Abstract The use of artificial intelligence-driven decision-support systems (AI DSS) to assist human calculations on the resort to military force has raised concerns that automation bias may displace human judgments. Such fears are compounded by the complexities and pathologies of organisational decision making. Discussions of AI often revolve around better training AI models with more copious amounts of technical data, but this article poses research questions that shift the focus to a human-centric and institutional approach. How can governments better train human decision makers and restructure institutional settings within which humans operate to minimise the risks of automation bias and deskilling? This article begins by exploring how governments have invested in AI literacy education and capacity-building. Second, it demonstrates how the need to question groupthink and challenge assumptions in decision making becomes even more relevant as the use of AI DSS become more prevalent. Third, human decision makers operate within institutional structures with internal audit trails and organisational cultures, inter-agency networks and intelligence-sharing partnerships that may mitigate the risks of human deskilling. Bolstering these three inter-locking, mutually reinforcing elements of education, challenge functions and institutions offers some avenues for managing automation bias in decisions on the resort to force.
Yee-Kuang Heng (Wed,) studied this question.