AI key for future NATO deployment and training-visualization, analyzation, integration, equipment protection
Husain et al. 18(Amir Husain, August Cole, and Wendy R Anderson. CEO of Spark Cognition & Board of Adivisors for UT Austin’s Department of Computer Science & Member of Council on Foreign Affairs, Senior fellow of Scowcroft Center on Strategy and Security at the Atlantic Council & Fellow of Brute Krulak Center for Innovation and Creativity at Marine Corps University. " As Budget Polemic Drives Headlines, Do Not Lose Track of NATO’s Approach to AI". 07-27-2018. Royal United Services Institute. https://www.europarl.europa.eu/cmsdata/155282/WendyRAnderson_RUSIArticle.pdf. 6-22-2022.)-cg
Indications and Warnings of Crisis Readiness at an Alliance level depends on the synchronized sharing of high-quality data to inform intelligence assessments and general situational awareness. One way to accomplish this is to establish AI-focused all-source data processing that is centralized by NATO headquarters but produced by, and available to, member states. An electronically shared view of the operational environment, frequently referred to as a common operating picture, can be developed in seconds, not days, to aid overloaded human analysts. Data sources must be expanded beyond conventional defense-related sources, to include open-source and commercially available imagery, metadata, and social media. Existing machine-learning systems already make this possible, but their use is nascent. The current generation of analytical software tools are a step in the right direction, but plenty of unprocessed data is not transformed into national and Alliance-level actionable intelligence reports due to a shortage of human analysts. This causes gaps in the intelligence picture, which can be exploited by an adversary. Taking full advantage of AI can smooth out the escalation of a crisis by combining the analytic efforts of NATO command organizations and member states to ascertain how, and when, a scenario will develop. Machine-learning crisis simulation systems offer improved visibility into the causes and drivers of a crisis that might otherwise be overlooked by conventional analysis, which can be too narrow to capture the true complexity of a situation. Integrating AI into coalition operations is a global challenge, not only for NATO, but also its supporting allies, many who, in the hyper war era, may not be organized militaries. In a hybrid, multi-domain context, a real-time common operating picture – that draws upon simulations but is also predictive – must extend into the information domain, including modelling peacetime public narratives. The volume and velocity of information during the early phases of a crisis will be nearly overwhelming. Anything other than an AI system has little hope of success in keeping up, while paring away irrelevant information to ensure that human decision-makers are tracking the right information. As with the assessment phase, AI-driven simulations and scenario planning offer substantial insights. With those insights, such systems can create a comprehensive picture of force readiness and logistical imperatives to inform NATO’s response options. AI-driven analysis can process massive amounts of unstructured data, maintenance logs, reports, and logistical information to create a detailed and accurate picture of force readiness. Machine-learning systems can also fold-in data from civilian, commercial-sector and non-governmental sources in order to produce a more accurate ‘whole of society’ capabilities view than what is currently possible. Employing this approach across the 29 NATO states is especially critical in the context of high-intensity conflict scenarios that, although unlikely, would be potentially catastrophic. Additionally, being able to include civilian resources and infrastructure as part of this response option analysis is something that AI systems are better suited to do than conventional database analysis. In addition to readiness driving response options, another major consideration is developing plans and evaluating their effectiveness. In more tactical scenarios, such as in the development of mission optimized auto-pilot capabilities for fighter aircraft, AI techniques such as reinforcement learning are already proven in creating optimized operational plans for a single platform in a similar way that can be applied for entire combat elements. The operational execution of a plan by NATO forces is currently dependent on mere assumptions of the consistency of training, resources, and the joint force’s ability to accomplish their mission. AI, particularly when combined with virtual- and augmented-reality visualization, can play a significant role in providing advanced training and pre-deployment unit-level preparation for NATO-led forces during peacetime, to ensure a rapid yet smooth transition into conducting operations. During those operations, autonomous software can be used to assist with maintenance, logistics management, and targeting of offensive and defensive systems. It can also ensure that a NATO force is successfully integrated with autonomous, unmanned ground, air and sea vehicles to provide a standardized, and ever increasing, level of operational competence and consistency of execution. During operations, machine-learning systems can use sensor data, entire technical libraries, and advanced models to accurately predict and then prevent equipment failure; given the danger now posed by improvised explosive devices and precision munitions to supply lines, such efficiency has profound strategic importance. In more tactical scenarios, such as in the development of mission optimized auto-pilot capabilities for fighter aircraft, AI techniques such as reinforcement learning, which allows machines to share their experiences and optimal solutions among themselves, have shown their utility. Adversarial AI systems running within a simulator can assist in the evolution of a highly optimized, robust mission intelligence that is effective at fulfilling defined objectives. In a similar vein, adaptations of these tactical AI-driven simulator frameworks can be used to gauge likely public reactions to proposed response options. What all of these approaches permit is to potentially identify – and address – operational pitfalls before they actually occur. Given the importance of public opinion and political support for NATO, what happens after a crisis is as important as the operations undertaken during it. Disengagement is likely to be an ongoing real-time competition over information and digital narratives, as contested as an operation’s military elements. The AI tools will be familiar, not unlike the machine-learning algorithms that today automatically run behind the-scenes markets placing political ads or marketing campaigns onto social media feeds.