industrialist Mr. N Chandrasekaran (Sarangi 2019). The taskforce submitted its recommendations in a report in June 2018 (Press Information Bureau 2018), on the basis of which a number of measures were undertaken to expand the scope of India’s defence AI.

2 Developing Defence AI

One of the first institutional changes made in the aftermath of the defence AI taskforce’s recommendations was the establishment of the Defence Artificial Intelligence Council (DAIC) under the Ministry of Defence (MoD), as well as a Defence AI Projects Agency (DAIPA) created under the Department of Defence Production (DDP) of the MoD (Press Information Bureau 2022c). In 2019, the MoD also created an AI roadmap for each Defence Public Sector Undertaking (DPSU), with about 70 initial defence-specific AI projects identified for development in the coming few years (Press Information Bureau 2022c).

The development of defence AI during the next few years gained momentum, as about 75 AI-based defence products and technologies, many of which were identified by the AI roadmap (Government of India 2022), were unveiled during the first “AI in Defence” (AIDef) symposium and exhibition organized by the MoD in July 2022 (Press Information Bureau 2022d). Later in the same year, the DDP published a full catalogue of their features, capabilities, applications, and advantages in the public domain, including forewords by key political entities and introductory facts and figures, titled “Artificial Intelligence in Defence: Presenting AI Preparedness of the Country in Defence” (DDP MoD 2022).

The catalogue mentions an array of frontier defence AI products that have already been developed and/or deployed—these include a few types of lethal autonomous weapons systems (LAWS), 3D-printed sentry systems, rail-mounted robots, autonomous intercept boats, AI-based swarm and storm drones, cognitive radars, unmanned vehicles, motion and anomaly detectors, target identification systems, facial and gesture recognition technologies, instantaneous translators, and monitoring and predictive systems (DDP MoD 2022). These products are being employed both on and behind the frontlines, in combat as well as in logistical and administrative roles.

This focus on the indigenization of defence technology production is also part of the larger Atal Innovation Mission (AIM), which is a whole-of-government approach to promote a culture of innovation across industries (Government of India n.d.-a). The resulting efforts of AIM mean that instead of simply importing, India has started prioritizing international partnerships and agreements around several types of military objectives and new technology, including defence AI. This process has also showcased a trend of altered relations with existing security partners and a renewed effort to establish new avenues of collaborative development and transfer of technology (ToT). An analysis of how India is organizing its international defence AI procurement and partnerships has implications vis-à-vis its foreign policy as well.

In terms of great power politics, India remained principally neutral throughout the Cold War, even though it shared friendlier relations with Russia—including defence ties. The Russian position retains a continued interest in India’s expanding defence AI sphere, as evidenced by articles on Russian state-owned media portals like Sputnik News which highlight the importance of India recognizing Russia as a potential academic and research partner for defence AI applications (Trivedi 2023). However, India has recently looked to diversify from Russian imports and collaborations on defence technology primarily due to the cascading effects of the Russia-Ukraine war (Waldwyn and Solanki 2023). While India and Russia have traditionally had close defence ties it is not clear that these ties impact all weapon systems at all times. India still relies on Russia for conventional weaponry and nuclear power (Sharma 2023) but seems to have diversified in terms of emerging military technology; within which the US, China, and their allies seem leagues ahead (Heikkilä 2022). Despite this shift, India and Russia continue to align politically on defence AI and were two of only five states that voted against a 2023 United Nations First Committee resolution expressing concern about the possible negative consequences of LAWS (United Nations 2023).

The lull in the India-Russia defence relationship has made space for other actors, especially Western powers, to forge closer alliances with India. The US, for instance, has convened and rekindled several strategic partnerships on defence technology with India over the past few years, including the Quad (US, India, Japan, Australia), I2U2 (India, Israel, US, UAE), a flagship U.S.-India AI initiative (USIAI), and the bilateral initiative on critical and emerging technologies (iCET). Both countries have also conversed several times (U.S. Department of Defense 2023; Ministry of External Affairs 2023) and pursued collaborations around defence AI, such as through critical technology working groups (Vergun 2023) and partnerships to counter common adversaries like China (Singh 2023). In 2023, the Indian MoD approved the procurement of 31 MQ-9 reaper unmanned aerial vehicles (UAV) from US-based General Atomics, which will be assembled at a new planned General Atomics facility in India (Renshaw 2023).

Another notable partner country that has allied with India on defence AI is Israel. The Indian Defence Minister and his Israeli counterpart adopted the ‘India-Israel Vision on Defence Cooperation’ in 2022 to strengthen the existing framework of bilateral defence cooperation on futuristic defence tech, deepen military cooperation, and co-produce new-age weapon systems (Press Information Bureau 2022e). Later the same year, India held a Defence Expo (Press Information Bureau 2022b) where several countries were present at ministerial levels in addition to businesses, investors, and startups. Here, the Indian company Adani Defence and Aerospace, along with its partner Israel Weapon Industries (IWI), unveiled ARBEL, India’s first AI-based small weapons-embedded Intelligent Fire Control System (IFCS) with motion sensors (ET Now 2022).

India is now also in talks with other potential defence tech collaborators at various levels. France and India concluded their fifth Annual Defence Dialogue in 2023 with a promise to work together in niche military domains such as space, cyber, and AI (Press Information Bureau 2023b). Australia and India explored the possibility of co-production and joint skilling in defence AI in their 2023 2 + 2 Ministerial Dialogue (Suman 2023). Preceding a similar 2 + 2 Dialogue in 2022, Japan and India also identified key areas of defence cooperation, including UAVs, anti-drone systems, robotics, and intelligence systems (Singh 2022).

India’s interest in these conversations around defence AI is only expected to increase given the nascent status of its ecosystem. Currently, many of these defence tech conversations happen at ministerial levels or higher and may or may not trickle down to actual cooperation. And if they do, the former may not always be reported. For most such conversations, either the Ministry of External Affairs (MEA) or the MoD are the torch bearers. Since the AI roadmaps for DPSUs mentioned earlier are not publicly available, it cannot be said with certainty whether these conversations and potential implementation plans are all part of one constructive engagement policy around defence AI.

3 Organizing Defence AI

A distinct characteristic of India’s defence AI organization is its mixed approach model which lacks a central decision-making body. Design, development, deployment, proliferation, and control of defence AI is largely spread across the central government through the MoD (and rarely, the Ministry of Electronics and Information Technology or MeitY), the 16 DPSUs that work on specific projects under the MoD, the DRDO and its sub-centres, and the three services (army, navy and air force) themselves. India has no private military contractors (PMCs), so any other inputs or action around development and regulation of defence AI is often supplemented by the established industry, start-ups, academia, and the civil society. However, interplay between these external actors and the government is usually limited and largely unidirectional.

The services, while ultimately within the ambit of the MoD, are also individually forthcoming in their defence AI engagements and have established their own institutional mechanisms in the form of an AI Sub Committee and a Joint Working Group on AI for the Tri-Services (Press Information Bureau 2022f). To ensure that their foundational structures are AI-ready, especially considering the copious amounts of data that AI will require, they have also formulated a data policy, convened a Data Management Framework, and appointed Data Management Officers (Press Information Bureau 2022f).

While defence AI research is usually unilaterally undertaken by academia and civil society actors independent of government oversight, the MoD has recently showed more interest in collaborating with the industry. The Ministry has started commissioning defence AI projects from startups through a new AIM avenue called Innovations for Defence Excellence (iDEX, Government of India n.d.-b). Notable examples include AI-based ground and air-mounted systems developed for the Indian military by indigenous startup Skylark Labs (Gandharv 2022), and unmanned marine vehicles (UMVs) developed by Pune-based startup Sagar Defence Engineering (Nath 2023). iDEX has been established by the conjunction of two DPSUs (Hindustan Aeronautics Limited and Bharat Electronics Limited) and is guided by a new Defence Innovation Organization (DIO).

4 Funding Defence AI

While India’s scale of engagements around defence AI both nationally and internationally has been substantial, the litmus test of its prioritization with respect to the larger defence and tech architectures in the country can be seen through its monetary evaluation by the government. In 2022, the MoD announced the earmarking of a specific ‘AI budget’ from the amount disbursed as part of the yearly defence budget, with a corpus of INR10bn (approx. USD120M) to be provided each year for the coming 5 years to support defence AI activities (Press Information Bureau 2022f).

While this may seem significant, it barely makes up a fraction of India’s 2023 defence budget—India allocated INR5.94tn (approx. USD73bn) for defence in the 2023-24 financial year, which makes the AI budget amount to less than 0.002% of the overall defence budget (Press Information Bureau 2023a). This is in spite of the fact that about INR1.62tn (approx. USD20bn) of the total budget was allocated to capital outlays pertaining to modernization and infrastructure development, of which a record 75% of the defence capital procurement budget was earmarked for domestic production (Press Information Bureau 2023c). Additionally, INR23bn (approx. USD2.8bn) was allocated to defence R&D through DRDO, which makes up almost 4% of the total defence budget, but DRDO’s spending on and outputs in terms of defence AI have not been significant.

While India’s magnitude of spending on defence makes it the fourth largest spender globally after the US, China, and Russia (WiseVoter n.d.), some estimates have calculated that over 50% of its defence budget is actually spent on its 1.4 million active personnel and their pensions, limiting the scope and resources for defence procurement and modernization (McGerty et al. 2023). Moreover, contrasted with other digital initiatives by the Indian government in 2023, the defence AI budget is a paltry sum. For instance, there was massive government support for and spending on digital public infrastructure (DPI) in 2023, and INR15bn (approx. USD180M) was allocated for the promotion of digital payments alone (Aryan 2023).

5 Fielding and Operating Defence AI

Despite limited funding, AI has begun to be deployed by each of the three services in various battlefield, logistical and analytical functions. The army, driven by border concerns, has mostly experimented with fusing AI with legacy systems and conventional warfare capabilities. The navy and the air force, in comparison, have been more forthcoming in testing AI and related technologies for a number of more creative applications, many of which do not involve conventional warfare.

5.1 Army

The army has installed about 140 AI-based surveillance systems—including high-resolution cameras, sensors, UAVs, and radars—to get live feeds from the Pakistani and Chinese borders (IndiaAI 2022). These feeds are aggregated and help the army get a more comprehensive idea of potential border intrusion detection, target classification, and enhance the accuracy of defence operations. This system is supplemented on the Line of Control (on the India-Pakistan border) and the Line of Actual Control (on the India-China border) by a Proactive Real-time Intelligence and Surveillance Monitoring (PRISM) system, which also assists in threat identification by generating multiple real-time audio-visual feeds of disturbed areas and generating alerts for suspicious movements (Mishra 2022).

Another breakthrough by the DRDO which is now in use by the Indian Army is the evolution of the Seeker Monitoring and Analysis System, which is touted to be a self-contained system with facial recognition technology (FRT) in addition to a surveillance, monitoring, and analysis system (DDP MoD 2022). It uses a novel facial recognition system under disguise (FRSD) which can identify individual faces even in low resolution settings with the addition of different clothing articles and accessories (Verma 2022). Several other similar tools have been employed by the army to track vehicles (Project V-logger), identify intruders using intrusion detection systems (Sarvatra Pehchaan), as well as autonomously detect humans using facial recognition rail-mounted robots (Silent Sentry) at the northern and western borders of India for real-time threat monitoring (Bommakanti et al. 2023).

The army has also developed its own swarm and storm drones with Beyond Visual Line of Sight (BVLOS) attack capabilities, as well as drone feed analysis systems which India may consider exporting (DDP MoD 2022). Additionally, to better understand and collect information from adversaries, the army has started to equip soldiers with Natural Language Processing (NLP)-based wearable language translation devices. These are light, convenient to wear, and have a high battery life, and have been used to bidirectionally translate from Mandarin to English and vice versa (DDP MoD 2022).

5.2 Navy

The navy has had plans to integrate big data and AI within its systems since 2018, with concrete plans to develop and deploy AI into its operations (The Hindu 2018). In the same year, it also began looking into acquiring several types of unmanned underwater vehicles (UUVs)—significant among these was the memorandum of understanding (MoU) signed by India-based Mahindra Defence and Israeli company Aeronautics Ltd. for the latter to offer a maritime version of the Orbiter 4 UAV for the Indian Navy (Naval Technology 2018). The Orbiter 4 is an advanced multi-mission platform with an ability to carry and operate two different payloads simultaneously. It will be shipborne by the Indian Navy to carry AI-based sensor payloads which can be launched and recovered from small warships that do not have a helicopter deck (IndiaAI n.d.).

The navy unveiled its indigenous flagship Autonomous Fast Intercept Boat (AFIB) in 2022, which can perform autonomous operations for special forces, search, and rescue, patrolling and surveillance, as well as interception of high-speed vessels, even in dense maritime traffic and shallow water (DDP MoD 2022). Other functions within which the navy has recently deployed AI-based systems includes maritime motion pattern recognition & anomaly detection, acoustic and magnetic signature analysis, and AI-enabled voice transcription software (DDP MoD 2022).

More recently, the navy launched its first UUV, called ‘Neerakshi’ (‘eyes in the water’ in Sanskrit), developed by state-owned Garden Reach Shipbuilders and Engineers (GRSE) and Aerospace Engineering Private Limited (AEPL), which is deployed to help with underwater surveys, and mine detection and disposal (Malin 2023). The navy is also currently testing its AI-powered Combat Management System (CMS) for naval ships, which enables rapid threat assessment and algorithmic decision-support tools and will be built into all warships commissioned from 2024 (Malin 2023).

5.3 Air Force

The Indian Air Force (IAF) has been employing AI in various domains and workstreams as well. On the battlefield, an Enemy Aircraft Activity Recognition & Classification system has been developed for air defence systems, which uses AI to help identify enemy aircraft and employs predictive analytics to chart their plan of action (DDP MoD 2022). The DRDO has also developed novel manned-unmanned teaming (MUM-T) capabilities for a yet-to-be-released Twin Engine Deck Based Fighter (TEDBF) program, which will replace Russian Mig-29 K fighter jets currently in use by the IAF (Indian Research Wing 2023). MUM-T capabilities are also being explored by the DPSU Hindustan Aeronautics Ltd. (HAL) for the IAF’s ‘Loyal Wingman’ warrior drone, which will be tested in 2024 (Chopra 2023).

HAL has also helped to develop a Voice Activated Command System (VACS) for IAF pilots, which recognizes their voice commands and sends the codes to the mission computer for actions like radio tuning, mode selection, navigation, etc. (DDP MoD 2022). Also on the logistical end, the IAF has delved into employing predictive monitoring for the maintenance of weapons and aircrafts through a system called PRO-HM+, which can identify trends, patterns, and relationships of aircraft behaviour, equipment failure and other future events (DDP MoD 2022). It uses descriptive, predictive, prescriptive, and prognostic layers of data analytics and has been known to predict maintenance requirements of aircrafts with a high confidence level (DDP MoD 2022).

On the administrative front, the IAF created an AI-based Campaign Planning and Analysis System (CPAS), which is utilized to provide functionality and efficacy for campaign planning and debriefing solutions for all aircrafts under its ambit (Bordoloi 2022). The IAF has also deployed an application that integrates all electronic intelligence gathered by it and other intelligence agencies to create a comprehensive electronic order of battle (EOB) (Bordoloi 2022). MeitY has collaborated with IAF to develop an imagery intelligence analysis tool to identify assets and assess enemy targets by employing AI on the reconnaissance data (Bordoloi 2022). Creatively, the IAF has also experimented with using virtual reality (VR) for wargaming and cadet training at various air force stations (Raksha Anirveda 2023).

6 Training for Defence AI

Most of the Indian military personnel are still used to legacy systems, which is why training them for new technology like defence AI systems is an essential part of military transformation. This has been taken up by both the government as well as the three services, and new centres and programs are now being created to ensure optimum training. However, since the introduction of defence AI systems is recent, it is difficult to say how appropriate, effective and/or reliable the training will be– there has been limited, if any, monitoring, and evaluation of either the deployed AI systems or their training modules (Off-the-record conversation with an Indian defence policy expert December 2023).

Since most defence AI systems run on massive amounts of data, which several sections of the commanders and operators have not had to directly deal with until recently, a focus on training administration personnel for data management and the establishment of data centres has been the first step towards institutionalizing defence AI training. On the recommendation of the defence AI taskforce from 2018, the government acknowledged a need to scale the existing capabilities of military data centres; establish a centrally facilitated network of test beds; create a framework to work with industry partners; and encouraging start-ups to help the services with developing AI systems to manage their intellectual property (Press Information Bureau 2022f). The services have implemented some recommendations from the taskforce as well. They launched a Data Management Framework in 2022 which included, inter alia, agreeing on a data policy, establishing a Data Management Office, and appointing officers to manage it (Press Information Bureau 2022f).

Active defence personnel are also being trained through universities and training centres set up by the three services. The Army established an AI Centre at the Military College of Telecommunication Engineering in 2021, with support from the National Security Council Secretariat (NSCS) (The Indian Express 2021b). The Navy set up an ‘AI core group’ and designated its training institution INS Valsura as an AI centre of excellence, within the ambit of which it holds regular webinars, training sessions and workshops around AI (Indian Navy 2022). The Air Force founded its own Centre of Excellence for AI under the aegis of UDAAN (Unit for Digitisation, Automation, Artificial Intelligence and Application Networking, the word ‘udaan’ also means ‘flight’ in Hindi) in 2022 (Press Information Bureau 2022a) and has also held workshops on the integration of AI and other frontier technologies (Press Information Bureau 2023d). Most of these centres train cadets on how to use existing AI systems, while some like the DRDO may also delve into conducting research to develop new ones. Sometimes, civilian experts and think tanks are brought in to discuss dual use tech and conduct short-term training for certain systems, but these are entirely dependent on the services taking the initiative.

The DRDO has also taken up training and combined it with innovation defence AI innovation. It announced in 2020 that it would create five new laboratories around AI and other frontier technologies, named the DRDO Young Scientist Laboratories (DYSL)—the labs would focus on AI at Bengaluru, quantum technologies at IIT Mumbai, cognitive technologies at IIT Chennai, asymmetric technologies at Kolkata, and smart materials at Hyderabad (Press Information Bureau 2020a). As per the norms laid out, everyone at these labs, including the directors, should be under 35 years of age (Shukla 2020). However, it has been observed that it is difficult to find capable talent to staff these laboratories, given that most young scientists and engineers from premier tech institutions end up in the private sector which has a variety of other benefits and pays much higher salaries than the DRDO (Off-the-record conversation with an Indian defence policy expert December 2023).

7 Recommendations for India’s Future Defence AI Trajectory

India has made considerable advances in defence AI for a nation that did not have an institutionalized conception of the technology until a few years ago. While there is no institutionalised metric to measure how significant the advancement has been, the magnitude of advancements itself seems considerable when viewed in the context of the rise in AI-related developments over the past 5 years or so. Whether all these developments, most of which only have singular visibility on paper, have been successful or worked as expected is a different performance-based metric altogether. This underlines the limits of evaluation since we have no concrete monitoring and evaluation data.

Given this context, it is plain to see that several gaps still exist in India’s defence AI ecosystem. It aspires to be a leader in defence AI, but the country’s military leadership still devotes limited resources to its development. This also means that there is little incentivization for India’s AI talent to join the defence AI architecture. There is also a lack of interoperability amongst different AI applications and subsystems across civilian and military spaces, especially since India’s only national AI strategy from 2018 is now decidedly outdated, leading to contrasts in AI policy priorities and approaches.

Going forward, India needs to ensure that it looks at AI as the highly diversified technology it is, instead of the siloed approach it has adopted with respect to its development and policy so far. An initial recommendation is to adopt a hydra approach to its AI ecosystem—this would consist of amalgamating its core AI philosophy and roadmap into one apex body and creating an updated national AI policy; additionally retaining and producing separate sectoral guidelines, bodies, standards, and strategies to ensure effective policy synthesis and execution at various levels, including defence (Mohan 2023).

It will also be crucial for India to ensure that it can establish systems for the prioritization of its defence AI to be both robust and responsible, efficient, and ethical. This should ideally include more government investment in military transformation, better training and incentives to ensure that quality talent can innovate and operate new systems, and a well-functioning cycle of monitoring, performance measurement, and evaluation, especially since the current defence AI paradigm in the country noticeably lacks the latter. There should also be an optimum constructive interaction amongst technologists, military personnel and policy experts who understand this nexus of issues that defence AI will continue to present. Consistent and multi-faceted actions around responsible innovation in all major tech and military developments is on the way to becoming a requirement, at least at a policy level, to participate in global networks, and India will need to carefully consider which priorities it values as it builds upon its defence AI ambitions.