AI in Public Service: Stakeholder Framing of Implementing Robotized Process Automation in Financial Aid in Swedish Municipalities.

Sammanfattning: Artificial Intelligence has become a hot topic as improvements in available technologies previously considered science fiction, like image recognition, language processing, and self-driving vehicles, are becoming financially viable. Further, several studies point to how automation of processes will increase efficiency in a variety of white-collar sectors. The public sector is no exception, and the future is already here. Consider, for example, the application of Robotized Process Automation, a form of narrow AI, in financial aid through the so-called "Trelleborg model." This paper studies the experiences of early adopters of this form of process automation. Interviewing Politicians, Private Sector Consultants, Senior bureaucrats, Street-level bureaucrats, while building on technological frames theory, this study explores stakeholders' perception of RPA implementation on financial aid in 3 Swedish Municipalities. A diverse set of experiences is essential to include, as co-ordination is a common challenge to implementation. At the same time, different groups can add additional pieces to the puzzle on implementing AI in the public sector. Stakeholders identify some critical elements to successful implementation, including framing the change as a means to substitute routine tasks and spending more time on client interaction, finding ways to bridge gaps in communication between different groups of knowledge workers, and identifying realistic defined goals with implementation.