FROM ALGORITHMS TO BATTLEFIELDS
By Ritika
Integrating AI technologies into the military involves a web of economic, cultural, and social challenges.
Historically, new technologies have consistently reshaped the landscape of warfare– from the rifle to the thermo-nuclear device. Integrating AI and robotics represents the latest chapter in this evolution, poised to redefine decision-making and the execution of force, fundamentally altering military strategy.
Contemporary AI systems operate within narrow machine-learning algorithms, but advancements in neuroscience, quantum computing, and biotechnology are pushing the boundaries of what is considered "intelligent machines." Cognitive processes such as knowledge acquisition, critical thinking, and decision-making, significantly impacted by advanced computing, will exert a greater influence on warfare than technologies enhancing command and control capabilities.Due to its cross-cutting nature, AI will impact both traditional military capabilities and the realm of hybrid threats. It is projected to influence all core areas –collective defence, crisis management, and cooperative security systems.
Considering the opportunities and the risks involved, this foundational technology is too complex for any single actor, like the Armed Forces, to manage alone. Consequently, cooperation is inherently needed to capitalize on AI's potential to transform enterprise functions, mission support, and operations.
From Tactics to Strategies
It is obvious to military analysts that all three levels of warcraft- tactical, operational and strategic– will be affected to varying degrees by this disruptive technology. This necessitates fundamentally restructuring military tactics, operations, strategy, logistics, and organisations. At the tactical level, superiority is derived from the continual evolution of tactics around new weapon systems and technology.
First and foremost is achieving informational dominance- the ability to generate, manipulate and exploit information sufficient to afford its possessors military dominance over its enemy. This transcends all levels of warfare. Historically, information has been a crucial battle-winning factor. In today's networked battlefield, with blurred frontlines, "last-mile warfare,” is where battles would be won with split-second decisions by field commanders at both tactical and operational levels.
At the tactical level, obtaining real-time insights, predicting enemy movements, and understanding immediate surroundings are essential for mission success and risk mitigation. Traditional intelligence methods, such as human patrols and drones, are no longer sufficient for modern conflicts' fast-paced, information-centric nature. Technological advancements are revolutionizing battlefield information gathering. Soldiers now utilize highly portable, advanced tools that provide enhanced situational awareness. With real-time data on enemy positions, potential threats, and terrain features delivered to pads, soldiers are empowered to make informed decisions. A notable development in this area is using nano drones, miniaturized UAVs designed for tactical reconnaissance and intelligence gathering in close-quarter combat. Nano drones offer offensive, defensive, and reconnaissance advantages, allowing soldiers to outmanoeuvre the enemy in last-mile warfare.
At the operational level of warfare, the depth of the battlespace has expanded exponentially due to the deep strike capabilities of warring sides armed by advanced computer processing, microelectronics, 24 x 7 surveillance, and precision-guided munitions (PGMs). To transmit target data between 'sensors and shooters,' robust communication networks are critical.
New information technologies support a network-centric approach to warfare, emphasizing acquiring and exploiting superior information. This has significant implications for synergy among platforms and the reorganization of command-and-control structures, including C4ISR tasks. The digitization of information gathering and dissemination presents significant technological challenges and promises radical changes to managing real-time "dominant battlespace knowledge," enhancing battlefield performance. These advances can render traditional trade-offs obsolete, widening the gap between militaries that effectively integrate new technologies and those that lag.
AI systems are increasingly capable of processing vast amounts of data, recognizing patterns, and making complex decisions. Initially, AI focused on general search strategies, but it soon became evident that integrating extensive domain knowledge was necessary for achieving higher performance. This shift led to the development of highly specialized expert systems, marking a transition from a search-based paradigm to a knowledge-based one. Such a move would greatly enhance the ability of commanders at the strategic and national level to hammer together an effective battle winning war strategy.
The Paradox of Transformation
Military transformation is inherently paradoxical, involving the reconciliation of the drive for dynamic shifts towards new methods of warfare within organizations inclined towards continuity. Given the complexity and unpredictability of warfare (referred to as the fog and friction of war in the military lexicon), militaries are structured to promote standardization and predictability. Traditionally, militaries prefer the status quo ante, avoiding uncertainty and risks associated with uncoordinated/uncalled-for actions. It is, therefore, not surprising that traditional military hierarchy, especially in large conventional armies, remains hidebound and ultra-cautious (some would say suspicious) of change. This creates a dilemma for defence planners, who must be prepared for resistance and confrontation to transformational changes from within the military’s ranks.
The military is increasingly being pushed into other domains it has traditionally been uncomfortable dealing with. Integrating AI technologies involves a web of economic, cultural, and social challenges.
The pressure on the military to change comes from different avenues. Most acute is the one emanating from policymakers being fed a constant diet of ‘policy initiatives’ generated by academics, system innovators and think tanks. Internally, militaries have their own 'new age strategists' who, with their 'new-fangled’ ideas, make military hierarchies uncomfortable, if not outright hostile. Externally, the progress reported by known and perceived adversaries creates even more excruciating pressure points. Clearly, the enemy, too, has a vote in this account.
Transformation, especially driven by cutting-edge technology, does not come cheap. Diverting precious resources, even for prosperous nations, is a matter of intense public scrutiny. Countries with deep pockets can only afford to venture into this: the U.S. has heavily invested in AI, funding projects like a robotic submarine system and allocating billions to AI, Big Data, and cloud technologies. Similarly, its emerging rival, China, aims to become a global leader in AI by 2030, with a significant footprint in the defence sector.
Is it Cost-Effective?
AI offers unparalleled efficiency and numerous advantages in the military sector. Originating from wartime research in cryptography, ballistic calculations, and cruise missile development between the tumultuous period 1940-45, AI is now enhancing security, military tactics, and defence strategies. AI is set to impact three key areas of national security: military superiority, information superiority, and economic superiority.In the military domain, AI integration with strategic weaponry, biotech, and computer systems will immensely add to a nation’s comprehensive national power.
At the national level, emerging technologies like machine learning and network technologies will give nations a decisive edge in intelligence gathering, data analysis, and cyber defence. AI in cybersecurity automates tasks for defensive and offensive objectives, becoming crucial in national security infrastructure. As AI technologies become more accessible, smaller nations invest in these advancements to maintain a competitive edge.
Measuring the effectiveness of AI technologies in a military context is challenging due to cognitive biases and the difficulty of isolating the impact of a single technology in a complex system. Expert human judgment plays a significant role, especially in uncertain situations. Testing AI generative imagery platforms and language models reveals challenges, particularly in rendering text accurately and creating original content. Despite the potential applications of AI systems, current capabilities remain in their early stages of development.
Risk Mitigation
Are we venturing blindly into a space of which we know very little- rapid convergence of physical, digital, and biological technologies marks the early stages of a profound technological revolution. Effectively governing these emerging and disruptive technologies on a global scale will be essential to mitigate the risk of future conflicts.
A word of caution would perhaps be in order. Establishing parameters for developing military AI and risk-reduction measures associated with military AI is crucial. It may be more manageable to regulate AI before these capabilities become fully integrated into military systems globally."The potential consequences of AI malfunctioning are significant, making it essential for us to adopt a proactive approach rather than a reactive one," said Elon Musk. It is, therefore, crucial to address the key uncertainties, threats, and obstacles posed by AI in today's military applications and practices to ensure its safe and effective use. Some threat areas are outlined in succeeding paragraphs.
Adversarial Attacks. As AI and ML technologies become more widespread, adversarial evasion attacks have emerged, highlighting the need to study these algorithms' reliability, privacy, and security. In imaging systems, imperceptible noise can induce classification errors. Noteworthy attacks include "white-box" attacks, where the enemy knows the algorithm, and "black-box" attacks, where the adversary only knows the input and output. The rise of these adversarial attacks underscores the importance of ensuring AI and ML system reliability, privacy, and security, especially in mission-critical applications like military operations. Developing robust defence mechanisms and comprehensive evaluation frameworks is crucial to mitigate these evolving threats.
Limited and Uncertain Information. During a battle, incomplete or unreliable information can hinder decision-making. To address this, strategies must enable the discretization of time and action, the creation of temporary windows of advantage, and the seizing of the initiative to achieve strategic objectives. Command staff relying solely on routines or established protocols will be at a tactical disadvantage and must be prepared to adapt and innovate as necessary. Additionally, transferring historical knowledge from military experts, presented as facts and rules, is crucial for developing effective machine-learning algorithms in tactical environments. These systems must thoroughly understand the military domain, including weapon performance, warfare models, decision models, rules of armed conflict, and operational protocols between combatants or allies.
Cybersecurity. Malicious exploitation of AI can inflict significant harm on military infrastructure, personnel, and operations by automating and amplifying cyber assaults, making them more sophisticated and harder to detect and thwart. AI-driven cyberattacks can emulate normal network activity or exploit vulnerabilities within AI systems, evading traditional cybersecurity defences and allowing attackers to access military systems, steal sensitive information, or disrupt critical functions.To counter AI-driven cyber threats, military entities must enforce stringent access controls, employ encryption and other security protocols, and conduct regular security audits and vulnerability assessments. Additionally, developing AI systems resilient to cyber threats is essential, involving creating AI models resistant to deception and adversarial attacks and deploying mechanisms to detect and mitigate disruptions.
Data Scarcity & Lack of Clear Values.The quality and quantity of data available significantly influence the performance of an ML algorithm. Despite the U.S. Army's extensive combat experience, it has faced challenges in applying ML effectively in real confrontations due to limited sample data. Unknown data about adversaries and the difficulty of real-time data collection during combat add to these challenges. Due to security and cost constraints, existing databases, often derived from exercises, differ significantly from actual battle conditions. One solution to bridge this data gap is conducting extensive fieldwork to capture and label real combat values according to ML algorithm requirements. Another innovative approach is to use real-time strategy games, allowing commanders to simulate various roles and scenarios, thereby generating valuable data and experience for ML applications.
Poor Evaluation Criteria. Developing ML algorithms aims to create systems that use accumulated experience and new scenarios to inform decision-making. The main challenge is ensuring the algorithm's validity, reliability, and applicability to different scenarios. Achieving this requires numerous experiments and simulations to test and refine its effectiveness.
Key Takeaways
The future of AI in military operations holds significant potential, but it demands a thoughtful and responsible approach. By acknowledging and addressing the challenges and risks associated with AI, policymakers can foster its ethical and effective use, maximizing its benefits while mitigating potential threats.
Integrating AI in the military should focus on augmenting human capabilities rather than replacing them, ensuring that technological advancements enhance global security and stability rather than undermining it.
By proactively addressing the challenges and risks associated with AI while leveraging its capabilities responsibly, the military can enhance its overall effectiveness and adaptability within the rapidly evolving technological landscape. This will necessitate the development of robust legal and ethical frameworks and ongoing dialogue and cooperation among policymakers, military leaders, technology experts, and the broader public. Only through this multifaceted approach can the military harness the transformative potential of AI in a manner that strengthens global security and stability.
