Policymakers must have a sound understanding of the objectives that AI seeks to achieve in the strategic context of India to disseminate artificial intelligence in defence. What kind of AI do we want?
By Ambuj Sahu*
Artificial Intelligence (AI), also dubbed as the Industrial Revolution 4.0, has been making giant strides in scientific and technological innovation across varying fields. It is capable of bringing significant transformations in the way civilian activities and military operations are conducted. Till now, the idea of attaining military superiority was conceivable only to a few countries like the US, China and Russia, who maintain large armed forces. AI, being a dual-use technology, may have interesting implications on the distribution of military power in the future. The possibility of AI-ushered advancements has opened the scope of an arms race where the conventional military capabilities will matter much less as time progresses. As a result, middle powers leading in civilian AI-tech also has the field wide open to compete for hard power. In this light, India is hard-pressed to enter the AI race in defence sooner rather than later. In January 2019, Army Chief Gen. Bipin Rawat was noted to have said , that India will be too late if the armed forces do not embrace AI soon enough.
Although there is a broad consensus on what AI is, i.e., carrying out tasks that can be performed by humans through a computer or digitally-controlled robots, there are diverse opinions about how AI can be achieved. At a time when data science is the new norm in the tech industry, it is perceived — in popular understanding — that it is inseparable from artificial intelligence. On the contrary, machine learning is only one of the tools that has contributed to the creation of AI technologies in addition to natural language processing (NLP), robotics, autonomous locomotives and other technology mediums. Hence, it is vital to take up a holistic understanding of the AI landscape in India and not limit it to the lack of data sciences infrastructure in the country.
This article discusses the developments in artificial intelligence in the defence sector and analyses the prospects and challenges that might be faced by India in the near future. It surveys the institutions and initiatives around which the AI policy of India can be expected to revolve around. Subsequently, it lists the barriers against the proliferation of AI in India’s defence and highlights the fundamental questions that policymakers should address before embarking on an AI programme.
What has India done so far?
The Ministry of Defence (MoD) constituted a multistakeholder task force for Strategic Implementation of Artificial Intelligence and Defence in February 2018 that submitted its report in June. The MoD implemented the recommendations by providing an institutional framework for policy implementation, issuing guidelines to the defence organisations, and laying out a vision for capacity-building.
In February 2019, the ministry established a high-level Defence AI Council (DAIC) under the chairmanship of Minister of Defence assigned with the task of providing strategic direction towards the adoption of AI in defence. The DAIC will guide the partnership between the government and industry and also review the recommendations concerning the acquisition of technology and startups. It also envisions the formation of a Defence AI Project Agency (DAIPA) as the central executive body.
The ministry ordered to focus on the capacity building within defence machinery. The tasks range from the knowledge production in the form of data collection, patents etc to acclimatising the personnel on-duty through internships, training programmes and sabbaticals. Each Service Headquarter (SHQ) will be provided with a window of Rs 100 crores for AI specific application developments from the ministerial budgetary allocation. The task force recognised AI as a ‘force multiplier’ and emphasised that all the defence organisations lay down their strategies of AI appropriation.
As discussed earlier, robotics is also one of the ways to achieve AI. The Centre of Artificial Intelligence and Robotics (CAIR) in the Defence Research and Development Organisation (DRDO) has also developed autonomous technology-based products. It has focused on the net-centric communication systems for tactical command control. For surveillance and reconnaissance purposes, CAIR has developed intriguing probes like snake robots, hexa-bots, and sentries. It has a comprehensive library for AI-based algorithms and data mining toolboxes that can potentially be used for image/video recognition, NLP, swarming. However, in a data-based approach to artificial intelligence, efficient learner algorithms can only serve a limited purpose without the hardware that can collect and process a large amount of data.
Few questions to be addressed..
The key challenges for the adoption of artificial intelligence, in general, are elucidated in the Niti Aayog’s 2018 document titled ‘National Strategy for Artificial Intelligence #AIforAll.’ In addition to this, there are many challenges that AI brings up in the military sector.
Firstly, policymakers must have a sound understanding of the objectives that AI seeks to achieve in the strategic context of India to disseminate artificial intelligence in defence. What kind of AI do we want? Do we require fully autonomous drones to engage with the adversary aircraft in a dogfight or deploy autonomous patrolling vehicles at the borders for getting the job done? How much autonomy should be given to the machines on the battlefield? A clear vision of the AI programme is necessary for a middle-income country like India that cannot afford to invest heavily in this sector. It is a matter of guns vs. bread and butter — there is a trade-off between spending in national security and public welfare. Unlike technological programmes in developed countries which can afford to fail first and fail faster, India does not have this luxury due to paucity of resources.
Secondly, the lack of critical infrastructure is one of the biggest impediments in the prospects of AI in India for both civilian and military uses. As AI runs complex algorithms on loads of data, it is essential to have robust hardware and enabling data banks within the country. If a critical AI-based military technology harnesses the data on a remote server located beyond the borders, it can potentially hinder the goal of Indian foreign policy from preserving its strategic autonomy in a way that it might be compromised.
Thirdly, the role of the private sector will be pivotal in making the AI accessible and efficient. As AI demands high-skills and capital, innovations need an ecosystem supporting the free flow of both money and skill. However, as per the existing FDI policy in defence, only 49% of the foreign investment is permitted within the automatic route above which it needs government approval. India has traditionally been conservative in handing over the reins of the defence industry in the hands of the private sector.
Leading powers like the US, China and the EU (and France) have their vision documents for research and development programmes in artificial intelligence. To begin with, India should envisage a clear strategic vision regarding the AI on similar lines. Despite resource limitations, India is home to world class academicians in computer science and engineering spread across the IITs, IISc, NITs and IISERs. An academia-industry-policy synergy is of utmost importance to realise the strategic, societal and cultural implications of AI in defence. It will help us to find answers to questions raised in the previous section.
The government should create a supportive ecosystem in which the AI industry in India can thrive. There is a dire need to invest in critical infrastructure so that the data servers lie within the territory. Apart from ensuring strategic independence, it will also address data privacy concerns.
The AI-market for civilian purposes in the country is on the rise. For instance, India ranks third in G20 countries in AI-based startups. Last month, PM Modi unveiled his “5-I” vision to maximise the societal benefit at the G20 summit on digital economy and artificial intelligence. India should realise the dual-use nature of artificial intelligence and thus also open its market to AI investments in defence. The 49% cap on FDI policy in defence should be revisited to account for this fact. Policymakers should brainstorm on the prospects of marrying the flagships initiatives of the current government — Make in India in Defence and Digital India — to bring a technological revolution in the defence industry.
New Delhi, whose major security priorities lie in the subcontinent, cannot ignore the progress made by China in the field of artificial intelligence. The State Council AI plan, released in June 2017, outlines China’s ambitious AI policy which aims to create an AI-industry of worth 150 billion RMB that is ten-fold its 2017 figures. At the same time, India should avoid setting their benchmarks taking Chinese investments in mind. Since India is a late entrant in the field, it could capitalise on the late-movers advantage, i.e., mimicking the existing narrow-AI technologies, to fulfil its basic security needs (like border patrols and intel-gathering) rather than seeking innovation. The ‘AI in defence’ vision should be centered on achieving military modernisation over balance of power considerations. Hence, this is the time for New Delhi to step up in the arena of what hardliner security experts refer to as an AI arms race.
*Ambuj Sahu is research intern in the Strategic Studies Programme at ORF New Delhi.