The War Before the War: Predictive Intelligence and Human Control in Warfare

Abstract

Artificial intelligence (AI) is increasingly being integrated into military intelligence, targeting, and decision-support systems in part because it can process information faster than human decision-makers. Yet, there is a less examined problem regarding this integration, that being the fact that humans may retain formal authority over military decisions while losing the time necessary to exercise that authority meaningfully. This paper argues that deliberative time, defined as the interval necessary to independently evaluate and when necessary intervene in an algorithmic recommendation, is a necessary condition of human control. Existing approaches tend to locate human control in authority, system design, explainability, or institutional regulation. These safeguards remain important, however each depends on humans having sufficient time to use it. This problem becomes particularly consequential in predictive warfare because probabilistic assessments of an adversary’s future behavior can encourage military action before uncertainty has been resolved. Such actions can themselves become new military signals, creating feedback loops in which algorithmic predictions contribute to producing the behavior they were intended only to anticipate. This paper draws on modern and historical conflicts to examine how accelerated prediction can compress interpretation and response. It argues that speed should therefore not be treated as an independent military advantage. When AI responds to an observable physical threat, reducing latency may protect security. When it interprets uncertain political intentions, the same compression can increase the risk of escalation. Maintaining human control in this reality and these technologies ultimately depends not only on whether a human remains in the decision process, but whether there remains enough time for that decision to actually be human.