AI capabilities in military situations increase the risk of inadvertent escalation and instability.
James Johnson, 2020 (Postdoctoral Research Fellow at the James Martin Center for Nonprolifera¬tion Studies, WASHINGTON QUARTERLY, “Artificial Intelligence in Nuclear Warfare: A Perfect Storm of Instability?”)
A broader point here is that under crisis and conflict conditions, the deterrent effect of AI is predicated on the perceived risks associated with a particular capa¬bility it enables or enhances. The higher the uncertainty generated by a capacity, deploying AI-augmented capabilities in a crisis might actually encourage an adversary to act more cautiously, and in turn, bolster stability. Thus, the uncertainty caused by the introduction of AI into a situation might incentivize states (especially those facing a superior adversary) to delegate decisions to machines for the per¬ceived deterrence effect.26 Because of the diffi¬culty of demonstrating a posture like this before a crisis or conflict, however, this implicit threat could equally worsen crisis instability. AI systems that are pro¬grammed to aggressively pursue tactical and operational advantages, for example, might misperceive (or ignore) an adversary’s bid to deescalate a situation as prelude to an imminent attack. These dynamics would increase the risks of inad¬vertent escalation and first-strike instability.27