Street-Level Algorithms and AI in Bureaucratic Decision-Making

Abstract

Studies of algorithmic decision-making in Computer-Supported Cooperative Work (CSCW) and related fields of research increasingly recognize an analogy between AI and bureaucracies. We elaborate this link with an empirical study of AI in the context of decision-making in a street-level bureaucracy: job placement. The study examines caseworkers' perspectives on the use of AI, and contributes to an understanding of bureaucratic decision-making, with implications for integrating AI in caseworker systems. We report findings from a participatory workshop on AI with 35 caseworkers from different types of public services, followed up by interviews with five caseworkers specializing in job placement. The paper contributes an understanding of caseworkers' collaboration around documentation as a key aspect of bureaucratic decision-making practices. The collaborative aspects of casework are important to show because they are subject to process descriptions making case documentation prone for an individually focused AI with consequences for the future of how casework develops as a practice. Examining the collaborative aspects of caseworkers' documentation practices in the context of AI and (potentially) automation, our data show that caseworkers perceive AI as valuable when it can support their work towards management, (strengthen their cause, if a case requires extra resources), and towards unemployed individuals (strengthen their cause in relation to the individual's case when deciding on, and assigning a specific job placement program). We end by discussing steps to support cooperative aspects in AI decision-support systems that are increasingly implemented into the bureaucratic context of public services.