Scientific Machine vs. Power Politics: Conflict Anticipation and the Black Box of AI

How does artificial intelligence shift power in international security? A burgeoning literature in international politics and security studies has documented its effects on the balance of power, strategic stability, and the future of warfare. In this work, power is largely material, if not kinetic, and the specifics of technologies are treated mostly as peripheral. By recovering classical International Relations theory in the form of Hans Morgenthau’s work on the role of scientific rationalism in guiding political decision-making and combining it with insights from Science and Technologies Studies, this paper investigates the role of so-called intelligent technologies, in particular machine learning, in the knowledge production for conflict prevention. Such technologies are met with enthusiasm in the policy sphere, prompting a wide range of actors in the field of conflict prevention to integrate them into their analyses. Leveraging original elite interviews with conflict modelers, practitioners, and policymakers, this paper tentatively argues the rush towards integrating AI and ML is not primarily about improving predictive analytics in terms of scale, speed, and cost, but about creating options and justifications for (in)action. Due to the internal opacity (‘black-boxing’) of machine learning, policymakers can delegate the responsibility of the analysis from the human to the machine, thus transforming problems of politics and power into problems of process and technology. This research has implications for appreciating the internal mechanisms and characteristics of emerging technologies, as well as their  underlying rationalities, to understand how they shape actors’ options for decision-making.

Speaker: Johanna Rodehau-Noack, Stanford University

Attend in person or online (see weblink)

Perry Conference Room