The Challenges of Artificial Intelligence in Military Intelligence

At present, most processes of military intelligence agencies in the world have been digitized, and the number of data sources continues to increase. Analyzing and understanding all the data received will help achieve information superiority, which is the basis of artificial intelligence.

It is not possible for us to manually process all data. Artificial Intelligence(AI) can mimic many human cognitive skills and can do so infinitely faster and more consistently. While AI can predict trajectories, identify patterns and apply theory to provide threat analysis and predictions, this is an iterative process as AI learns from experience with reinforcement learning techniques used to improve models. So, data availability can be a challenge for training AI models, which requires large amounts of defense-specific data, and collection and classification constraints can make this difficult.

Furthermore, due to the explanatory power that humans possess, it is highly desirable for AI intelligence systems to keep humans involved. This could allow AI to make decisions when their consequences are well understood, especially when uncertainty is low. As in the case where dynamic activities are involved, the need for human involvement is even greater. This is a two-way test for human intelligence agents and artificial intelligence systems. Artificial intelligence systems need more reasonable and reliable analysis and decision-making capabilities, which rely on commanders who understand relevant knowledge and strategies to plan, supervise and regulate the intelligence analysis process. A military intelligence analysis system with a high level of intelligence requires AI to be properly used by commanders who can understand complexity, solve problems, and ask the right questions for the situation. In general, future intelligence personnel and artificial intelligence military intelligence systems will constitute a new human-machine relationship.