AI Applications in the Public Sector: Start Now or Wait for the State Budget?

DNVN - According to experts, one of the major barriers in deploying artificial intelligence (AI) in the public sector today is funding. In the context of limited budgets, is it possible to apply AI immediately with a suitable deployment model and available resources to take advantage of opportunities instead of waiting for the State budget?

Start with available resources

According to Mr. Dam Hai Dang - Expert of the Department of Science and Technology of Tay Ninh province, funding is the biggest challenge when applying AI to digital transformation in the locality. To deploy a basic virtual assistant system for public services, the minimum cost is up to 10 billion VND. If relying only on the State budget, the implementation process will be prolonged.

Instead of waiting for funding, Tay Ninh chose to start from its own available resources. The locality focuses on applying AI to specific areas, including tourism. Currently, tourists often search for information via Google, but the accuracy of these sources of information is not always guaranteed. Therefore, Tay Ninh has developed an AI system that provides official data on restaurants and hotels that meet standards recognized by authorities, while also assisting tourists in building reasonable itineraries.

Within the state agency, Tay Ninh also utilizes open source AI models such as Gemini, DeepSeek, Meta... to develop virtual assistants. By applying enhanced retrieval techniques (RAG), virtual assistants can quickly search for information in the province's document management system, helping officials save time searching without having to invest too much in infrastructure.

By 2025, the cost of AI applications is expected to continue to decrease thanks to the development of more optimal models. Tay Ninh is aiming to integrate AI Agents into the document management system, helping to automatically receive and classify documents, thereby reducing the workload for administrative staff. At the same time, the locality also aims to expand AI into public services, creating conditions for people to access information more quickly and accurately.

Treat AI as an investment, not just a cost

From the perspective of the management agency, Mr. Tran Anh Tu - Deputy Director of the Department of Science, Technology and Technology (Ministry of Science and Technology) said that integrating AI into state management activities not only helps save resources but also improves work processing speed. In addition to popular applications such as ChatGPT, the public sector needs tools to support decision-making and effective data synthesis to reduce complex manual processes.

Implementing AI in the public sector still requires a change from leadership thinking to how organizations operate.

The government is currently strongly promoting AI as a fundamental technology, through three main directions: developing data infrastructure, training human resources, and building AI models suitable for the Vietnamese language.

Regarding data, the National Data Center in Hoa Lac is being built to store and provide data from ministries, branches and localities, of which a part will be opened to serve research and application development.

In terms of human resources, there are currently more than 50 AI-related training programs at universities, but it is necessary to increase the dissemination of AI knowledge in the public sector to ensure effective application.

In terms of technology, Vietnam is implementing many projects to build a large-scale Vietnamese data warehouse, supporting the development of language models suitable for the country's cultural characteristics.

However, Mr. Tu emphasized that despite supportive policies, implementing AI in the public sector still requires changes from leadership thinking to organizational operations.

According to Associate Professor, Dr. Nguyen Xuan Hoai - Deputy Director of the National Innovation Center (NIC), to promote AI projects, it is necessary to first change the way of thinking and implementation methods. Some AI projects cannot be considered as just costs but need to be considered as long-term investments.

In addition, the way AI projects are costed needs to be adjusted, because the AI ​​implementation model is different from traditional IT projects. For example, AI can help optimize operational efficiency, so why doesn’t the State apply a performance-based payment mechanism instead of a fixed fee method? If an AI solution saves 30% of the State’s operating costs, can a small portion of that savings be used to pay the solution provider?

According to Associate Professor, Dr. Nguyen Xuan Hoai, to do this, we first need to change our mindset and be ready to accept new things. Pricing AI is still a difficult problem, even for developed countries, so Vietnam can refer to the way they are doing it. More importantly, there needs to be a testing mechanism (sandbox) to verify and gradually adjust, instead of expecting a perfect model from the beginning.

Thus, it can be seen that applying AI in the public sector does not necessarily have to start with large-scale projects with huge capital. Tay Ninh has proven that, with a flexible approach and taking advantage of available resources, state agencies can completely deploy AI right now.

The key is to change your mindset, find the right deployment model, and be willing to experiment. When AI is no longer just a distant concept, but becomes a practical tool to help optimize management and serve people, investing in this technology will become more necessary and urgent.

Moonlight

Source: https://doanhnghiepvn.vn/cong-nghe/ung-dung-ai-trong-khu-vuc-cong-bat-dau-ngay-hay-cho-ngan-sach-nha-nuoc/20250323085933977