Policy dialogue: An AI-Ready public administration: Building inclusive, accountable, and anticipatory governance

The policy dialogue featured international experts, including Assoc. Prof. Dr. Eduardo Araral, Lee Kuan Yew School of Public Policy, National University of Singapore; Member of the UN Committee of Experts on Public Administration (CEPA); Dr. Prabin Maharjan, Programme Management Expert at UNPOG/DPIDG/UNDESA; and Ms. Georgina Curto Rex, Head of Research at the UN University Institute in Macau, Macau SAR, China.

The speakers discussed various issues related to building AI-ready public administration.

Opening the dialogue, Prof. Dr. Alex Brillantes emphasized that AI readiness is not merely a technological issue but, first and foremost, a matter of institutional preparedness, governance capacity, and the adaptability of public administration. Accordingly, preparations for AI need to be situated within science, strategic foresight, systems thinking, risk analysis, and institutional innovation. They must also be linked to the requirements of inclusiveness, capacity development, human rights protection, accountability, and international cooperation in AI governance. This perspective highlights the need to shift the approach from solely focusing on technology adoption towards building sufficient institutional capacity to govern the changes brought about by AI.

In terms of institutional capacity, Assoc. Prof. Dr. Eduardo Araral affirmed that public administration’s readiness for AI is not only identified by access to technology but, more importantly, by human capacity, data quality, governance mechanisms, and implementation capacity. Drawing on his experience teaching AI in public policy and public management, he pointed out that one major barrier is the concern that AI may change or replace certain jobs, thereby undermining incentives for data sharing and organizational collaboration. Therefore, building AI-ready public administration requires building confidence, strengthening workforce capacity, and establishing effective ways for humans to work with AI simultaneously.

According to  Assoc. Prof. Dr. Eduardo Araral, data is the foundation of AI capabilities.  Agencies in the public sector need to develop shared databases that can be interconnected and used with AI tools, supported by a clear data governance framework to ensure data quality, accuracy, security, and appropriate use. As AI increasingly contributes to analysis, generates recommendations, and supports decision-making, human accountability and oversight become particularly important. It is necessary to clearly define the scope of AI use, identify the stages at which human supervision is required, and determine who bears ultimate responsibility for decisions made with AI assistance.

In addition, Assoc. Prof. Dr. Eduardo Araral stressed that AI capabilities must be developed in ways that reflect specific institutional contexts, languages, and data characteristics. Standardizing prompts, developing tool libraries, and enhancing civil servants’ skills are essential conditions for AI to deliver tangible value. Above all, leadership plays a decisive role in driving change and creating momentum for AI adoption across the entire administrative system.

Regarding anticipatory governance and the capacity to proactively prepare for the future, Mr. Prabin Maharjan argued that a government cannot be considered truly AI-ready without the capacity for foresight and proactive governance. Many existing AI readiness assessment frameworks primarily examine whether governments can apply AI at present, without fully capturing their ability to anticipate, govern, and adapt to the changes AI may bring in the future. Meanwhile, AI is evolving rapidly, even faster than many governance systems can adjust. Therefore, AI-ready public administration must do more than simply “know how to use AI”; it must be able to “anticipate, prepare, and adapt proactively.”

To move from “readiness” to “resilience and adaptability,” Mr. Prabin Maharjan identified four key pillars: (1) continuous anticipation capacity to identify emerging opportunities and risks at an early stage; (2) systems thinking to assess AI’s impacts holistically, not only within individual agencies but also across sectors, social groups, and policy outcomes; (3) the development of flexible, inclusive, and accountable institutions, in which humans retain control and responsibility for decisions supported by AI; and (4) the development of an AI-literate civil service capable of evaluating, verifying, and using technology responsibly.

In other words, truly AI-ready public administration is not simply an administrative system capable of deploying existing technologies. It must be a governance system that can anticipate the future, proactively manage risks, ensure inclusiveness and accountability, and continuously learn and adapt to changes in AI. This is the foundation for moving from “readiness” to “resilience and adaptability” in an ever-changing technological environment.

Discussing solutions to narrow gaps in AI capacity-building amid disparities between countries and social groups, Ms. Georgina Curto Rex emphasized the need to adopt anticipatory governance to identify emerging risks of exclusion, inequality, and data bias. According to her, building AI readiness must go hand in hand with inclusiveness, prioritizing investment in digital infrastructure, data, and AI talent in localities, communities, and countries with limited resources.

At the same time, coordination among governments, the private sector, academia, and international organizations needs to be strengthened to ensure that the voices of vulnerable groups are adequately and clearly reflected in policymaking. AI capabilities must also be developed across the entire governance system, from central to local levels, so that all citizens have opportunities to access and benefit from the technology. This would help narrow the digital divide and promote equitable and inclusive development.

The dialogue also addressed the need to change how government performance is measured in the AI era. In addition to indicators such as service delivery speed and cost, the speakers proposed incorporating measures of governance quality, public trust, institutional resilience, inclusiveness, and sustainability.

Regarding implementation in practice, the speakers agreed that governments only beginning to establish digital foundations should not rush to deploy AI across all areas. Instead, they should adopt a gradual approach, starting with digital infrastructure, high-quality data, connectivity, and an appropriate institutional framework. They should then strengthen the digital and AI capabilities of civil servants and pilot AI applications in selected areas before scaling them up.

At the same time, as AI becomes increasingly involved in decision-making and public service delivery, governments need to clearly define human responsibilities and ensure transparency, verifiability, data protection, and the protection of citizens’ rights. Anticipatory governance should also be incorporated from the design stage to identify emerging risks of exclusion, inequality, and data bias through impact assessments, system testing, and ongoing monitoring of implementation outcomes. More importantly, vulnerable groups must be involved in the process to ensure that AI is not only effective but also implemented fairly and inclusively, without exacerbating existing inequalities.

Concluding the dialogue, Assoc. Prof. Dr.  Nguyen Thi Thu Hà summarized the key points raised by the participants and emphasized that AI readiness is not merely about the ability to apply technology; it requires transformation across the entire governance system. Technology can only deliver value when embedded in an appropriate institutional framework and accompanied by accountability, data governance, inclusiveness, and strategic foresight. This provides a basis for governments to proactively develop adaptation roadmaps, harness AI’s potential to improve governance effectiveness while identifying and managing risks, and ensuring that sustainable development objectives are achieved.