Understanding the Battle for AI in Warfare through the Practices of Assemblage: A Case Study of Project Maven

Abstract

Artificial Intelligence (AI) has been widely-heralded as a revolutionary technology across a range of domains, including defence. In this sphere, its potential for automating low-cost and low-risk forms of warfighting known as “remote warfare” has given rise to fears of Lethal Autonomous Weapons Systems (LAWS). While the use of LAWS has been subject to intense speculation, far less attention has been paid to their development. Notably, cutting-edge AI innovation is not to be found in government-funded research labs or traditional defence contractors, but consumer technology companies. Because of this, unique controversies have arisen over its development. In the case of Project Maven, a US Department of Defense initiative seeking to leverage AI for automating drone footage analysis, employee protests eventually forced one of its contractors, Google, to end their involvement. Using Project Maven as a case study, this research therefore seeks to understand the development of an emerging Military-Technological Complex developing these technologies and an oppositional Civil Society Coalition seeking to regulate them. Based on documentary analysis of key texts and five semi-structured expert interviews, this thesis uses an assemblage approach to examine the discourses surrounding Project Maven, the interplay of power between its elements, and their resulting configurations. It finds a tension between the compulsory powers of the Military-Technological Complex, exercised through structural, material relations, and the productive powers of Civil Society, exercised through the production and sanctioning of knowledge. It concludes that the development of consumer technologies for warfare and the associated emergence of a Military-Technological Complex reflects a broader unravelling of conventional ties between war, space and time.