Data in New Delhi's predictive policing system

Abstract

In 2015, Delhi Police announced plans for predictive policing. The Crime Mapping, Analytics and Predictive System (CMAPS) would be implemented in India's capital, for live spatial hotspot mapping of crime, criminal behavior patterns and suspect analysis. Four years later, there is little known about the effect of CMAPS due to the lack of public accountability mechanisms and large exceptions for law enforcement under India's Right to Information Act. Through an ethnographic study of Delhi Police's data collection practices, and analysing the institutional and legal reality within which CMAPS will function, this paper presents one of the first accounts of smart policing in India. Through our findings and discussion we show what kinds of biases are present within Delhi Police's data collection practices currently and how they translate and transfer into initiatives like CMAPS. We further discuss what the biases in CMAPS can teach us about future public sector deployment of socio-technical systems in India and other global South geographies. We also offer methodological considerations for studying AI deployments in non-western contexts. We conclude with a set of recommendations for civil society and social justice actors to consider when engaging with opaque systems implemented in the public sector.