Artificial Intelligence without Cyber Resilience in South Asia

Artificial intelligence is becoming one of the defining technologies of 21st century. Today AI is being deployed in health care systems, financial trading, translation and transportation and military technology massively. The technology and terminology “artificial intelligence” are not the product of 21st century, rather the term was coined in 1956 at Dartmouth summer workshop organized to develop thinking machines. However, there is no single definition of this technology, thus is quite difficult to define. According to a definition proposed by European Commission, artificial intelligence (AI) is a system with capability to achieve given goal by acting physically or digitally, after perceiving their environment by interpreting the structured or unstructured data, reasoning the knowledge derived from this data and deciding the best actions to perform the given goal. This definition identifies the capability to perceive, interpret and reason as the pre-requisites for AI enabled systems. Due to its capability of intelligence monitoring, reconnaissance, target recognition, communication and navigation, automated command and control, precision strikes, AI enable systems are becoming necessary for militaries. Artificial intelligence is not a stand alone technology, rather enhances or adds new features when integrated into military systems. AI’s integration into military systems is a double edge sword. On the one hand it is improving the existing systems by providing precision, intelligence, detection and decision making tools, on the other hand it is increasing risks and vulnerabilities for the existing structures and systems.

Today policy makers are discussing issues like whether to trust machines, as machine speed is enhanced manifold due to AI, which in asymmetric contested environment would make it difficult for the commander to contain, control and terminate the event because of the enhanced speed of the contest. Moreover, under the pressure of limited time and speed, state might accept a higher risk and escalate under the imperative of use it or lose it situation. Moreover, another issue with utilization of AI in military systems is the biasness of the algorithms encoded into them by human engineers, which could trap human operators into machines’ biased and flawed assumptions. The integration of AI into anti-submarine warfare would reduce the survivability of SSBNs, which would be damaging for the nuclear deterrence and strategic stability between nuclear rivals. By improved ability to collect and recognize different submarine signatures, AI would play an effective role in anti-submarine warfare.

Most of the research related to AI in military systems is associated with their contribution as catalyst in offensive or defensive operations. However, one less discussed fact in this regard is the security of the artificial intelligence based systems. Today artificial intelligence is used in missile defence and reconnaissance, which enhances the target recognition, image and pattern recognition and trajectory calculation. Moreover, it can also assist in analysis of damage effects. Enhanced intelligence and trajectory calculations will assist states in more guided and precision strikes. All these technological developments reveal that artificial intelligence is enabling states to gather massive information and processing it to achieve desired objectives. However, artificial intelligence does not exist in vacuum. It is essential to recognize that all the data amassed, processed and utilized with the help of artificial intelligence needs protection and security. Moreover, besides data, machines and their algorithms also need to work effectively to avoid manipulation and breach.  This brings us to the point that though artificial intelligence is a necessity for national security of states and is largely adopted worldwide, this technology also needs to be secured against cyber-attacks. To protect such systems it is necessary that states build resilience against them. It is necessary that AI systems should be protected from cyber-attacks and whole infrastructure of state must have cyber resilience.

With increased dependence on information technology and rapid digitization of systems, term cyber security gained momentum. However, these systems not only need to be securitized but they should be resilient against the threats. Cyber resilience is the ability of the system to operate during an attack and achieve a minimum level of operationalization while responding to an attack. It also enables system to develop a back-up system that works in case of attack. Cyber resilience is a step forward from cyber security because it not only ensures the security of system, but also identifies the threats to it and then proposes the system that could work amidst such attacks. Most military systems are resilient against kinetic attacks because resilience and survivability go hand in hand. But, with modernizations in military it is necessary that state’s cyber networks which are working on artificial intelligence must be resilient against kinetic and non-kinetic attack.

Today states are in race to use the AI in their military systems to achieve maximum military gains and denying their adversary the same. Situation is not so different in South Asia where two nuclear rivals of the region are paving the way towards use of artificial intelligence for military purposes. India has developed Center for Artificial Intelligence and Robotics (CAIR) in DRDO, with the aim to develop AI within the military systems to improve geographical information system technology, decision support systems and object detection and mapping. Moreover, companies like Bharat Electronics Limited (BEL) are already in the process of developing and incorporating AI into military equipment. This includes AI-enabled patrol robot developed by BEL built in the hope to be utilized by the Indian military. Moreover, in 2019 India’s Gen. Bipin Rawat said adversary in north is spending huge amount on AI and cyber warfare, so we cannot be left behind in this race. It is mostly projected by the Indian policy makers and many international scholars that India is facing adversaries at two fronts (China-Pakistan), to justify India’s military expenditure and modernization. However, recently, events like Galwan Valley clash evidently exposed that India’s military capabilities are mostly against Pakistan. Moreover, South Asia’s security dynamics are heavily characterized by the action-reaction chain. To avoid security dilemma vis-à-vis India, Pakistan would also invest in AI. At the moment Pakistan has also started working towards achieving expertise in AI. In 2019 President of Pakistan launched PIAIC with focus on development of skills in AI to strengthen economy and defence systems. Moreover, there are centers like National Center of Artificial Intelligence and Department of Robotics and Intelligent Machine Learning  in NUST, which are working to improve AI based knowledge in Pakistan. Besides that Pakistan recently launched a program named “Digital Pakistan” to increase access and connectivity, digital infrastructure, e-government, digital skilling and training  and introduce innovation and entrepreneurship.

There are many studies done on the implications of AI on nuclear deterrence and strategic stability in South Asia. These studies highlight that due to prevalent asymmetry in conventional military build-up, introduction of AI into military technology would worsen the already fragile deterrence stability of the region. This assumption is based on the argument that due to AI in reconnaissance systems, high-level intelligence collection would affect the survivability of nuclear weapons, which is based on diversification and concealment. However, AI would also enable both states to have more response options in a short time with the help of decision making tools in case of crisis, especially in aerial battle.

Moreover, both states are moving towards the massive digitalization of their military systems and society without building cyber resilient systems. Resilience can be built against the vulnerabilities like human factor, massive speed of the systems, protection and storage of data and advanced persistent threats (ATPs). Artificial intelligence based systems must be incorporated in societies and militaries along with mechanisms to strengthen the cyber security systems. A front runner in AI like US has also expressed concerns over the need for modern equipment to operate on “internet-like networks” and subsequent increased vulnerabilities due to their applicability. Therefore, military modernization can happen effectively through cyber resiliency in military systems, network processes and cyber architecture. Cyber resilient system would enable state to develop a system that would remain functional during a phishing attack. Steps like cyber deception, agility and clone defense could increase resilience in the existing systems. This is important to understand in already lacking strategic stability, military systems based on artificial intelligence would be an ideal target of AI advanced persistent threats in South Asia.

Therefore, as the process of digitalization is increasing in Pakistan-India equation, it is also becoming very important that both states should develop resilience in their cyber systems so that the technologies could give them advantage rather than becoming a security peril for them.

Ahyousha Khan
Ahyousha Khan
Research Associate at Islamabad based think-tank Strategic Vision Institute