AI in Education and Public Administration

Education and public administration share a silent commonality: both are large-scale human-serving systems where one individual (a teacher, an official) often has to meet the vastly different needs of dozens or hundreds of others. This is precisely where artificial intelligence demonstrates its value — not to replace humans, but to free them from repetitive tasks and expand their service capabilities. This article presents real-world applications of AI in teaching-learning and public services in 2025-2026, along with the accompanying opportunities and risks.

Part 1: AI in Education

General Overview in Vietnam

Vietnam’s education technology (EdTech) is rapidly digitizing. According to the Vietnam EdTech White Paper 2024, up to 60% of domestic EdTech products have integrated AI technology. 2025 is considered the “breakout” year for AI in education, with many applications entering actual classrooms.

AI in education focuses on three main application groups:

1. Personalized learning assistants (virtual tutors). This is the most anticipated application. A chatbot acts as a “virtual tutor,” closely tracking each learner’s progress, providing detailed explanations, practice exercises, and suggesting learning methods tailored to their proficiency level. The key is personalization: instead of a general lesson for the entire class, each student receives support appropriate to their strengths and weaknesses.

2. Automated grading. In large-scale online classes, AI supports quick and consistent grading, especially for multiple-choice questions and structured essays. This frees up immense time for teachers.

3. Teacher lesson preparation support. AI tools suggest topics, create slides, compile lesson plans, and design exercises at various cognitive levels, helping to standardize and update content.

Global Example: Khanmigo

To understand the potential of virtual tutors, let’s look at Khan Academy’s Khanmigo — currently the most prominent AI-powered learning assistant. In the 2024-2025 academic year, the number of K-12 students using Khanmigo surged from 40,000 to 700,000, with projections to exceed one million in the 2025-2026 academic year.

A noteworthy design aspect of Khanmigo: unlike typical chatbots, it does not directly provide answers but rather guides learners to discover solutions themselves, with “infinite patience.” This is a crucial pedagogical difference — effective AI in education must teach how to think, not just do the work.

Opportunities and Risks in Education

Opportunities: If implemented responsibly, virtual tutors can alleviate teacher workloads, personalize learning for millions of students, and narrow the gap in access to quality education between urban and rural areas — a particularly significant point for Vietnam.

Risks to manage:

• Academic dishonesty: students using AI to complete assignments instead of learning. Solutions like Khanmigo (guiding instead of giving answers) are a correct approach.
• Data privacy: learning systems collect data about individual students. Khanmigo, for instance, advises users not to input Personally Identifiable Information (PII) into conversations and displays a reminder “do not share personal data.”
• Algorithmic bias: if training data is imbalanced, AI may treat certain student groups unfairly.
• Access inequality: if only affluent schools and well-off students can access AI, the technology could widen the gap instead of narrowing it. This is a policy risk that needs attention.
• Erosion of human interaction: AI cannot replace the teacher’s role in inspiring and fostering character development.

Part 2: AI in Public Administration

”Digital Officials” Working 24/7

Public services are where citizens often encounter barriers: complex procedures, uncertainty about whom to ask, and limited office hours. AI chatbots precisely solve this problem — operating as a 24/7 “digital official”, guiding procedures, retrieving information, and receiving feedback, without time or request quantity limitations.

Practical Implementation in Vietnam

2025 sees many localities introducing virtual assistants to public service portals:

• Binh Phuoc Province (February 2025) officially launched an AI chatbot system on its Public Service Portal, operating as a “digital official” to support citizens and businesses in looking up information, guiding procedures, and receiving feedback around the clock.
• Hanoi supports citizens via chatbots and virtual assistants on the Zalo OA platform, the iHanoi application, and the city’s Public Service Portal.
• Ho Chi Minh City (HCMC) integrated chatbots into its Public Service Portal to answer questions about business registration, construction permits, and other procedures, allowing citizens to submit questions via an online interface and receive instant responses.
• UNDP Vietnam developed a chatbot at dichvucong.me, integrating both text and voice, supporting 15 basic administrative services 24/7. By early September 2025, one system had supported over 1,000 inquiries from citizens in looking up information and completing procedures.

From a policy perspective, public administration journals in 2025 have discussed the integration of AI into public administrative management as an inevitable modernization trend, extending beyond chatbots to data analysis and process automation.

Opportunities and Risks in Public Administration

Opportunities: Administrative chatbots reduce the burden on one-stop officials, shorten waiting times, and make services more accessible for people in remote areas or those unfamiliar with procedures. Voice support (like UNDP’s chatbot) further expands accessibility for the elderly and those with literacy difficulties.

Risks to manage:

• Accuracy and legal liability: chatbots providing incorrect procedural advice can cause harm to citizens. Information provided by AI must be verified and updated according to the latest legal regulations.
• Citizen data security: public services handle sensitive information (ID cards, personal records). Data security must be a top priority.
• Digital divide: individuals who do not use smartphones or are unfamiliar with technology may be left behind. Smart public services must be accompanied by traditional channels, not completely replace them.
• Transparency and complaints: when AI makes decisions affecting citizen rights, there must be a mechanism for human review and appeal.

Common Denominators

Both education and public administration show that AI is most useful when acting as an assistant that extends service capabilities, not as the final decision-maker. Virtual tutors support rather than replace teachers; “digital officials” guide rather than replace accountable personnel.

Three overarching principles for responsible implementation:

1. Human-in-the-loop: always have humans verify and take responsibility for important decisions.
2. Data protection: especially for student and citizen data — transparency in collection, usage, and storage.
3. Equitable access: design to narrow, not widen, the digital divide among population groups.

In 2025-2026, Vietnam is taking its initial yet substantial steps: from public service chatbots in Binh Phuoc, Hanoi, and HCMC to virtual tutors in EdTech platforms. Success will not only be measured by the extent of technology deployed but by whether it truly serves people fairly and reliably.

References

• 12 AI Applications in Education 2025 — PHX SmartSchool
• AI Applications in Education — Explosive Trend in 2025 — Doi Moi Giao Duc
• Meet Khanmigo: Khan Academy’s AI-powered teaching assistant & tutor
• Student data privacy and security: Khan Academy’s Khanmigo — Public Services Alliance
• Hanoi uses AI, virtual assistants to help citizens with online public services — VietNamNet
• AI Chatbot to serve as a “digital official” in online public services — VnEconomy
• [Applying Artificial Intelligence in Public Administrative Management — Journal of State Management](https://www.quanlynhanuoc.vn/2025/08/19/ung-dung-tri-tue-nhan- tao-trong-quan-ly-hanh-chinh-cong/)