1. TURN: AI tech increases the risk of accidents and false alarms—leading to nuclear conflict.
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?”)
AI technology fused with advanced weapons (potentially simultaneously across multiple combat zones) could allow these systems to react at machine speed and boost the overall pace of combat. Despite the tactical advantages of being able to react in real time, especially in asymmetric contested environments, experts warn that massive increases in the speed of combat could result in machines react¬ing to combat situations at a pace that surpasses human comprehension—so much so that commanders might be unable to control, contain, or terminate events.5 Because human commanders would be unable to react quickly enough, the decision to delegate control to an autonomous system would be a challenging ethical and tactical dilemma. Thus, while AI-enabled autonomous early-warning systems would theoretically allow defense planners to identify and monitor threats faster and more reliably than before, the lack of human judgment and supervision coupled with the inherent brittleness (or lack of real-world common sense to deal with new situations) and black box (or opaque and unex¬plainable) characteristics of AI machine-learning algorithms mean that the risk of destabilizing acci¬dents and false alarms will likely rise. As strategist Thomas Schelling argued, “when speed is critical, the victim [in this case a nuclear-armed state] of an accident or a false alarm is under terrible pressure.”6 This point is especially true of the time pressures associated with AI. Fear created by the unpredictability and uncertainty of an enemy conducting warfare at machine speed may, therefore, tempt nuclear states to automate their nuclear retaliatory capability. That is, all else being equal, a nuclear-armed state less confident in its second-strike capabilities (i.e., China, North Korea, Pakistan, and perhaps Russia) will be more inclined to use automation. The high speed associated with AI and autonomy can have unexpected and escalatory outcomes. For example, the 2010 stock market flash crash—which reduced the stock market value by one trillion dollars within a matter of minutes—was, according to the US Securities and Exchange Commission (SEC), enabled and amplified by the use (or misuse) of autonomous financial trading algorithms.7 Unlike the financial markets, however, there is no overarch¬ing authority in international relations to enforce pre-programmed fail-safe mech¬anisms based on a shared set of rules. Thus, the prospect of a flash crash-like event occurring in adversarial, loosely regulated, offense-dominant, and strategically competitive domains, such as cyberspace, missile defense, or anti-satellite weapons (ASATs), is especially alarming. Military AI systems functioning at machine speed could push the pace of combat to a point where the actions of machines eclipse the ability of human decision-makers to control (or even comprehend) events. In extremis, human commanders might lose control of the outbreak, course, and termination of warfare. Were humans to effectively lose (or pre-delegate) control of warfare to machines, inadvertent escalation pathways and crisis instability would increase, potentially with catastrophic results.8 Compelled by the speed and pre¬cision of AI to make decisions in a compressed timeframe, a state might accept higher risks and escalate a conflict with the belief it was in a use it or lose it situation, or a lack of confidence in its ability to guarantee the safety and control of its nuclear arsenals.9