Christian Sandvig and his colleagues (2016) helped to set the research agenda for communication and information scholars concerned about the impact of algorithmic techniques for the generation of strategic intelligence for corporate and government decision-makers. Much of the research that followed was focused on the nature and extent of the biases and errors that emerged when assessments and recommendations affected the life chances of racial and ethnic minority population segments (Barocas & Selbst, 2016). Attention to the impact of these systems has just begun to be developed with regard to the challenges associated with the law, and its defense of the fundamental rights of members of those groups. This paper examines those concerns as they apply to the use of algorithmic systems by urban police, judges, and other central actors within the criminal justice system (CJS) in the United States (Kroll, et al., 2017; van Brakel & De Hert, 2011; Whittaker, et al., 2018; Winston, 2018).
Although the use of cameras for the surveillance of target areas within urban centers has been the subject of critical assessment almost from the beginning of their use, much of that work was focused on the behavior of the human monitors that determined what the central focus of those cameras would be, as well as the nature of the behaviors that would trigger the movement of officers to the scene (McPhail, B. and A. Clement, et al., 2013). Increasingly, however, the work of human monitors has been re-assigned to semi-autonomous computer systems, guided by artificial intelligence resources, updated routinely through the use of machine learning techniques (Berman, 2018; Mateescu, et al., 2015; Sackler and Sackler, 2017). The use of cameras, especially those by officers on foot patrol, or in motor vehicles is described, but a primary focus of this paper is on the computer-aided analysis of the images captured by these devices.
The capture and use of images from mobile cameras, the analysis of social networks as well as affective assessments of individuals and members of groups derived from automated analysis of social media text and images, as well as other transaction-generated information (TGI) that has come to be referred to as “big data,” has been recognized as contributing to the development of a transformative moment in the nature of policing (Brayne, 2017; Degeling and Berendt, 2017; Hu, 2017; Manovich, 2018). The application of these and other informational resources to the development of predictive policing has been recognized as presenting a genuine threat to the traditional meaning of “reasonable suspicion” and related justifications for the application of First, Fourth, Fifth and Fourteenth Amendment rights to the targets of police attention (Cohen, 2019; Ferguson, 2015, 2016; Maharrey, 2018).
This paper will explore these developments, with special regard to their likely impact upon the life chances, well-being, and social construction of members of racialized population segments in the foreseeable future (Carney & Enos, 2017; Turow and Hennessy, et al., 2018).