Description
How to stop “killer robots” is a popular way of framing the international debate about weaponised artificial intelligence (AI). Most media coverage features sensationalist images of humanoid robots out to get humanity. For scholars writing in this field, avoiding editorial inclusions of Terminator-like imagery, in particular, can feel like swimming against the stream. To address these issues, the paper asks: how is weaponised AI constructed across widely accessible visuals and what are the effects of these imaginaries? IR has increasingly become attentive to visuals as central to political meaning-making and contemporary warfare. In this context I argue that visuals in the debate about weaponised AI communicate narrative, sociotechnical imaginaries (Jasanoff 2015; Cave, Dohal, and Dillon 2020). Following an interpretive methodology, I analyse visuals found in Google image searches of key words such as "weaponised AI" and "AI weapons". This analysis shows three main narrative imaginaries: (1) visuals serve to associate problems of “killer robots” with far-off, futuristic scenarios, masking the extent to which autonomous features are already part of weapons in use today; (2) visuals privilege a Western reference-point as the imaginary for a technological revolution that is decidedly global in nature; and (3) visuals often paint a ‘clean’ and orderly picture of AI when the debate about weaponised AI would benefit, instead, from recognizing the ‘sweatiness’ of AI (Ahmed 2017). The paper concludes with reflecting on what common visuals on weaponised AI render visible/invisible and the boundary-drawing consequences of these performative moves.
