Ariel Maschke

In 2018, the United States Congress passed the Family First Prevention Services Act, for the first time linking federal reimbursement to child welfare organizations’ use of and demonstrated fidelity to evidence-based practices (EBPs). Subsequently, child welfare organizations face a novel accountability demand from the state: fidelity, or the degree to which individual professionals implement EBPs according to tested protocol consistently. Policy stipulates that organizations must monitor professionals’ EBP fidelity, but how organizations do it is their choice. Some child welfare organizations are adopting artificial intelligence (AI) tools to support fidelity monitoring. AI tools offer a seemingly simple solution: train an algorithm to listen to and produce evaluations of professionals’ work with clients. Conceptualizing AI tools as institutional actors, I set aside questions of tools’ technicality (e.g., how they work) and effectiveness (e.g., whether they improve professionals’ fidelity). Instead, I foreground tools’ sociality, exploring what happens when AI turns its gaze upon professionals rather than the people professionals serve. Accordingly, this study will produce: 1) a historical analysis of how AI emerged as a fidelity monitoring possibility and 2) an ethnographic account of how AI fidelity monitoring tools shape and are shaped by child welfare practice and social policy.