Democratic legitimacy of algorithmic public services across citizen–system interfaces: the case of Bürokratt’s algorithmic governance in Estonia (2020–2025)

Abstract

This thesis examines the degree of democratic legitimacy of Bürokratt’s governance arena that involves the algorithmic mediation of public services in Estonia between 2020 and 2025. Bürokratt is analysed as an interoperable, LLM-based virtual assistant that aims to facilitate citizens’ access to public services and strengthen Estonia’s digital government infrastructure. The thesis combines literature on algorithmic governance, transparency and legitimacy with the framework of democratic anchorage in governance networks. Methodologically, it adopts a qualitative single-case study design based on public documents and public-facing textual materials that directly mention Bürokratt. The corpus includes government strategies, policy documents, public webpages, technical reports, European Union materials, academic studies, opinion articles and newspaper articles. The analysis assesses Bürokratt’s governance across four anchorage points: democratically elected politicians, participating groups and organisations, territorially defined citizenry, and democratic rules and norms. The findings show that Bürokratt’s democratic legitimacy is institutionally strong but democratically uneven. The strongest evidence appears in its anchorage in elected politicians and politically accountable institutions, especially through ministerial sponsorship, national AI strategies, agenda-setting and public-sector coordination. The evidence is more limited regarding participating groups and organisations, where actors appear mainly as implementers, adopters, or technical partners rather than as representatives connected to membership bases. In the citizenry anchorage, citizens are highly visible as users and beneficiaries of improved public services, but much less visible as actors able to participate in, contest or influence the system’s governance. Democratic rules and norms appear only partially, mainly through stakeholder involvement, openness, and references to rights protection, but are not consistently translated into explicit procedures for deliberation, justification, or contestation. The thesis concludes that Bürokratt’s legitimacy is largely constructed through state capacity, service efficiency and Estonia’s broader digital government narrative. Its democratic anchorage remains more limited where legitimacy would require clearer citizen influence, membership-based representation and visible mechanisms of public justification.