The GovLab’s Akash Kapoor argues that while the Western world is making massive investments in the pursuit of AGI (artificial general intelligence), the majority world is pursuing a different direction, which he calls AAI (Applied AI). Applied AI asks: Does the model work on patchy 4G or refurbished phones? What yield can it bring farmers, or which bureaucratic bottleneck can it cut?

Faced with budget constraints and language barriers limiting access to global learning, Vietnam's Academy of Public Administration and Governance (APAG) embraces “working with what you have” to transform the public sector with AI.

At the recent  UN Public Sector Forum in Samarkand, I met Ms. Pham Thi Quynh Hoa, APAG's Acting Director of Science Management and International Cooperation. Alongside the APAG President, Associate Professor Nguyen Ba Chien, Ms. Pham is coordinating this pragmatic "learning by doing" journey to explore AI's role in ambitious public sector transformation.

Ms. Pham explained the context behind APAG's new approach. In 2024–2025, Vietnam launched a "streamlining revolution," consolidating 22 ministries to 17, cutting over 3,000 sub-departments and 22,000 positions, and abolishing entire district tiers. But organizational restructuring alone wasn't enough.

Vietnam's Politburo introduced resolutions calling for a fundamental shift in civil servants' roles, from implementers to adaptive, tech-savvy professionals capable of shaping policy, not just executing it. This reflects a shift from control to collaboration, procedures to performance.

At the heart of this transformation lies Vietnam's "Know – Want – Be Able" model, inspired by Boterf's competency triad. Civil servants must first know why change is necessary, want to drive change through intrinsic motivation, and be able to access the necessary tools, data and institutional support. This model now provides a common framework for transformation across all government levels.

Ms. Pham started on a note of humility, underlying that what they have done is a modest experiment, testing the waters with a new technology to fill a gap in the Academy’s offering, but perhaps more importantly, to support cultural change. Learning by doing is very much their motto.

APAG’s mission goes beyond training civil servants; it also includes a mandate to connect the Vietnamese public sector with global learning opportunities. As such, APAG serves as both a training hub and a center for collecting, contextualizing, and disseminating international knowledge to support Vietnam’s comprehensive civil service reform.

However, the reform process itself creates new and more complex demands for learning from international experience. This learning can no longer stop at mere “referencing.” It must be directly tied to Vietnam’s specific and pressing challenges, she says. 

Yet APAG faces significant constraints in fulfilling its mission to explore how global lessons can be meaningfully applied in Vietnam’s own context. The country lacks the resources to send civil servants abroad for study tours. Moreover, limited English and foreign language proficiency among staff presents a major barrier to accessing global materials. Traditional methods of sharing international experience often suffer from lack of depth, consistency, and systematic analysis, making it difficult to translate foreign insights into actionable knowledge.

The breakthrough came when Vietnam's General Secretary visited Singapore in March 2025 and asked a pointed question of why haven't Vietnamese civil servants actively studied and leveraged the valuable experiences of their neighboring nation.

In response, APAG worked with Professor Vu Minh Khuong at the Lee Kuan Yew School of Public Policy in Singapore, to use AI to support the Academy in fulfilling its mission.

As a Harvard-trained economist whose PhD focused on the application of information technology to enhance labor productivity, Professor Khuong developed a simple “case study” template that APAG staff could use to prompt ChatGPT to summarize international examples ( e.g. how did the Singapore government tackle a particular challenge in water management, or air pollution, etc). The template has 7 sections, starting with a question on rationale: “Why did the Singapore government get interested in this issue.”

APAG works with Prof. Dr. Vu Minh Khuong on the implementation of the project to build an excellent civil service

Ms Pham emphasized starting with "why" rather than jumping to "what was done," fostering curiosity and critical thinking by understanding root problems and motivations. The template also includes sections on applicability to Vietnam's priorities and implementation suggestions.

APAG then trained colleagues in AI's critical use: effective prompting, but also limitations like hallucinations and biases. For many, this was their first AI exposure. Crucially, recent improvements in Vietnamese language capabilities removed a major barrier to accessing international practices.

With Professor Khuong's coaching, APAG's first Friday evening training on May 16th drew 20 lecturers. The second session (less than 3 weeks later) attracted 262 participants—including the entire Board of Directors—out of 350 total faculty. Remarkably, these voluntary summer sessions offered no financial incentives, yet demonstrated a high level of engagement.

After the training session, APAG lecturers used Professor Khuong’s template to create 136 case studies that included best practices from Singapore and other countries, including China. While Ms. Pham initially tested the use of AI to select the cases, she wanted APAG faculty to tap into their initiative, curiosity, and passion. The faculty were encouraged to choose cases themselves from a longer list of 240 topics, as well as create their own based on their interests and expertise. An editorial board of 12 APAG faculty and 8 Lee Kwan Yew alumni peer-reviewed all 136 cases, with academic departments and practitioners providing additional validation.

Ms. Pham shared two insights: peer review quality varied, given the volume and the novelty of reviewing AI-generated text. However, human assessment by scholars and practitioners proved decisive in determining practical applicability. AI outputs were impressive, but human expertise was essential.

APAG and Professor Khuong aim to prune cases to around one hundred. Remarkable progress for an Academy that previously developed just three international case studies over decades.

Did colleagues become more curious about international practices? "Absolutely," she replied.  The experience also prompted Ms. Pham’s deeper reflection on how AI can complement human expertise: “For more than 30 years in the public service, I have evaluated countless documents, and I was confident I had become very skilled at it. Yet when I started using AI, I was struck by the fact that it could perform even better than I did. Imagine after three decades of relentless effort and quiet pride in mastering a craft, you suddenly discover that AI can do the same task in an instant and at a higher quality.”

But, she went on to add: “Naturally, I felt a moment of frustration. But that feeling lasted only a second."

Very soon, I soon realized that my expertise allowed me to critically evaluate and select the best among AI-generated suggestions. Without that background, the tool would not have been nearly as effective.

APAG will publish a website with the case studies later this month, and adopt a “proactive, systematic, and practice-oriented” approach to reform inspired by international best practices.

Building on this initial effort, APAG, in collaboration with Professor Khuong, has three additional projects planned:

1. In December 2025, they will publish a handbook “Accelerating Vietnam’s Development in the Era of National Ascent – Harnessing Artificial Intelligence to Learn from Global Best Practices in Building a High-Performing Civil Service.” With chapters by the Vietnamese Ambassador to Singapore and the Singaporean Ambassador to Vietnam, the book will cover international lessons for government innovation.

2. The Academy also plans to set up a chat tool to collect questions and challenges practitioners face in their daily work, so as to ground insights from case studies into the reality of Vietnam. Combining AI with the knowledge of APAG scholars and practitioners, the chatbot will offer a kind of “rapid response team” to help address practical problems in real time.

3. APAG intends to integrate the cases into its leadership and management training programs. The goal is to equip central and local leaders with the capacity to use AI to identify and analyze problems, learn from international experience, and apply it to practical solutions.

For Ms. Pham, this pragmatic “Applied AI” approach, working with constraints rather than against them, embodies a larger truth about development. “I know that we have to stand on the shoulders of giants while creating uniquely Vietnamese solutions”,  Ms Pham Thi Quynh concluded. “Yet the metaphor I prefer is that we must  learn to ride the waves - joyfully, if possible- rather than being drowned by them.”