Artificial intelligence as a mode of ordering. Automated-decision making in primary care

This paper examines the process of organizing primary care from the increasing production of digital data and algorithms, leading to decision-making being delegated to automated processes. Based on a case study of primary care in Catalonia (Spain), we develop an interpretative framework to analyze the effects and controversies of the introduction of AI in primary care, stressing the need to understand these technologies as historically situated trends, as part of a growing automation process of ordering frontline clinical care. To foster a nuanced discussion on the mode of ordering of automation, we replace the notion of AI with Automated Decision-Making (ADM) and use the conceptual distinction between striated and smooth spaces, as developed by Deleuze & Guattari. With the analysis we identify four dynamics that order and organize health, illness, and public health systems: (1) The capacity of Electronic Health Records and associated software to shape the daily routines of primary care; (2) The dual role of healthcare professionals, who serve as both data users and data collectors; (3) The system of values and priorities configured by the network of classification systems in use and; (4) The emerging tensions during the progressive automation processes. These dynamics configure a mode of ordering characterized by diminishing the importance of experiential knowledge, the reduction of patient negotiation capacity, the professionals’ discomfort with being constantly monitored and controlled, as well as the gradual neglect of factors related to socio-economic determinants of health.