In this paper, we report results from fieldwork in the context of municipalities and governmental institutions looking to implement algorithmic decision-making in public service provision. We empirically investigate bureaucratic decision-making practices in the context of governmental job placement, a core public service in many countries, from the perspective of caseworkers. Acting as participants in a large cross-disciplinary research project between 2019-2020 (ongoing), we set up a participatory workshop with caseworkers. This was followed up by in situ interviews that allowed the caseworkers to think-aloud while guiding us as we talked through the decision-making process in governmental job placement. The paper’s contribution is a conceptualization of the characteristics of bureaucratic decision-making in the context of human-AI collaboration: 1) processual decisions that move forward the caseworker’s understanding of the individual case; 2) formal decisions whereby caseworkers close a bureaucratic process or individual’s application; 3) balancing decisions in which the caseworker weighs the potential consequences when a decision is uncertain or questionable. The application of human-AI collaboration in job placement, we argue in this paper, must take the kind of bureaucratic decision (processual, formal, or balancing) into consideration, as we consider from a CSCW perspective how to integrate AI into the already-complicated human workflow of bureaucratic decision- making.
https://doi.org/10.1145/3449114
The 24th ACM Conference on Computer-Supported Cooperative Work and Social Computing