New Paper: The Algorithmic State Architecture (ASA) – An Integrated Framework for AI-Enabled Government

Mar 17, 2025

Data for Policy is pleased to present a new preprint, “The Algorithmic State Architecture (ASA): An Integrated Framework for AI-Enabled Government,” authored by our Founding Director Dr Zeynep Engin (University College London), alongside esteemed co-authors Prof Jon Crowcroft (University of Cambridge, The Alan Turing Institute), Prof David Hand (Imperial College London), and Prof Philip Treleaven (University College London). Published as a preprint in March 2025, this paper introduces the Algorithmic State Architecture (ASA), a transformative four-layer framework designed to guide governments in harnessing artificial intelligence for public sector innovation.

As AI and data-driven technologies rapidly reshape the public sector, governments face significant challenges in integrating these innovations into cohesive systems. The ASA framework addresses this by providing a structured, four-layer model that explains how key components of AI-enabled government interact and reinforce each other:

  • Digital Public Infrastructure (DPI): The foundational layer that includes data exchange systems, cloud infrastructure, and secure platforms essential for delivering public services.
  • Data-for-Policy (DfP): The intelligence layer where data is collected, analysed, and used to inform policy decisions, improve service delivery, and measure impact.
  • Algorithmic Government/ Governance (AG): The process layer where AI systems enable automated decision-making and operational improvements while maintaining human oversight and accountability.
  • GovTech: The service layer that delivers AI-powered public services directly to citizens, ensuring accessibility, efficiency, and innovation.

Key Insights and Case Studies

What makes the ASA framework unique is its emphasis on interdependence—these layers are not standalone; they function as an integrated system where advancements (or weaknesses) in one layer directly affect the others. The paper draws on case studies from Estonia, Singapore, India, and the UK to demonstrate how this layered approach plays out in practice:

  • Estonia’s X-Road system shows how robust infrastructure enables advanced data-driven services.
  • Singapore’s Smart Nation initiative highlights how AI-driven analytics improve urban planning and service delivery.
  • India’s Aadhaar infrastructure illustrates the challenges of scaling foundational systems without fully developed governance frameworks.
  • The UK’s GovTech platforms reveal the importance of aligning technical innovation with strategic governance.

Policy and Practice Implications

The ASA framework highlights the need for integrated governance in AI-driven government transformation. Governance mechanisms should be embedded within technical systems to balance innovation and accountability. Key considerations include infrastructure, data privacy, algorithmic governance, and ensuring accessibility. Strategic implementation requires identifying system gaps, prioritising investments, and ensuring effective change management through stakeholder engagement. Performance metrics should assess cross-layer integration, governance effectiveness, and societal impact.

  1. Develop integrated governance across all ASA layers.
  2. Prioritise robust digital infrastructure and data governance.
  3. Build internal public sector expertise.
  4. Implement gradual, strategic development.
  5. Address digital divides and ensure equitable access.
  6. Strengthen democratic oversight in AI transformations.
  7. Foster cross-sector collaboration for ecosystem development.

These recommendations underscore the importance of governance, inclusion, and democratic values for successful AI-enabled transformation in government.

Looking Ahead

The paper highlights that AI integration is not just a technical challenge—it is a governance challenge. Future research should explore how different political and institutional contexts shape ASA implementation and how rapidly evolving technologies, such as foundation models, influence public sector development. The ASA framework also raises important questions about AI agency, democratic oversight, and the evolving relationship between AI systems and public trust.

This work marks a significant contribution to both theory and practice, bridging gaps in digital government research and offering a roadmap for assessing and advancing AI-enabled systems. As AI technologies, including foundation models, reshape public administration, the ASA framework stands as a timely tool for navigating this evolution responsibly and effectively. It is essential reading for policymakers, public sector leaders, and technologists seeking to navigate the complex landscape of AI-driven government transformation.

The full preprint is freely accessible on arXiv, and we invite our community to explore this innovative framework as it sets a new standard for AI-driven governance. Read the full paper.