Implementing AI in the Swedish public sector – a struggle on knowledge?

Over the past decades, Swedish politicians have initiated ambitious plans that aim to establish the country as ‘world class’ in terms of digitalisation, specifically with respect to public administration (N2018.14). This development also fits the Swedish self-image as a high-tech nation, with a strong belief in technological knowledge as a strategy for creating a prosperous nation (Petersén, 2019; Bergwik, S. et.al., 2014). Lately, this task has increasingly been linked to the so-called 4th industrial revolution – the rise of artificial intelligence (AI) and advanced robotics as part public and private organisations. This evolution is also increasingly documented in research which shows that data driven and automated systems that incorporate new technological developments in the areas of AI, algorithms and big data have rapidly become an integrated part of public governance and administration (Wirtz et.al 2019). While the presence of such systems in the context of governing is often communicated and understood as enhancing neutrality and accuracy in decision making, they also produce quite profound political challenges for conceiving and implementing a fair, inclusive, and impactful 21st century politics and public sector (Bucher, 2018). In this paper, we are interested in focusing on how knowledge is perceived and produced in these processes, as a strategy of investigating aspects of power in the implementation of new technologies in the public sector. Drawing on Foucault’s (1982) understanding of the knowledge-power nexus, we believe this approach to be a fruitful strategy to analyse patterns of privilege when new technology is to be implemented. Here, we also believe Sweden, with its long tradition of technology optimism in combination with a strong trust in the state, to be a crucial case. The aim of the paper is to analyse the different rationalities of knowledge that are activated when new technology is brought into a public organization. More specific, we are interested in both how knowledge is produced in policy, as well as is given meaning by different actors, like politicians, civil servants and different groups of professionals, including engineers. In addition, we will also focus on how knowledge is negotiated between private and public actors, as the number of private-public-partnerships are growing in this field. Empirically, we will make use of the National Network on AI and Public Authorities, where a growing number of Swedish public authorities are members, through which we both will collect policy material as well as find informants from both the public and the private sector.