Studying Surveillance AI-cologies in Public Safety: How AI Is in the World and the World in AI

Abstract

Technological surveillance for the sake of safeguarding public safety (e.g., cameras, sensors, mobile phones, OSINT) pervades the lives of individuals on many levels. In this article, we advance the idea that the addition of AI changes the way surveillance ecologies function and thus deserves to spawn its own concept: the surveillance AI-cology. Surveillance AI-cologies are made up of interconnected collections of disparate actors (technological, human, more-than-human, organisational, etc.), all implicated in AI-aided surveillance tasks. They contain not only the usual complexities of any technological ecosystem but also the added complexity of AI, with emergent characteristics, both technically and socially. We argue for the utility of multi-faceted perspectives in doing work within AI-cologies, and we describe (anthropologically inspired) methodology for understanding and unpacking AI surveillance ecosystems. The development of democratically controlled AI surveillance requires the systematic consideration of ethical, legal, and social aspects (ELSA) within the quintuple helix (public, private, civil society, academia, nature). We stress the relevance of clearly defining which perspectives of the quintuple helix are considered in AI surveillance, and which not, to achieve a transparent set of (ELSA) values that guide AI surveillance development and implementation. We provide an example of the way we have developed and applied (some of) these methodologies in the context of a test-site for the development and application of smart city technology, a so-called “Living Lab.” Here we take the stance of active involvement of academics as “critical friends” into complex innovation and assessment processes. Together with our conversation partners in the field, we tease out and reflect upon the (public safety) values embedded in the setup of the Living Lab we explore. We end with a call to understand surveillance AI-cologies not as a problem to be solved, but as a continuing process to be discussed among highly diverse stakeholders.