Gaps in Deterrence Theory And Artificial Intelligence Solutions
Daniel Sazhin, Tyler J. Gandee, Christopher Stevens, Lorraine Borghetti · 2025
With recent developments in Artificial Intelligence (AI) and the importance of deterrence strategies for maintaining peace, it remains unknown how AI-enabled capabilities could impact deterrence decision making. Additionally, there are noteworthy gaps in deterrence theory that validates mechanisms of deterrence decision making, models how beliefs change over time, and predicts how psychological preferences will influence decisions to escalate or respond with conflict. Addressing these gaps could adjudicate the credibility debate between classical and perfect deterrence theories. We hypothesize that experimental investigations into how people develop beliefs about belligerence and how these beliefs change over time could address 1) Gaps in deterrence theory, 2) Identify empirical approaches to test how people behave in deterrence situations, and 3) Could help leverage AI and human-machine co-learning to understand deterrence decisions. Specifically, Large Language Models could contribute to developing better awareness of a decision context and interact with both humans and reinforcement learning derived agents to assist with developing models of adversarial intent. We argue that such AI tools could eventually serve as decision aids in wargaming and operational situations. The implementation of AI-enabled deterrence theory could allow leaders to achieve decision superiority, effective signaling, and achieve the avoidance of unnecessary escalations.