Description
This paper examines the puzzle of why, despite the Pentagon’s experiments with these technologies dating back to the 1970s, establishing US global leadership in the field of weaponised AI has only emerged as a goal of US defence planning following the institutionalisation of great power competition as the focus of American foreign policy during Donald Trump’s presidency. Building on three existing sets of explanations for this puzzle – strategic, cultural, and organisational – this paper involves an interpretivist content analysis of two major sets of US defense planning documents published after the Cold War: National Military Strategies and Quadrennial Defense Reviews/National Defense Strategies. Three technological anxieties are shown to have predated – and subsequently been grafted onto – the Pentagon’s more recent thinking about weaponised AI, these being that: (1) advances in dual use technologies have the latent potential to alter the character of war; (2) the diffusion of Anti-Access/Area Denial capabilities threatens the US’ global power projection capabilities; and (3) only overwhelming US technological dominance can prevent destabilising geopolitical competition. Through this intervention, this paper highlights the need for both a greater historical sensitivity toward the technological dimensions of great power competition and an awareness that the Pentagon’s approach to the development of weaponised AI has been shaped by anxieties about both the capabilities of strategic competitors like China and global technological developments since the end of the Cold War. For the various stakeholders contributing to the international debates on this subject, this suggests that a fuller understanding of the dynamics involved with the weaponisation of AI requires consideration of both what US policymakers have been and is currently worried about.
