An Arms Race frame for AI increases the risk of accidental wars – the speed of decision making and the vulnerability to hacking make miscalculation and escalation inevitable
Vestner, 2021 - Head of Security and Law Programme at Geneva Centre for Security Policy [Tobias, July 8 “Warfare and Artificial Intelligence” in Robin Geiß and Henning Lahmann (eds), Research Handbook on Warfare and Artificial Intelligence -forthcoming https://www.gcsp.ch/publications/military-operations-and-artificial-intelligenceGCSP Acc 5/27/22 TA]
Most importantly for military strategy, AI applications may assist decision-makers to monitor the battlefield and develop scenarios. Indeed, AI could be developed to predict the behaviour and reactions of foreign countries or generate simulations of the progression of ongoing conflicts.32 AI may also be useful to assess threats, provide risk analyses, and suggest courses of action, ultimately guiding decision-makers on the best response to take.33 In addition, AI may support the alignment of the armed forces’ ways and means with the given political and strategic objectives - a major function of military strategy. A consequence of such developments would be an increased speed and quality of military processes. While this would provide significant advantages to those states with the most performant AI,34 this may also pressure armed forces to increasingly delegate the orchestration of military operations to AI systems.35 Indeed, the use of AI for military strategy may also lead to challenges. Reliable AI systems would need to be trained with vast data sets.36 Furthermore, it has been warned that AI may exacerbate threats, transform their nature and characteristics, and introduce new security threats.37 A tabletop exercise on the integration of AI into nuclear C2 systems showed that such systems were ‘vulnerable to malicious manipulation that can severely degrade strategic stability’,38 for instance. Such vulnerabilities would derive mostly from the risk posed by third actors using techniques to deceive, disrupt or impair C2 systems,39 which indicates the importance of system safety for AI to be used for military strategy. Another significant challenge is that AI may accelerate the speed of warfare to the extent that humans cease to be able to follow the developments, ultimately leading humans to lose control. 40 This phenomenon has been termed battlefield ‘singularity’ or ‘hyperwar’.41 This may lead to strategic errors and accidents, including involuntary conflict escalation. Even if such risks can be alleviated, the increased reliance on AI would reduce the human element of military strategy, in particular psychology and human judgment. It has been argued that this could lead to a ‘gap between how the AI solves a problem framed by humans, and how those humans would solve it if they possessed the AI’s speed, precision, and brainpower’.42 Yet it has also been argued that strategy development would require the understanding of values, the balance of costs, and the understanding of the complex social system in which war operates, thereby significantly limiting AI’s use for military strategy.43 Yet it is also possible that when enemies possess high levels of rational prediction power provided by AI systems, the decisive factor in warfare will not be the AI systems’ capabilities but the human judgment, in particular concerning critical and difficult choices.44 This, however, presumes a certain level of meaningful human involvement. In sum, AI may enhance military strategy development and strategic decision-making, notably if able to process more data and make sense of complexity with more precision and at a higher speed than humans and simple computing. A likely result is an acceleration of military operations, which may increase pressure on armed forces to integrate AI and may marginalize human judgment. States’ recent adoption of defence strategies on and related to AI indicate that states increasingly intend to develop, acquire, and operationalize AI for military purposes. As such, the possession and use of AI is a strategic objective itself. In light of secrecy around the development of new technologies, states’ investment in military AI can become a strategic liability as it may increase the risk of destabilizing arms races, misperceptions, and miscalculations.