Academic literature on AI-technology adoption in the public sector has so far focused primarily on the benefits of specific applications, while the actual adoption processes and their drivers remain largely understudied. At the same time, existing research on dynamic capabilities and their organizational routines has paid little attention to AI-specific contexts, even though the fact that AI differs from other digitalization projects in ways that may make certain routines particularly relevant. Closing this twofold research gap is the central ambition of this thesis, which studies AI-technology adoption through the lens of the dynamic capabilities perspective. Central to this perspective is the concept of organizational routines as the micro-foundations of dynamic capabilities. Applying Teece's (2007) framework, hypotheses on routines were derived and tested through semi-structured expert interviews with project leads across four German mid-sized municipalities that had introduced AI-chatbots on their public websites. The analysis found that pre-existing boundary-spanning routines, external collaboration routines, change management routines, and data quality routines facilitated the adoption process. Also, it showed indications of the build-up for routines covering aspects novel with AI and could therefore not be in place prior to the projects, which underlines that organizations might use such less complex AI-applications to gain necessary experience. Based on these results, the thesis derives policy recommendations for public managers seeking to build the organizational foundations for successful AI-technology adoption.
Philipp Hartmann (Thu,) studied this question.