The end of the beginning of widespread Artificial Intelligence (AI) adoption has arrived. Instead of seeming like a futuristic scene from a movie, today’s advancements are making a marked difference in processing large data sets.
These advances in capability and reliability demonstrate that AI should have a role in any unified threat intelligence strategy for the U.S. military, because of its incredible potential to accelerate the production of actionable intelligence and decision-advantage workflows.
Unified threat intelligence describes a comprehensive approach to collecting, analyzing and sharing threat information across an organization. While the U.S. military has used the concept for years describing efforts to gather intelligence from multiple sources, introducing AI tools drives operational efficiency and accelerates the response to threats.
The concept includes two tenets: ensuring the joint team has a singular and comprehensive view of the threat landscape and that teams can respond by combining data from cyber, geopolitical and physical intelligence sources into a single, actionable intelligence stream.
Even with the improvements afforded by AI, the two fundamental challenges still exist: an overabundance of data to process and false positive/false negative errors. AI continues to improve and attempts to keep pace with vast amounts of data collected, but the need for humans in the loop has not decreased. The military still requires humans to ensure AI performs well enough to “minimax” type one errors (false positives) and to try to eliminate type two errors (false negatives) altogether.
Military embraces AI platforms
As early adopters of proven technology, the U.S. military has embraced AI platforms for analyzing imagery, sensing and tracking and large language models that can search vast data sets and legacy documentation. These AI tools do hundreds of hours of precise work in seconds.
The public sector has identified or is developing bespoke models for both broad and niche use cases. An exciting addition to the process is the development of unconstrained virtual workspaces. This is where these models can be applied, inviting stakeholders to share and fuse their information and data streams while providing practical boundaries and high security. The virtual workspaces create a highly informed mission command environment to support multi-domain operations (MDO).
AI heavily benefits the initial stage of unified threat intelligence strategy detections by improving insights derived from large, disparate data sets. It can accelerate supervised learning and completely transform unsupervised learning—doing what humans cannot—to identify correlations across dissimilar data sets.
For example, AI can correlate textual insights, with insights derived from imagery, with measurement and signature intelligence. It can fuse data on the physical movement of troops and assets, geopolitical climate and cyber activity to determine actors, actions, and threat objectives.
Improving analysis and decision-making
A unified threat intelligence strategy allows the military services into a collaborative environment to share intelligence, and the assessment phase integrates insights into decision workflows. While sorting through inputs and intelligence can quickly overwhelm human analysts, AI works tirelessly.
The fusion of data and intelligence from different networks and sources requires governance. It’s vital that stakeholders are assigned permission to contribute within their competencies and roles to create early warnings about threats. It’s also important to have controls and maintain governance so AI tools and collaborators stay within their boundaries.
A good example of how a well-implemented unified threat intelligence strategy might work in a real-world situation is in helping allies determine who was responsible for the recent sabotage of communications cables in the Baltic Sea and their intent.
Powered by AI, the military could include data sets for the movement and speed of ships in the area before, during and after the incident; weather feeds, water depth, currents and ship information; ship equipment manifests, registration, recent port calls and the national origin of crewmembers.
Adding geopolitical data might illuminate motivation. Implementing AI tools across these disparate data sets can produce insights that guide further analysis or even point to the cause of the cut cable, accelerating decision cycles. The same AI capabilities could be used to map and surveil underwater environments and channels.
The military can use AI tools to develop options, helping teams evaluate responses and forecast resources required for support. AI’s speed is especially helpful in eliminating analysis paralysis and in making efficient, confident decisions.
Because this phase demands human oversight to exercise quality control over AI inference, the team’s collaboration environment is critical in detecting and minimizing both false positives and false negatives. Data-driven decision-making is the goal but so is ensuring AI tools don’t solve a large problem (too much data) by creating an even larger problem (too many false positives and false negatives).
AI tools reduce military reliance on analysts to find correlations across data relating to threats in the physical, cyber and geopolitical domains and reduce errors in detecting threats identified in cross-domain intelligence fusion. This helps to speed up workflows, intelligence production and decisions.
Although AI is still progressing through adolescence, it is mature enough today within a unified threat intelligence strategy to address challenging military scenarios and accelerate the identification and response to threats.
