Smaller Chance of Mistake, More Innocents Dead: A Different Problem with Using AI Systems in War (and Elsewhere) 

Abstract

Using AI systems in identifying targets may very well increase accuracy, in the following sense: For each person suspected to be a legitimate target, the chance of a false identification is lower if such systems are used than if the analysis is done by humans. Using such systems is also incredibly cheap in time and other resources, so that many more targets may be identified. Introducing such AI systems may reduce the chances of a mistake, but radically increase the number of innocents killed. What is the moral status of such a shift?

In this talk, David Enoch will argue that the key to an answer is to understand that—and how—the required evidential threshold for a justified local attack may depend on the global features of the case. The discussion will proceed via analogies to questions about enforcement levels in the criminal justice system, and will engage more general questions about local-global coherence. David Enoch will show how the main insight and conclusion apply much more widely than merely to warfare.

David Enoch is Professor of the Philosophy of Law at Oxford, and the Rodney Blackman Professor of Philosophy and Law at the Hebrew University of Jerusalem. He works primarily in moral, political, and legal philosophy.

The Passmore Lectureseries at the ANU School of Philosophy honours the legacy of Professor John Passmore, a pioneering thinker whose work shaped modern philosophy in Australia and beyond. Held annually since 2000, the series invites distinguished philosophers to explore enduring questions in thought, reflecting Passmore’s spirit of intellectual boldness, critical clarity, and historical depth.