Deterrence vs Resilience & AI

The deterrence theory was stressed during the Cold War, and its aim was to acknowledge the existential threat posed by nuclear weapons. The primary foundation of this theory is credibility and communication to accept that rival states have a firm belief in one another’s determination and ability to respond if attacked. Although nuclear war is prevented by Mutually Assured Destruction (MAD, proxy wars are intensified due to a precarious balance among the states. AI has minimized the feasibility and viability of conventional deterrence theories. Arms control agreements aim is limited to nuclear arsenals, in contrast, AI has no limit pertaining to quantifiable, covered, or observation, as it is working solely on algorithms and available data. Resilient Artificial Intelligence systems ensure that AI systems are protected to such an advanced degree that they have the potential to resist cyberattacks, and quickly bounce back fast while maintaining national security, economy, and survival. It doesn’t mean that an AI system cannot be attacked, but it means the capacity to recover from attacks, hacking, and data theft very more quickly without a data breach. In the nuclear area, international security relied on deterrence. It is grounded on the idea of using a credible threat of retaliation to prevent the enemy from taking hostile action. Now, after the advent of AI in the field of political, economic, and security domains, various states face significant difficulties in deterrence. What went wrong? The feasibility and capability of conventional deterrence strategies are replaced by AI, which brings autonomy, speed, accuracy, opacity, and unpredictability to decision-making. Paradigm and pattern shift in deterrence is evaluated from human to machine learning (ML), threats from states to hybrid, and the looming threat is now assessed by algorithms.AI is a double-edged sword. On the one hand, AI can become very useful in transforming industrialization, agriculture, education, innovation, and, more importantly, the healthcare sector while on the other hand, AI has the leverage to harm state activities in different forms, such as disinformation, manipulation, data theft, and cybersecurity vulnerabilities. The improvement and mishap hinge on AI’s resilience to address endangering situations, protect millions of citizens, and prepare autonomous systems to respond to existing threats in no time. AI technologies lessen decision-making rounds by operating at machine speed. Autonomous systems may yield action during emergencies earlier than humans can. This increases the likelihood of unintentional escalation, destabilizing the controlled pace that deterrence has relied upon until then. Threats that are apparent, credible, and signaled are vital to deterrence. However, AI systems are not transparent, and visible in nature. In Conventional deterrence, threats are signaled by movement of traditional troops, ammunition, or nuclear arms tests. But in AI systems, mostly work on black boxes, and the main foundation is based on algorithms. Now the overwhelming threats are that there is no advancement of universal laws to deter the threat of AI systems, and the farcical danger is that Do Algorithms can differentiate and not get obfuscate between strategic menace (threat) and peaceful innovation? It is well known fact that AI systems have brought a paradigm change in nuclear deterrence. AI systems enhance missile guidance and surveillance, thereby compromising second-strike capabilities. The conviction in nuclear command and control could be diluted by hacking, manipulation, or bluffing of AI systems. Political stability poses distinctive threat due to Disinformation tactics powered by AI. Here, deterrence is supplementary to resilience, generating social barriers in contradiction of manipulation rather than military retaliation. The main question is: by what means will the states prompt the AI to play its role in the security of the state? Deterrence is passive to perception and psychology. To further emphasize this, AI, machines, and algorithms are unable to differentiate between threat, deterrent, or peace. If AI is continuously fueled by numerous threats, its response will be ambiguous as these AI systems are impalpable, ubiquitous and fast developing. There is no endpoint of development and improvement in AI systems, and algorithms and machines will decide the fate of the country. Ultimately, the AI arms race is increased among the states due to the unavailability and constraints of open treaties and inspections, like in traditional or nuclear deterrence. AI and machines have no layer that can distinguish real war from mere threats. To ensure stability among states, it is the universal obligation of all states to pledge for the inclusion of people (human in/on the loop) in critical decision-making processes, predominantly in nuclear command and control.Deterrence hinge on resilience in the AI era. Adversaries are less likely to attack when societies, militaries, and infrastructure are exposed as resilient to AI-enabled attacks. To deal with AI-driven threats, perceptions, and disinformation, it is the primary responsibility of every state to communicate sensibly, make AI systems more resilient, and there must be human assessments both at the time of peace and threat, if any. The writer is M.Phil Scholar.

Author , Mujeeb ur Rehman